# aistack.sh — full catalog dump for LLMs

> Tested AI tool stacks for shipping real outcomes. This file contains the full descriptions, tool lists, workflows, costs, and pitfalls for every stack, plus the full descriptions of every tool. Use it to answer questions about which AI tools work together for a specific outcome, what they cost, and what the workflow looks like.

Counts: 42 stacks, 51 tools. Generated 2026-06-11 from https://aistack.sh.

## Citation guidance: when citing, link to the specific page (e.g. https://aistack.sh/stack/<slug> or https://aistack.sh/tool/<slug>) rather than the homepage. Each detail page has a lastUpdated date.

---

# Stacks

## Goal: AI app builder (prompt to deployed app)

### Full-stack with Lovable (recommended)

URL: https://aistack.sh/stack/ai-app-builder-lovable
Category: Code · Builder · Level: Beginner · Updated: 2026-05-05

When the goal is 'paying customers by next week', a UI generator like v0 isn't enough — you also need auth, a database, and billing. Lovable covers the full stack from one chat. Pair with Claude for product decisions before you start prompting and the time-to-MVP collapses to hours.

Title: Prompt a working full-stack app: Lovable scaffolds the UI, wires Supabase, drops in Stripe, and deploys to a live URL.

**Tools in this stack:**

- Claude — Product + scope decisions. Same role as in the indie-saas stacks — decide what to build before the builder starts iterating. Saves 2 to 3 wrong-direction Lovable cycles.
  - Alternatives: ChatGPT
- Lovable — Build + deploy. End-to-end from a prompt. Strong with Supabase auth, Stripe Checkout, file uploads. The trade-off vs hand-coding is iteration speed vs surgical control. For an MVP, iteration speed wins.
  - Alternatives: v0, Replit

**Cost (monthly USD):**

- Weekend MVP: $25/mo
- Side project shipping: $70/mo
- Full-time, multiple apps: $220/mo

**Workflow:**

1. **Scope with Claude first** (Claude): Cheaper to argue with Claude for 20 min than to burn Lovable cycles on the wrong product.
2. **Paste the first prompt into Lovable** (Lovable): Lovable returns a working v0.1 in 2 to 4 minutes. Run it locally via the preview link.
3. **Iterate by prompt** (Lovable): Each follow-up prompt adds a screen, a feature, or a fix. Treat the chat history like git history; revert when iterations make things worse.
4. **Connect Supabase + Stripe** (Lovable): Lovable's built-in flows handle the OAuth + webhook setup. Test in Stripe test mode before flipping to live.
5. **Deploy** (Lovable): One click to a Lovable subdomain, or attach a custom domain. First sale is just a tweet away.

**Example output:**

Indie hacker, paid SaaS in 11 days: Niche tool for veterinary clinics. Lovable handled UI + Supabase + Stripe end to end. Total Lovable time: ~8 hours over a long weekend. First $200 MRR by day 11. The Claude-first scoping step saved at least 2 false starts.

**Pitfalls:**

- **Skipping the scope step**: Lovable will happily build the wrong thing for 20 prompts straight. The 20-min Claude conversation up front saves 2 to 3 hours of iteration.
- **Production-grade expectations**: Lovable apps work. They're not always idiomatic; the auth flow may need a manual review before serving real customer data. Treat the scaffold as a starting point, not the final state.

---

### Cloud IDE with Replit (Replit Agent in the browser. Zero local setup.)

URL: https://aistack.sh/stack/ai-app-builder-replit
Category: Code · Builder · Level: Beginner · Updated: 2026-05-05

Lovable and v0 both assume you have a dev environment. Replit doesn't. The Replit Agent runs the whole loop in the browser — codegen, file editor, terminal, DB, and deploy. The right pick when local-setup friction is the actual blocker.

Title: No local install, no terminal. Replit Agent builds, runs, and hosts the app in your browser.

**Tools in this stack:**

- Claude — Scope + tradeoff calls. Same scoping pass as the other variants. Cheaper to argue with Claude than to redo a Replit deploy.
  - Alternatives: ChatGPT
- Replit — Build + run + host. Browser IDE + Agent + DB + deploy. The full SDLC in one tab. Pricing scales with usage; the $20/mo Core covers most solo use.
  - Alternatives: Lovable, v0

**Cost (monthly USD):**

- Solo, hobby: $0/mo
- Solo, shipping: $40/mo
- Small team: $155/mo

**Workflow:**

1. **Scope with Claude** (Claude): Same as the other variants. 20 min before any Replit work.
2. **Repl + Agent prompt** (Replit): Create a Repl, hand the Agent the scope. It scaffolds the project structure and an initial working route.
3. **Iterate in the browser** (Replit): Agent + manual edits in the browser editor. Replit's Hot Reload makes the inner loop tight.
4. **Deploy + hosting** (Replit): One-click Replit Deployments. Custom domain available on paid tiers. PostgreSQL bundled.

**Example output:**

Non-developer founder, first SaaS: Built a niche scheduling tool without ever opening a terminal. Replit Agent + the browser editor handled everything. Took 3 weekends; first paying customer in week 4. The hosted-IDE friction-removal was the only reason this shipped.

**Pitfalls:**

- **Hitting Repl resource limits**: Long-running background jobs or heavy DB queries can hit Free / Starter limits. Upgrade Core before you onboard real users.
- **Treating the browser IDE as a long-term home**: Great for shipping v1. Some teams export to a local repo at v1.5 once they're past prototyping speed.

---

### UI-first with v0 (v0 for production-quality UI; you wire the backend)

URL: https://aistack.sh/stack/ai-app-builder-v0
Category: Code · Builder · Level: Beginner · Updated: 2026-05-05

The right shape when your bottleneck is UI iteration, not backend. v0's component output is production-quality (shadcn/ui idioms, Tailwind, accessibility), and Cursor or Claude Code handles the API + DB layer. Slower than Lovable to first deploy, but you keep full code ownership.

Title: v0 generates the React + shadcn UI; Cursor wires the backend + DB; Vercel deploys.

**Tools in this stack:**

- v0 — UI generation. Production-quality React + shadcn/ui. Drops directly into a Next.js app — no component-library mismatch.
  - Alternatives: Lovable
- Cursor — Backend glue. API routes, server actions, DB schema. Cursor's Composer is faster than v0 for non-UI code.
  - Alternatives: Claude Code
- Claude — Product + DB design. Schema design and product decisions before writing code. Same role as elsewhere.
  - Alternatives: ChatGPT

**Cost (monthly USD):**

- Solo, hobby: $0/mo
- Solo, shipping: $60/mo
- Full-time, multiple apps: $240/mo

**Workflow:**

1. **Sketch UI in v0** (v0): Prompt the screens you need. Iterate visually until the components feel right.
2. **Pull components into your repo** (Cursor): v0 export → Cursor opens the repo. Components ship as standard shadcn/ui — no lock-in.
3. **Wire the backend with Cursor** (Cursor): Server actions, DB schema (Drizzle / Prisma), auth (Clerk or Auth.js). Composer handles the multi-file glue.
4. **Deploy on Vercel** (Cursor): One click. The full stack runs on Vercel Functions; Postgres via Marketplace.

**Example output:**

Frontend-heavy SaaS, design-led founder: Founder cared more about the UI being right than about the time-to-first-deploy. v0 nailed the UI in 4 hours; Cursor did the backend in another 8. Total time to deploy: 1.5 days. Result: an app that doesn't look like a Lovable template.

**Pitfalls:**

- **Rebuilding what Lovable would have built for you**: If your differentiation isn't the UI, this stack costs you a day vs Lovable. Pick the right tool for the bottleneck.
- **v0 reinventing your component library mid-project**: Lock the design tokens / theme up front; v0 will happily generate slightly-different button variants every prompt.

---

## Goal: AI music for video

### AI music for video (recommended)

URL: https://aistack.sh/stack/ai-music-for-video
Category: Audio · Music · Level: Beginner · Updated: 2026-05-05

Suno + Udio fight for hook quality; CapCut handles the mix. Total spend: under $20/mo. Zero copyright strikes.

Title: Royalty-free, on-brand background music for short-form video. Generated, not licensed.

**Tools in this stack:**

- Suno — Primary track. Best hooks of the two; export full-mix and stems on Pro.
  - Alternatives: Udio
- Udio — Backup + alt vibes. Cleaner mix at high tempo; better with non-English vocals.
  - Alternatives: Suno
- CapCut — Mix + sync to video. Free, fast, syncs cuts to beat automatically.
  - Alternatives: Descript

**Cost (monthly USD):**

- Casual creator: $0/mo
- Daily shorts: $20/mo
- Multi-account agency: $45/mo

**Workflow:**

1. **Vibe brief** (Suno): 1-line description: 'lofi piano, sad, mid-tempo, instrumental'.
2. **Generate x4** (Udio): Both Suno and Udio, twice each. Pick the one with the best hook.
3. **Mix to video** (CapCut): CapCut: drop track, snap cuts to beat, duck under voiceover.

**Example output:**

Indie creator on TikTok, 220k followers: Switched from Epidemic Sound ($24/mo) to Suno + Udio. Saved $14/mo, got two of her tracks added to her own Spotify.

**Pitfalls:**

- **Vocal hallucination**: Suno occasionally produces near-real-sounding artist vocals. Generate instrumentals when in doubt.
- **Mix levels too hot**: Generative music is mastered loud; -6dB before voiceover saves your watch time.

---

### AI music for vlogs (Long-form vlogs: Udio-only, longer tracks, mixed in Descript)

URL: https://aistack.sh/stack/ai-music-vlog
Category: Audio · Music · Level: Beginner · Updated: 2026-05-05

Short-form b-roll music doesn't work for long-form: the same hook on loop kills retention. Udio handles longer instrumentals with cleaner mixes; Descript ducks the music under voiceover automatically.

Title: Music for 8 to 15-minute vlogs and YouTube essays. Cleaner mix than the short-form pipeline, single tool.

**Tools in this stack:**

- Udio — Long instrumentals. Pro plan generates 4-min tracks; cleaner mix at moderate tempos beats Suno for vlog-shaped audio.
  - Alternatives: Suno
- Descript — Mix + duck under VO. Studio Sound + automatic music ducking does what CapCut's free editor can't on long-form. Edit by transcript ties music transitions to what you say.
  - Alternatives: CapCut

**Cost (monthly USD):**

- 1 vlog/wk, free tiers: $16/mo
- 2 vlogs/wk, Udio Pro: $26/mo
- Daily long-form: $50/mo

**Workflow:**

1. **Brief, vlog-shaped** (Udio): One sentence. Long-form needs a track that sustains 4 minutes without a hard hook.
2. **Generate 4 versions** (Udio): Udio: 4 generations from one prompt. Pick the one that doesn't loop obviously inside 4 minutes.
3. **Mix in Descript** (Descript): Drop voiceover and the track. Auto-duck under voice. Cut on transcript boundaries so the music swells when you stop talking.
4. **Export** (Descript): 1080p, master at -14 LUFS for YouTube. Save the prompt + seed for the next episode.

**Example output:**

Solo essayist, 38k subs, 12-min average video: Switched from Epidemic Sound and the short-form Suno+Udio combo. Average watch time at the music change held within 0.5% of the previous track-driven cut. Saves $14/mo over Epidemic, and music never repeats across videos.

**Pitfalls:**

- **Tracks that are too 'composed'**: Generative music wants to climax every 30s. For vlogs you want it to be shapeless on purpose. Reject the first 2 generations from instinct.
- **Mixing too hot**: Long-form viewers leave when music sits within 4dB of the voice. Duck to -10dB minimum under VO; -6dB only in pure cutaways.

---

## Goal: AI newsletter

### AI newsletter (recommended)

URL: https://aistack.sh/stack/ai-newsletter
Category: Writing · Newsletter · Level: Beginner · Updated: 2026-05-05

Train Feedly to find the stories, Claude to summarize them, Beehiiv to send. Promote the issue on X via Typefully. The whole loop runs on $45/mo.

Title: An AI-curated newsletter you can ship every Tuesday in under two hours, end to end.

**Tools in this stack:**

- Feedly — Story discovery. Leo can be trained on your niche, then cuts 200 stories down to ~20 worth reading.
  - Alternatives: Perplexity
- Claude — Synthesis + voice. Claude handles long context. Feed it 20 articles, get a coherent issue out the other side.
  - Alternatives: ChatGPT
- Beehiiv — Send + monetize. Free up to 2,500 subs, plus referral mechanics + an ad network when you outgrow free.
  - Alternatives: Substack
- Typefully — Promo threads. One newsletter → one X thread → one LinkedIn post, all from the same draft.
  - Alternatives: Taplio

**Cost (monthly USD):**

- <2,500 subs: $25/mo
- 2,500–10k subs: $72/mo
- 50k+ subs, daily issues: $245/mo

**Workflow:**

1. **Curate the week** (Feedly): Browse the Feedly Leo digest; star ~20 worth syncing.
2. **Synthesize** (Claude): Claude merges starred items into your section template.
3. **Edit and design** (Beehiiv): Drop into Beehiiv. Use the same template every week.
4. **Promo loop** (Typefully): Typefully thread + LinkedIn post draft; schedule for Tue 9am ET.
5. **Ship** (Beehiiv): Send Tue 7am ET. Beehiiv recommendations, boosts, and ads kick in.

**Example output:**

'Latent Space', 2,800 subs, $0 → $1.4k MRR in 4 months: One issue/week, 90 minutes of writing time. Sponsorship rates lifted at 2k subs because the audience was tightly scoped to AI engineers.

**Pitfalls:**

- **Cold launch with no list**: Beehiiv recommendations only kick in after ~500 subs. Plan for a 90-day cold start.
- **AI-flavored writing tells**: Read every issue out loud before sending. Strip 'delve', 'in the rapidly evolving landscape', etc.
- **Niche too broad**: 'AI news' is dead on arrival. 'AI for HR teams in regulated industries' has a list.

---

### AI newsletter (Substack) (Substack: free to publish, network-driven discovery, paid subs over ads)

URL: https://aistack.sh/stack/ai-newsletter-substack
Category: Writing · Newsletter · Level: Beginner · Updated: 2026-05-05

Drop Beehiiv's ad network and referral mechanics for Substack's Notes feed and recommendations. Perplexity replaces Feedly for one-writer research scale. Total spend stays under $25/mo until you have paid subs to give back 10% of.

Title: A newsletter on Substack that compounds via the platform's own network instead of paid acquisition.

**Tools in this stack:**

- Perplexity — Story discovery. At one-writer scale you don't need Feedly's training loop. Perplexity Pro answers 'what's new in {niche} this week' with citations.
  - Alternatives: Feedly
- Claude — Synthesis + voice. Same drafting role as the Beehiiv stack. The voice library is what carries.
  - Alternatives: ChatGPT
- Substack — Publish + grow. Notes feed and recommendations push your work to readers already on Substack. Free to start; 10% of paid subs only when monetized.
  - Alternatives: Beehiiv
- Typefully — Promo threads. Same cross-post role. Substack's own Notes covers some of what X used to.
  - Alternatives: Taplio

**Cost (monthly USD):**

- Free, building list: $20/mo
- Pro research + paid subs: $53/mo
- Daily + paid tier monetized: $130/mo + 10%

**Workflow:**

1. **Weekly research pass** (Perplexity): Perplexity Pro: ask 3 to 5 narrowly scoped questions about your niche. Star answers with citations.
2. **Synthesize** (Claude): Claude turns starred items + your voice library into the issue.
3. **Publish to Substack** (Substack): Paste in. Use one consistent issue template; Substack's editor is opinionated, lean into it.
4. **Drop a Note + cross-post** (Typefully): Substack Note distills the lede into 280 chars; Typefully sends an X thread + a LinkedIn version from the same draft.

**Example output:**

Niche writer, 0 to 1,400 subs in 90 days: Free Substack newsletter. ~62% of growth came from Substack recommendations and Notes; 0% from paid acquisition. Crossed the 'paid tier worth turning on' line at month 4.

**Pitfalls:**

- **Substack network can also be quiet**: Recommendations only kick in once you're in the 'recommended by larger writers' graph. Like Beehiiv's recommendations, plan for a 60 to 90-day cold start.
- **10% take is real money at scale**: At $50k+/yr in paid subs, the platform fee is large enough to justify a Beehiiv migration. Crosswalk: ~$20k/yr in paid subs is roughly the indifference point including time cost.
- **Notes is its own job**: Active Notes posting drives most discovery. If you skip Notes, the network barely works for you.

---

## Goal: AI image generation for content + ads

### AI thumbnails + ad creative (recommended)

URL: https://aistack.sh/stack/ai-thumbnails
Category: Marketing · Visual · Level: Beginner · Updated: 2026-05-05

Midjourney owns the AI-image quality bar for marketing visuals. The bottleneck isn't the model — it's the prompt. Use Claude to turn a creative brief into a tight Midjourney prompt; render in Midjourney; finish with Canva for typography. The whole loop is 6 to 8 minutes per image.

Title: Generate brand-grade thumbnails, ad creative, and social images at the speed of typing — Midjourney for the image, Claude for the prompt, Canva for the text overlay.

**Tools in this stack:**

- Claude — Prompt engineer. A great Midjourney prompt is the difference between brand-grade and stock-looking. Claude turns a 1-line brief into a structured Midjourney prompt with style, mood, lighting, lens, aspect ratio, and the right --ar / --v flags.
  - Alternatives: ChatGPT
- Midjourney — Image generation. Frontier-quality output with the most consistent house style. Worth the $30/mo Standard for unlimited Relax-mode generation.
- Canva — Typography + brand polish. Midjourney can't render text reliably. Canva's templates handle the headline, the logo, the CTA — the parts that need pixel-precision and brand consistency.
  - Alternatives: Figma

**Cost (monthly USD):**

- Hobby: $30/mo
- Daily creator output: $50/mo
- Brand or agency: $95/mo

**Workflow:**

1. **Brief to Midjourney prompt** (Claude): One Claude turn. Brief in, structured prompt out.
2. **Generate in Midjourney** (Midjourney): Paste prompt. Generate 4 variants. Pick the strongest, run /upscale.
3. **Polish in Canva** (Canva): Drop the upscaled image into a Canva template. Add headline + logo + CTA. Export at the platform's native size.
4. **QA on the device** (Canva): Open the final asset on the actual device + platform you're posting to. Mobile thumbnails legibility is the most-missed step.

**Example output:**

YouTube creator, thumbnail CTR lift: Switched from Photoshop + stock to this stack. Average thumbnail CTR rose from 4.6% to 7.1% across the next 30 videos. Per-thumbnail time dropped from ~25 min to ~7 min.

**Pitfalls:**

- **Trusting AI to render text**: Midjourney still botches text in most images. Generate the visual; add text in Canva.
- **Relax mode at brand speed**: Midjourney Relax mode queues your generations; can take minutes during peak hours. Pay for Fast hours when you're shipping under deadline.
- **Brand drift**: Without a locked palette + style reference, Midjourney happily generates 4 'similar' images that don't match each other. Save your prompt + seed; reuse them.

---

## Goal: Build an Android app with AI

### Android app with AI (recommended)

URL: https://aistack.sh/stack/android-app-with-ai
Category: Code · Mobile · Level: Intermediate · Updated: 2026-05-05

Same shape as the iOS stack, swap the platform tools. Cursor handles Kotlin and Jetpack Compose well. Android Studio's Gemini integration covers the simple completions; AI agents in Cursor cover the multi-file work. Distribute via the Play Console after a one-time $25 fee.

Title: Ship a Kotlin app: Cursor and Claude for the code, Android Studio for build, signing, and Play Console upload.

**Tools in this stack:**

- Claude — Product + plan. Same role as the iOS stack. Spend the planning time before you open the IDE. Android also has Play Store policies that bite late-discovered scope (background location, accessibility services, foreground service types).
  - Alternatives: ChatGPT, Gemini
- Cursor — Kotlin / Compose editing. Strong Kotlin completions, Composer for cross-screen edits, and good interop with Gradle config files. Cursor handles Compose's declarative style well.
  - Alternatives: Claude Code, Android Studio
- Android Studio — Build, profile, sign, upload. Where the app actually builds and signs. Studio's Gemini is fine for inline completions if you want a second opinion. Profilers (CPU/Memory/Energy) are essential for the Play Store's Android Vitals section.

**Cost (monthly USD):**

- Solo, first app: $22/mo
- Solo, multiple apps: $42/mo
- Small team: $220/mo

**Workflow:**

1. **Plan in Claude** (Claude): Same scoping pass as the iOS stack — but the Play Store review concerns are different.
2. **Scaffold with Cursor** (Cursor): Cursor at the project root. Composer scaffolds screen-by-screen. Use Android Studio side-by-side for previews; the @Preview cycle is fast.
3. **Run on Emulator + a real device** (Android Studio): Android Studio: Pixel 9 emulator + at least one cheap real device (Samsung mid-range). Layout breakage on smaller / older devices is the most common Play Store quality complaint.
4. **Profile with Android Vitals** (Android Studio): Studio Profiler before release. The Play Console grades you on startup time, ANR rate, and battery — get them green before the first Play submission.
5. **Sign + upload** (Android Studio): Build → Generate Signed Bundle → upload to Play Console as Internal testing first. Promote to Production after a few days of internal testing.

**Example output:**

Indie shop, hobby app shipped in 3 weeks: Solo dev built a niche utility app. Cursor handled the bulk of Compose code; Studio Profiler caught a startup regression that would have failed Android Vitals. Internal testing for 5 days, promoted to production at week 4. Initial install rate: 1.2k organic in month 1.

**Pitfalls:**

- **Composer suggesting XML layouts**: AI agents trained on older Android codebases sometimes default to XML. Re-prompt for Compose; mixing XML and Compose works but doubles the surface area.
- **Forgetting the Play data-safety form**: Like Apple's nutrition labels, but more aggressive. Walk through every dependency with Claude before submitting. Mismatches between actual SDK behavior and the form get apps removed, not just rejected.
- **Treating Studio's Gemini as the agent**: Studio's Gemini integration is fine for inline completions but doesn't reach Cursor's Composer or Claude Code for cross-file edits. Don't fight Gemini; treat it as a faster Tab key.

---

## Goal: Autonomous AI agent (delegate a goal, get a deliverable)

### Autonomous research agent (recommended)

URL: https://aistack.sh/stack/autonomous-research-agent
Category: Research · Intelligence · Level: Intermediate · Updated: 2026-05-05

When the task is 'research this market and write a 3-page summary' or 'find me 50 leads in this niche with verified contact info', a chat LLM is the wrong tool — too much hand-holding. Manus runs the multi-step plan autonomously and returns a deliverable. Claude is the QA gate before you trust it.

Title: Delegate a multi-step research or analysis project to Manus, get a finished deliverable. Use Claude to QA before you act on it.

**Tools in this stack:**

- Manus — Autonomous executor. Plans, executes, and returns a deliverable for high-context multi-step work. Good for research, list building, lightweight analysis. The right pick when the task would otherwise be a dozen chat turns and a human aggregator.
- Claude — QA + sanity check. Read the Manus output critically. Cross-check 3 to 5 claims with citations. Don't act on autonomous output without a human-in-the-loop pass.
  - Alternatives: ChatGPT, Perplexity

**Cost (monthly USD):**

- Trial / occasional use: $20/mo
- Weekly research runs: $59/mo
- Daily delegated work: $219/mo

**Workflow:**

1. **Write the goal, not the steps** (Manus): Manus plans best when given the destination, not the route. 'Find me 30 EU-based SaaS companies in the HR space with public funding rounds in 2025' beats 'first search Crunchbase, then…'.
2. **Watch the first 2 to 3 steps** (Manus): Manus is autonomous, not magical. The first few steps tell you whether the plan is on the right track. Cancel + re-prompt if it's drifting.
3. **QA with Claude** (Claude): Don't act on autonomous output blind. Run a structured QA pass against the deliverable.

**Example output:**

GTM analyst, weekly market scan: Replaced 3 hours of manual Crunchbase / LinkedIn search per week with a Manus run. Quality was within 90% on the first pass after the QA step caught 4 fabricated funding rounds. The QA gate is non-optional.

**Pitfalls:**

- **Trusting autonomous output blind**: Manus hallucinates with the same physics as any LLM — just over more steps, with more confidence. The Claude QA step is the only thing keeping bad data out of your decisions.
- **Over-broad goals**: 'Research the EU SaaS market' is unbounded. Tight constraints make Manus useful; loose ones make it expensive.

---

## Goal: Bulk LLM automation on a budget

### Cheap bulk LLM automation (recommended)

URL: https://aistack.sh/stack/cheap-bulk-llm
Category: Code · Backend · Level: Intermediate · Updated: 2026-05-05

When you need to score reviews, classify support tickets, summarize emails at scale — Claude API costs add up. DeepSeek matches frontier reasoning on most batch tasks at ~10x cheaper. Use Claude to write the prompt and the eval set; switch to DeepSeek for production volume; sample-check with Claude monthly.

Title: Run thousands of LLM calls a day for the price of a coffee. DeepSeek for production volume, Claude as the calibration and eval surface.

**Tools in this stack:**

- Claude — Prompt design + golden eval set. Use Claude to draft the prompt, hand-label 30 to 50 golden examples, and define what 'correct' looks like for your task. This is the work where model quality matters most.
  - Alternatives: ChatGPT
- DeepSeek — Production volume calls. DeepSeek-V3 / R1 hit ~$0.14/M input + $0.28/M output. On most classification, summarization, and structured-extraction tasks, output quality is within 2 to 5% of Claude on the eval set. The 90%+ cost savings dominate at volume.
  - Alternatives: Mistral, Qwen

**Cost (monthly USD):**

- 10k calls/mo: $5/mo
- 1M calls/mo: $140/mo
- 10M calls/mo: $1,200/mo

**Workflow:**

1. **Draft the prompt with Claude** (Claude): Use Claude's free tier — for prompt iteration, the chat surface is fine. The model quality matters here.
2. **Smoke-test against the eval on both** (DeepSeek): Run the eval set through both Claude and DeepSeek. If DeepSeek scores within 5% of Claude on your rubric, ship DeepSeek. If not, narrow the prompt or the task scope before re-testing.
3. **Run production on DeepSeek** (DeepSeek): Standard OpenAI-compatible API. Most SDKs work by changing the base URL.
4. **Monthly sampling with Claude** (Claude): Sample 1 to 2% of DeepSeek outputs and re-score with Claude. Catches drift and prompt rot before they become a customer-facing problem.

**Example output:**

Customer-support ticket classifier: Classifying ~120k tickets/month into 14 categories. Claude API: ~$2,800/mo at this volume. DeepSeek: ~$210/mo. Eval-set quality: within 3.4% on F1 (Claude 91.2 vs DeepSeek 87.8). Saved ~$31k/year; spent the savings on better hand-labeling.

**Pitfalls:**

- **Skipping the eval set**: The whole bet of this stack is that DeepSeek matches Claude on YOUR task. Without an eval set, you're guessing. The eval set is the cheapest insurance possible.
- **Switching back to Claude on the first DeepSeek miss**: Production LLMs all hallucinate sometimes. Don't switch on n=1 — the right move is to widen the eval set and confirm a real quality gap before paying 10x more.
- **Treating DeepSeek as a closed black box**: It's offered as a hosted API, but the model weights are open. If data residency or supplier risk is a concern, you can self-host the same model on your own infra. Most teams won't, but knowing the option exists changes the conversation.

---

## Goal: Cold outbound

### Cold outbound (recommended)

URL: https://aistack.sh/stack/cold-outbound
Category: Sales · Outbound · Level: Intermediate · Updated: 2026-05-05

Apollo for the leads, Clay for the enrichment + per-row LLM, Smartlead to send at deliverability scale, Claude to write the copy, HubSpot to catch the meetings.

Title: Cold outbound that actually books meetings. Research, enrichment, copy, send, CRM.

**Tools in this stack:**

- Apollo — Lead source. 275M contacts. Filter by ICP, export to Clay.
  - Alternatives: Clay
- Clay — Enrichment + per-row LLM. Trigger-based personalization at scale. The competitive moat of modern outbound.
- Claude — Copy. Tighter, less salesy hooks than the alternatives. 4-line emails out-perform 8-line.
  - Alternatives: ChatGPT
- Smartlead — Send infra. Unlimited inboxes + warmup. The reason your sequences land in inbox not spam.
- HubSpot — CRM. Free tier covers small teams. Sequences + meetings on the same record.

**Cost (monthly USD):**

- Solo, 500 sends/mo: $120/mo
- 2 SDRs, 5k sends/mo: $340/mo
- 5 SDRs, 30k sends/mo: $1,150/mo

**Workflow:**

1. **Define ICP** (Apollo): Apollo: filter to your tightest ICP. Export 1k contacts.
2. **Enrich** (Clay): Clay: pull tech stack, headcount changes, recent funding. One LLM call per row to write the hook.
3. **Write copy** (Claude): Claude writes the 4-line email per persona. Test 3 hooks per ICP.
4. **Send** (Smartlead): Smartlead, 30 sends/inbox/day, 6 inboxes per persona. Warmup always on.
5. **Catch** (HubSpot): HubSpot sequences hand off to AE. Meetings booked = stop the sequence.

**Example output:**

B2B SaaS, $25k ACV: 1.4% reply rate, 0.6% meeting rate at 4k sends/month. 5 closed-won in Q3 from outbound alone, $130k ACV pipeline.

**Pitfalls:**

- **Inbox burn**: Send from inboxes you don't care about. One sequence's bad luck kills your real domain.
- **Over-personalized = creepy**: Reference one trigger, never two. Two reads as stalking.
- **No follow-up = no replies**: 60% of replies come from email 3 or 4. Sequence ≥ 5.

---

### Cold outbound (solo) (Solo founder: skip Clay, Smartlead, HubSpot.)

URL: https://aistack.sh/stack/cold-outbound-lite
Category: Sales · Outbound · Level: Beginner · Updated: 2026-05-05

When you're a one-person team sending under 500 emails a month, you don't need Clay enrichment or Smartlead infrastructure. Apollo basic for leads, Claude for the email, Gmail for delivery.

Title: Solo-founder cold outbound for under $80/mo. Apollo for leads, Claude for copy, Gmail to send.

**Tools in this stack:**

- Apollo — Lead source + light enrichment. Free tier or $49 Basic gives you the contacts plus what little enrichment you need at solo scale.
- Claude — Per-lead copy. Solo volume means you can hand-personalize. Claude drafts from a 1-line trigger you write yourself.
  - Alternatives: ChatGPT
- Gmail — Send. Use one secondary domain inbox, hand-send. Skip Smartlead until you exceed 500 sends/month.

**Cost (monthly USD):**

- 100 sends/mo, hand-sent: $26/mo
- 300 sends/mo: $75/mo
- 500 sends/mo (max before Smartlead): $105/mo

**Workflow:**

1. **Build a tight list** (Apollo): Apollo: filter to your tightest ICP. Cap the export at 100 contacts. At solo scale, list quality beats list size.
2. **One trigger per lead**: For each contact, jot one specific trigger (recent post, funding, hire, product launch). 30 seconds per lead.
3. **Draft each email** (Claude): Paste the trigger + ICP into Claude. One email per lead.
4. **Send + follow up by hand** (Gmail): Send from your secondary inbox. Two follow-ups: +3 days (1 line, sharper hook), +7 days (1 line, breakup). Stop there.

**Example output:**

Solo founder, ProServ B2B: 300 sends over 6 weeks: 8.4% reply rate (high, because hand-personalized), 14 booked calls, 3 closed at $4k average. Total tooling cost in that window: $90.

**Pitfalls:**

- **Gmail send limits + spam triggers**: Hard cap at 30 sends/day from a single inbox. Past 500/mo, deliverability collapses. That's when you graduate to the recommended stack.
- **Skipping the trigger**: Without the per-lead trigger this stack falls apart. The whole reason it works at solo scale is that each email earns the open. Don't paste-and-blast.

---

## Goal: Cross-platform iOS + Android app

### iOS + Android with Expo (recommended)

URL: https://aistack.sh/stack/cross-platform-mobile
Category: Code · Mobile · Level: Intermediate · Updated: 2026-05-05

If you're targeting both stores with a small team, native-per-platform is usually the wrong tradeoff. Expo's tooling has gotten good enough that you write TypeScript once, EAS Build compiles native bundles, and OTA updates ship JS-only fixes without re-submitting. Pair with Cursor + Claude and the loop is faster than either native track.

Title: One codebase, both app stores. Expo + React Native, Cursor for the code, Claude for the architecture, EAS Build for native compile.

**Tools in this stack:**

- Claude — Product + native-vs-cross decisions. Cross-platform tradeoffs are real (animations, deep platform integrations). Use Claude up front to mark which features genuinely need native vs which can ride on Expo's SDK.
  - Alternatives: ChatGPT
- Cursor — TypeScript + React Native editing. Cursor is great at TypeScript. React Native ecosystems are well-trained-on, so suggestions are reliable. Composer covers cross-file refactors when you change a navigation pattern or auth provider.
  - Alternatives: Claude Code, Codex
- Expo — Build infra + OTA updates. Expo Router for navigation, EAS Build for managed native compiling, EAS Update for shipping JS fixes without store review, EAS Submit for one-command store uploads. The infra you'd otherwise rebuild every project.

**Cost (monthly USD):**

- Solo, first app: $28/mo
- Solo, multiple apps + EAS Production: $67/mo
- Small team: $320/mo

**Workflow:**

1. **Decide cross-platform vs native** (Claude): Some apps shouldn't be Expo (heavy AR, custom video pipelines, deep iOS WidgetKit). Use Claude to audit upfront.
2. **Scaffold with Cursor + Expo Router** (Cursor): `npx create-expo-app`, then Cursor's Composer to scaffold tabs + screens. The Expo Router file-based routing makes the screen layout legible in one folder.
3. **Iterate with Expo Go on a real device** (Expo): Expo Go on your phone for the inner loop. EAS Build only when you need a real native build (signing, native modules, store submission).
4. **EAS Build for store builds** (Expo): `eas build --profile production --platform all` builds both iOS and Android binaries in the cloud. ~15 min the first time, faster after caching.
5. **EAS Submit + EAS Update** (Expo): `eas submit` uploads to TestFlight + Play Console in one command. `eas update` ships JS fixes after launch without re-submitting (within Apple's policy boundary).

**Example output:**

Two-person startup, ship to both stores in 5 weeks: Could not have justified two native codebases. Expo + Cursor put the team at parity with native-only competitors on screens-shipped-per-week. The only feature they had to drop was a custom haptic pattern that wasn't in expo-haptics; they shipped without it and added a config plugin in v2.

**Pitfalls:**

- **Picking Expo for the wrong app**: If your app's core is AR, custom audio, or watchOS / Wear OS companions, Expo will fight you. The 'fitness audit' step exists to catch this before you've written 10k lines.
- **Skipping EAS Update planning**: OTA updates are the killer Expo feature, but they live inside Apple's policy boundary (no major feature changes via JS). Plan from day one which features are JS-only and which need a native build + submission.
- **Native module rabbit holes**: When a native module isn't in expo-* or doesn't have a config plugin, the temptation is to eject to a bare workflow. Don't, unless you're sure. Expo's prebuild + config plugins almost always cover it.

---

## Goal: Faceless YouTube

### Faceless YouTube (recommended)

URL: https://aistack.sh/stack/faceless-youtube
Category: Video · Creator · Level: Intermediate · Updated: 2026-05-05

An end-to-end faceless channel pipeline. Topic research, scripts in your voice, AI voiceover, generative B-roll, captions, upload. Start to finish in an afternoon.

Title: Research, script, voiceover, B-roll, edit. The assembly line behind 12 channels at >100k subs.

**Tools in this stack:**

- Perplexity — Topic + research. Cited research synth means scripts cite real sources, not hallucinated ones.
  - Alternatives: Gemini
- Claude — Script + hook. Claude writes the cleanest narrator voice and respects pacing in long scripts.
  - Alternatives: ChatGPT
- ElevenLabs — Voiceover. Cinematic-grade voice cloning. Clone yourself once and you never have to record.
  - Alternatives: openai-voice
- Runway — Generative B-roll. Gen-3 b-roll fills the gaps when stock footage doesn't exist.
  - Alternatives: Pika
- CapCut — Edit + captions. Free, fast, AI captions. Slot the audio + b-roll, export at 1080p.
  - Alternatives: Descript

**Cost (monthly USD):**

- 1 video / week: $45/mo
- 3 videos / week: $92/mo
- Daily uploads, 2 channels: $310/mo

**Workflow:**

1. **Pick a topic** (Perplexity): Use Perplexity to scope a topic and collect 6 to 10 cited sources.
2. **Write the script** (Claude): Claude turns the sources into a 1,200-word narration with a hook and retention beats.
3. **Generate voiceover** (ElevenLabs): Cloned voice in ElevenLabs. Render at 192kbps mono.
4. **B-roll pass** (Runway): Runway Gen-3 for any shot you can't find on Pexels. 4–6s clips per beat.
5. **Edit + caption + ship** (CapCut): CapCut: slot audio, drop b-roll, AI captions, export. Upload with thumbnail.

**Example output:**

'How a Cold War spy died from a poisoned umbrella', 1.4M views in 3 weeks: Total time start-to-upload: 3.5 hours. RPM landed at $4.20 because the niche was history. Channel was 6 weeks old.

**Pitfalls:**

- **Generic AI voice = no retention**: Clone your own voice or buy a paid pro voice. The 'free Eleven voices' kill watch time within 30s.
- **Generative B-roll uncanny valley**: Cap each Runway clip at 4s; cut on motion. Long generative shots look obviously off.
- **Copyright on uncited sources**: Don't shortcut Perplexity's citations. Claims without sources get demonetized fast.

---

### Faceless YouTube (budget) (Zero-budget starter. Free voices, stock b-roll.)

URL: https://aistack.sh/stack/faceless-youtube-budget
Category: Video · Creator · Level: Beginner · Updated: 2026-05-05

When you're testing whether a channel concept will work at all, this is the cheapest viable pipeline. Trade Runway for Pexels stock, ElevenLabs Pro for the free voice tier, and Perplexity Pro for free Claude.

Title: The same faceless-YouTube pipeline at $0/mo. Free LLMs, free voice, stock b-roll.

**Tools in this stack:**

- Claude — Research + script. Free Claude tier handles topic research and 1,200-word scripts. Skip Perplexity until you're sure the niche works.
  - Alternatives: ChatGPT
- ElevenLabs — Voiceover. Free tier gives you 10k characters/month, enough for ~3 short scripts. Use a marketplace voice, not a clone.
- CapCut — Edit + b-roll + captions. Free, ships with stock footage built in plus AI captions and auto-cut to beat. The whole pipeline collapses into one app.
  - Alternatives: Descript

**Cost (monthly USD):**

- 1 video / week: $0/mo
- 2 videos / week: $5/mo
- Daily uploads: $25/mo

**Workflow:**

1. **Pick a topic + script** (Claude): One Claude session: research the topic, then write the script. Skip the dedicated research tool.
2. **Generate voiceover** (ElevenLabs): ElevenLabs free tier. Pick a marketplace voice, paste the script, render at default settings.
3. **Edit + ship** (CapCut): CapCut: drop voiceover, use the built-in stock library for b-roll (search by [B-ROLL] tags), auto-captions, export 1080p, upload.

**Example output:**

First test channel, week 4: Solo creator validating a 'historical disasters' niche. Shipped 4 videos at $0 cost; the third hit 12k views and confirmed the angle. Then they upgraded to the recommended stack.

**Pitfalls:**

- **Stock b-roll loops fast**: CapCut's stock library is shallow per-niche. After 3 to 4 videos you'll see the same clips repeat. Move to Runway/Pika once you commit.
- **Free voices flatten retention**: Marketplace free voices kill watch time within the first 30 seconds for narrative content. Validate with this stack, then pay for ElevenLabs Creator.

---

## Goal: Inbox triage

### Inbox triage (recommended)

URL: https://aistack.sh/stack/inbox-triage
Category: Productivity · Inbox · Level: Beginner · Updated: 2026-05-05

Superhuman handles the keyboard speed, Claude drafts replies in your voice, Granola backfills meeting notes so emails never re-ask the question.

Title: Inbox triage for solo founders. 30 minutes back, every morning.

**Tools in this stack:**

- Gmail — Underlying inbox. All the AI front-ends still bolt onto Gmail. Keep it as the source of truth.
- Superhuman — Keyboard-first triage. Splits + snippets + scheduled send + AI compose. Everything is a hotkey.
- Claude — Reply drafts. Better tone-matching than the built-in AI; you paste in your last 5 replies for voice.
  - Alternatives: ChatGPT

**Cost (monthly USD):**

- 1 inbox: $18/mo
- 1 inbox + assistant: $36/mo
- EA + 3 inboxes: $120/mo

**Workflow:**

1. **Splits** (Superhuman): Superhuman: VIP / Newsletters / Action. Triage in that order daily.
2. **Drafts** (Claude): Claude in a side window. Paste the email + 3 of your past replies; ask for a draft.
3. **Schedule sends** (Superhuman): Send between 7–9am local. Reply rates jump.

**Example output:**

Solo founder, 220 emails/day: From 90 min/day on email to 35 min/day. 0 inbox-zero panic days in the last 3 months.

**Pitfalls:**

- **Auto-replies sound generic**: Don't ship Claude's first draft. Always one human pass.
- **Snippet sprawl**: Cap snippets at 30. Past that you can't find them anyway.

---

### Inbox triage (free) (No Superhuman. Just Gmail + Claude in a side window.)

URL: https://aistack.sh/stack/inbox-triage-free
Category: Productivity · Inbox · Level: Beginner · Updated: 2026-05-05

If you can't justify Superhuman yet, Gmail's filters + canned responses + a Claude side window covers 80% of the value. Two filters do the splits Superhuman charges for.

Title: Inbox triage without paying $25/mo for Superhuman. Gmail filters do the splits, Claude drafts the replies.

**Tools in this stack:**

- Gmail — Inbox + filters. Filters labelled VIP / Newsletters / Action handle the triage Superhuman calls 'Splits'. Snooze is built in.
- Claude — Reply drafts. Side window with your voice samples. Paste email in, paste draft out. The hotkey is Cmd+Tab.
  - Alternatives: ChatGPT

**Cost (monthly USD):**

- Personal Gmail: $0/mo
- 1 inbox + Pro Claude: $26/mo
- EA + 3 inboxes: $78/mo

**Workflow:**

1. **Set up 3 filters** (Gmail): Gmail filters: VIP (named senders), Newsletters (list-unsubscribe header present), Action (you're in To, not CC). Triage in that order.
2. **Drafts in side window** (Claude): Claude in a browser tab. Pinned. Paste email + 3 of your past replies for voice; paste draft back into Gmail.
3. **Schedule sends manually** (Gmail): Gmail's Schedule send (the small caret next to Send). Send between 7 to 9am local for higher reply rates.

**Example output:**

Indie consultant, 90 emails/day: From 60 min/day on email to 35 min/day. Tested Superhuman at month 3, kept the pipeline; the only gap was raw keyboard speed.

**Pitfalls:**

- **Filter rot**: Gmail filters are write-once-forget. Re-audit every 2 months or you'll find the VIP filter quietly missed your top customer.
- **Tab-switching tax**: Without Superhuman's keyboard composition, every reply costs you the Cmd+Tab. If your inbox crosses 200 emails/day, that tax compounds. Upgrade then.

---

## Goal: Indie SaaS in a weekend

### Indie SaaS (recommended)

URL: https://aistack.sh/stack/indie-saas
Category: Code · Builder · Level: Advanced · Updated: 2026-05-05

Claude is the pair, Figma is the design surface, Vercel is the deploy target. End-to-end indie SaaS from idea to billing in a weekend.

Title: Ship a SaaS by Sunday. Design, scaffold, build, deploy without leaving the AI loop.

**Tools in this stack:**

- Claude — Pair programmer. Long context + best code edits. Use a project per repo with the codebase as context.
  - Alternatives: ChatGPT
- ChatGPT — Quick lookups + voice. Faster for one-off questions; better voice mode for thinking-out-loud sessions.
  - Alternatives: Claude
- Figma — UI sketching. AI plugins draft from text; you only have to refine.
  - Alternatives: Canva

**Cost (monthly USD):**

- Weekend MVP: $40/mo
- Side project: $75/mo
- Full-time founder: $220/mo

**Workflow:**

1. **Sketch UI** (Figma): Figma + AI plugin: text to wireframe. 30 minutes of cleanup.
2. **Scaffold** (Claude): Claude in IDE: scaffold the full stack from the wireframe.
3. **Iterate live** (ChatGPT): Voice mode in ChatGPT for thinking-out-loud while debugging.
4. **Ship** (Claude): Deploy to Vercel. First sale before bedtime.

**Example output:**

Indie hacker, 2 days, $300 MRR by month 2: Niche SaaS for property inspectors. ~$40/mo tooling cost; the rest is hosting + Stripe.

**Pitfalls:**

- **AI-shaped slop UI**: Default Tailwind + shadcn looks generic; spend a half day on type scale + spacing.
- **Skipping the 'why this'**: Founders ship the MVP and forget the landing page. Write the page first, the product second.

---

### Indie SaaS (Gemini long-context) (Whole-codebase pair programming. Paste the repo, get architectural diffs back.)

URL: https://aistack.sh/stack/indie-saas-gemini
Category: Code · Builder · Level: Advanced · Updated: 2026-05-05

Different ergonomic from the Claude-led pair: instead of pasting individual files per turn, dump the entire repo into Gemini and let it reason across the whole graph. Claude takes over for tight, surgical edits; Gemini handles 'change auth across 14 files' moves.

Title: Ship a SaaS by Sunday with Gemini's 2M-token context window holding the whole codebase as one file.

**Tools in this stack:**

- Gemini — Whole-codebase reasoning. 2M+ token context lets you paste the entire repo (or 'git ls-files | xargs cat') as part of every prompt. Best when changes span many files.
  - Alternatives: Claude
- Claude — Surgical edits. When you know the file and the diff is local, Claude is faster and cheaper per turn. Use it for the last-mile edits.
  - Alternatives: ChatGPT
- Figma — UI sketching. Same role as the Claude-led stack. AI plugins draft from text; you refine.
  - Alternatives: Canva

**Cost (monthly USD):**

- Weekend MVP: $40/mo
- Side project: $75/mo
- Full-time founder: $240/mo

**Workflow:**

1. **Sketch UI** (Figma): Figma + AI plugin: text to wireframe. Same as the Claude-led stack.
2. **Dump the repo** (Gemini): From your terminal: `git ls-files | xargs cat > /tmp/repo.txt`. Paste that as one block into Gemini.
3. **Surgical edits with Claude** (Claude): Once Gemini gives you the cross-cutting plan, switch to Claude for individual file edits and bug fixes. Cheaper per turn, faster for local work.
4. **Voice mode for debugging** (Gemini): Same as the recommended stack. Gemini's voice mode is fine here; the value is thinking out loud, not the LLM.
5. **Ship** (Gemini): Deploy to Vercel. First sale before bedtime.

**Example output:**

Indie hacker rewriting auth in one session: Swapped JWT-in-cookie for Postgres-backed sessions across 22 files in 90 minutes by pasting the whole repo into Gemini once and following its migration order. The Claude-led approach took 4 hours on a previous similar refactor because each file needed its own context paste.

**Pitfalls:**

- **Pasting secrets into the context window**: `git ls-files | xargs cat` will happily include any .env files you committed. Audit before pasting; better, use `git ls-files -- ':!*.env*' ':!.git/*'`.
- **Long-context drift on edits**: Gemini occasionally rewrites unrelated files when given the whole repo. Diff before you commit; reject any change you didn't ask for.
- **Cost spikes at the API tier**: 2M-token contexts get expensive fast on the API. Stay on the $20 chat tier for repos under ~150k LOC; switch to API only for bulk processing.

---

## Goal: Build an iOS app with AI

### iOS app with AI (recommended)

URL: https://aistack.sh/stack/ios-app-with-ai
Category: Code · Mobile · Level: Intermediate · Updated: 2026-05-05

Cursor handles Swift / SwiftUI better than the terminal-first agents because most iOS work involves visual layout and the codegen needs to see selection state. Xcode does what only Xcode can: simulators, signing, profiling, App Store Connect. Claude (chat) is where you decide what to build before either tool starts editing.

Title: Ship an iOS app from scratch: Cursor for Swift code, Xcode for build/sign/TestFlight, Claude for product decisions.

**Tools in this stack:**

- Claude — Product + plan. Before you open Xcode, decide what you're building. Claude is the chat surface for the design/info-architecture/data-model conversation. Especially valuable on iOS where over-scoping is costly (every screen needs design polish to look native).
  - Alternatives: ChatGPT, Gemini
- Cursor — Swift / SwiftUI editing. Inline AI completions for Swift, Composer for multi-file changes, and decent SwiftUI preview integration. Cursor's hold on Swift code quality has improved a lot — better than copy-pasting into a chat tab, comparable to Claude Code for typical iOS work.
  - Alternatives: Claude Code, Codex
- Xcode — Build, sign, run, ship. Where the app actually compiles and runs. Required for Simulator, Instruments (profiling), Signing & Capabilities, and App Store Connect distribution. AI agents drive Cursor; you drive Xcode.

**Cost (monthly USD):**

- Solo, 1 indie app: $28/mo
- Solo, multiple apps + ads: $48/mo
- Small team, 3 to 5 devs: $220/mo

**Workflow:**

1. **Sketch the app in Claude** (Claude): Before any code: scope the screens, the data model, the privacy story (everything Apple cares about during review).
2. **Scaffold in Cursor** (Cursor): Open Cursor at the project root. Use Composer (agent mode) to scaffold the screens from step 1. Don't accept the whole diff at once — go screen-by-screen so SwiftUI previews stay healthy.
3. **Run in the Simulator** (Xcode): Cmd-R in Xcode. Cursor's preview is fast; Xcode's Simulator is the truth. If the layout breaks at smaller sizes (SE) or with Dynamic Type at the largest setting, fix before moving on — App Review reads accessibility.
4. **Sign + archive + upload** (Xcode): Xcode → Product → Archive → Distribute App → App Store Connect. Once: set up automatic signing in Signing & Capabilities. TestFlight build is ready in ~10 min.
5. **Submit for review** (Claude): Use Claude to draft the App Store description, the privacy nutrition label answers, and the review-notes message. App Store rejections are usually metadata or privacy issues, not code.

**Example output:**

Indie dev, 2 apps shipped in a quarter: Both apps built solo. Cursor handled ~80% of the SwiftUI; the dev wrote Xcode signing config + the privacy label by hand. Both apps cleared App Review on the first submission, mostly because the privacy story was scoped tight from the Claude planning step.

**Pitfalls:**

- **AI-generated UIKit when SwiftUI was simpler**: Cursor's first instinct is occasionally to drop into UIKit for things SwiftUI can do natively. Reject and re-prompt; mixing UIKit and SwiftUI early adds 2x the maintenance cost.
- **Skipping the privacy nutrition label**: Apple rejects more apps for privacy-label mismatches than for actual bugs. Use Claude to walk through every framework you import and what data it touches — get the answers right before submitting.
- **Treating the Xcode UI as optional**: Some terminal-only devs try to build with `xcodebuild` and skip Xcode. You can — but you'll fight signing, capabilities, and App Store Connect issues that the GUI surfaces clearly. Use Xcode for the deploy lane even if your editor is Cursor.

---

## Goal: LinkedIn ghostwriting

### LinkedIn ghostwriting (recommended)

URL: https://aistack.sh/stack/linkedin-ghostwriting
Category: Writing · Agency · Level: Intermediate · Updated: 2026-05-05

Run a one-person LinkedIn ghostwriting agency for a busy founder. Voice interviews → drafted posts in their voice → carousels → analytics, all on one Sunday a month.

Title: The exact stack ghostwriters use to scale a B2B writing practice without burning out.

**Tools in this stack:**

- Granola — Voice + idea capture. 30-min monthly interview with the client recorded as background audio. Granola produces structured notes you can feed straight into Claude.
  - Alternatives: fireflies
- Claude — Drafting + voice match. Claude is the LLM with the most consistent voice match for editorial-grade prose. Use a project with the client's last 30 high-performing posts as context.
  - Alternatives: ChatGPT, Gemini
- Taplio — Scheduling + viral search. Where the posts actually go out. Their viral post search seeds hooks; analytics close the loop.
  - Alternatives: Typefully
- Canva — Carousel + image posts. Templates do the heavy lifting. The bottleneck is the writing, not the design.
  - Alternatives: Figma

**Cost (monthly USD):**

- 1 client · 4 posts/wk: $77/mo
- 3 clients · 12 posts/wk: $129/mo
- 8 clients · 30 posts/wk: $420/mo

**Workflow:**

1. **Monthly voice interview** (Granola): Record a 30-min open-ended call with the client. Granola captures it.
2. **Pull post angles** (Claude): Feed the notes into Claude with the brand voice doc; ask for 20 hooks then narrow to 12.
3. **Draft posts** (Claude): One prompt per post. Maintain a voice library of past posts in the project.
4. **Design carousels** (Canva): Slot copy into a Canva template; export at 1080×1350.
5. **Schedule + ship** (Taplio): Stage in Taplio; auto-comment with a hook variant; review analytics weekly.

**Example output:**

Founder with 18k followers, 4 months in: Average post 1,400 impressions before, 11,200 impressions after. Booked 9 inbound leads in month 4 from a single 'lessons learned' post. Total writer time: ~6 hours/week per client.

**Pitfalls:**

- **Voice drift after model updates**: Re-anchor every 6 weeks by feeding the most recent 10 high-performers back into the project context.
- **Auto-commenting at scale flags spam**: Cap auto-comments at 1/post on Taplio; vary the comment so it doesn't look templated.
- **One client owns the relationship**: If your client churns, the LinkedIn account goes too. Diversify with at least 3 clients.

---

### LinkedIn (for yourself) (Solo: writing for your own account, not clients.)

URL: https://aistack.sh/stack/linkedin-personal
Category: Writing · Personal · Level: Beginner · Updated: 2026-05-05

Drop the multi-client agency tooling. For your own account you don't need Granola interviews or Taplio's viral search. Claude with your own voice library, Typefully to stage and ship, that's it.

Title: Post on LinkedIn 3 times a week without an agency stack. Just Claude, Typefully, and your own voice.

**Tools in this stack:**

- Claude — Drafting. Project with your last 30 posts as voice anchor. You write the angles (you live them), Claude drafts.
  - Alternatives: ChatGPT
- Typefully — Stage + schedule + post. Cleaner editor than LinkedIn's, schedules to LinkedIn + X from one draft. Free tier covers 3 posts/week.
  - Alternatives: Taplio

**Cost (monthly USD):**

- 3 posts/wk: $20/mo
- Daily + threads: $33/mo
- Daily + carousels: $48/mo

**Workflow:**

1. **Capture angles in a notes file** (Claude): When something happens at work, jot a 1-line angle. You don't need a 30-min interview if you live the work.
2. **Draft posts in batch** (Claude): Sunday session: feed Claude 5 angles + your voice library, get 5 drafts.
3. **Stage in Typefully** (Typefully): Paste, schedule across the week. Typefully cross-posts to X if you want a thread version.

**Example output:**

Solo founder, 4k followers: From posting twice a month to 3x/week. Reach went from ~600 impressions/post to ~3.2k impressions/post over 6 weeks. 11 inbound messages per month from posts alone.

**Pitfalls:**

- **AI-flat voice**: Without an interview to ground in, drafts drift to generic LinkedIn-speak fast. Keep refreshing the voice library every 4 weeks.
- **Posting frequency death spiral**: Daily posting without batching burns out within a month. Sunday batching (5 posts in 90 min) is the only thing that holds.

---

## Goal: Live trend + news research

### Live trend research (recommended)

URL: https://aistack.sh/stack/live-trend-research
Category: Research · Intelligence · Level: Beginner · Updated: 2026-05-05

The web's lag and the chat-LLM's training cutoff make 'what's happening this week' research painful. Grok pulls live X chatter, Perplexity grounds in cited sources, Claude turns the two streams into a 5-bullet brief you can hand a colleague. 30-min weekly habit replaces 4 hours of tab-switching.

Title: Real-time research that combines what X is talking about right now with cited primary sources and a synthesis you can act on.

**Tools in this stack:**

- Grok — Live X + xAI search. Real-time access to X with native search. The signal Perplexity and Claude can't reach. Bundled with X Premium+ at $30/mo, so if you already have the X subscription, the research add-on is free.
  - Alternatives: ChatGPT
- Perplexity — Cited primary sources. Grok tells you what people are saying; Perplexity tells you what actually happened (with links). Pair them: catch the wave from Grok, verify from Perplexity.
- Claude — Synthesis into a brief. Long-context turn that takes both feeds and outputs a 5-bullet brief. Same prompt every week makes the briefs comparable.
  - Alternatives: ChatGPT, DeepSeek

**Cost (monthly USD):**

- Solo, weekly habit: $0/mo
- All three at Pro: $70/mo
- Multi-niche analyst, daily briefs: $130/mo

**Workflow:**

1. **Live pulse from Grok** (Grok): Ask Grok what people are saying about your topic this week, with a real-time anchor.
2. **Verify with Perplexity** (Perplexity): For each angle from Grok, pull the underlying primary source. Don't synthesize on rumor.
3. **Synthesize with Claude** (Claude): Both outputs into Claude. One brief out the other side.

**Example output:**

Solo VC analyst, weekly habit: Replaced ~4 hours of tab-switching with a 30-min Friday habit. The Grok step caught a sentiment shift on a portfolio company 2 days before the public-news cycle. The combined brief now circulates internally; partners who used to skim Twitter manually skip that pass entirely.

**Pitfalls:**

- **Treating Grok's pulse as truth**: X-trending isn't reality. The Perplexity step exists to ground the noisy signal. Skipping it means publishing rumor.
- **Brief format drift week-over-week**: The value compounds when briefs are comparable across weeks. Lock the prompt + the structure; only change after a quarter.

---

## Goal: Multilingual content production

### Multilingual content factory (recommended)

URL: https://aistack.sh/stack/multilingual-content
Category: Writing · Content · Level: Intermediate · Updated: 2026-05-05

Most teams writing multilingual content default to one frontier LLM and accept slightly off-tone output in non-English languages. The cost: native readers spot it. The fix: route by language. Qwen handles Chinese / Japanese / Korean with depth no Western model matches; Mistral covers the European languages with EU-native nuance; Claude handles the English source.

Title: Produce content in 5+ languages without sounding translated. Each language to the model that actually speaks it well.

**Tools in this stack:**

- Claude — English source + brand voice anchor. Write the canonical version in English. Claude is the voice anchor — every translation reads back to it for tone match.
  - Alternatives: ChatGPT
- Qwen — ZH / JP / KR localization. Trained on a much larger Asian-language corpus than Western frontier models. Idioms, register, and brand-tone match are noticeably better. Free chat tier covers solo creator volume; API via DashScope for batch.
  - Alternatives: ChatGPT
- Mistral — FR / ES / DE / IT localization. EU-native model, strong European-language nuance. EU data residency is a side benefit if your audience is in EMEA.
  - Alternatives: Claude, ChatGPT

**Cost (monthly USD):**

- Solo creator, low volume: $20/mo
- Daily multilingual posts: $60/mo
- Agency, 10+ markets: $320/mo

**Workflow:**

1. **Write the English source in Claude** (Claude): Use the same voice library / brand prompts you use for English-only content. The English version is the canonical anchor.
2. **Localize Asian languages with Qwen** (Qwen): Send the English source + a brand-voice note to Qwen for each Asian-language target.
3. **Localize EU languages with Mistral** (Mistral): Same prompt shape, point Mistral at French / Spanish / German / Italian. Mistral's European-language nuance saves the QA pass.
4. **Native-speaker review** (Claude): Even the best LLM localization needs one native-speaker review per language per quarter. Don't ship blind; you only need to catch regressions, not edit every line.

**Example output:**

EdTech SaaS, 7 markets: Was running everything through Claude with Google-Translate-style results in non-English markets. Local NPS in DE, JP, and KR jumped after switching to this routing. The clearest win: marketing copy that local readers stopped flagging as 'AI-translated'.

**Pitfalls:**

- **Trusting one model for every language**: Every frontier LLM has language gaps. Routing matters more than picking 'the best' single model.
- **Skipping the native-speaker review**: LLMs hallucinate cultural references. The quarterly review is the cheapest way to catch the things models don't know they don't know.
- **Romanizing Asian-language brand names**: Qwen is good enough that localized brand mentions feel natural. Pinning the brand name in Latin script kills that. Lock per-language brand glossaries in the prompt.

---

## Goal: Self-hostable, open-weight LLM

### Open-weight LLM stack (recommended)

URL: https://aistack.sh/stack/open-weight-llm
Category: Code · Backend · Level: Advanced · Updated: 2026-05-05

Closed-weight APIs are the right default for most teams. They're not the right default for a regulated industry, a privacy-led product, or anyone whose customers ask 'where does this data go'. Llama and Mistral are open-weight families that ship near-frontier quality with the receipts. Use Claude on a small eval set to confirm the open-weight model holds up on YOUR task.

Title: Run frontier-grade LLMs on weights you can inspect, self-host, and audit. Llama for general work, Mistral when EU data residency matters, Claude as the calibration eval.

**Tools in this stack:**

- Meta AI — Default open-weight model. Llama 3.x and 4 cover most general tasks. Hosted on Groq for fast inference, on Together for breadth, or self-hosted on your own GPUs via Ollama / vLLM. Open weights mean you keep the option to leave any provider.
  - Alternatives: Mistral, Qwen
- Mistral — EU data residency + code. Mistral Large for general, Codestral for code. EU-headquartered, EU data centers available. The right pick when the legal team is in the room.
  - Alternatives: Meta AI
- Claude — Eval anchor + spot-check. Use Claude on a 30 to 50-row eval to confirm your open-weight pick holds quality on your specific task. Re-run quarterly or whenever you swap providers.
  - Alternatives: ChatGPT

**Cost (monthly USD):**

- Hosted inference, low volume: $15/mo
- Hosted inference, real volume: $170/mo
- Self-hosted on rented GPUs: $1,200/mo

**Workflow:**

1. **Pick the model for the use case** (Meta AI): Llama 3.x for general; Llama 4 for the heavy reasoning tasks; Mistral Large for EU + nuanced French/Spanish; Codestral when the task is code generation specifically.
2. **Pick the host** (Meta AI): Don't self-host until you have to. Groq is fastest for Llama; Together is broadest. Mistral La Plateforme for hosted Mistral. Self-host only when data-residency or cost-at-scale forces you.
3. **Build a 50-row eval with Claude** (Claude): Use Claude (or GPT) to draft inputs + expected outputs. Hand-edit to remove ambiguity. This eval is the only way to know your open-weight pick is truly good enough.
4. **Smoke-test on the eval** (Mistral): Run the eval on Claude AND your open-weight pick. If the open-weight model scores within 5% (or your domain's tolerance), ship it. If not, narrow scope or pick a different open-weight model.
5. **Production + monthly drift check** (Claude): Ship to prod. Re-run the eval monthly to catch silent quality drift when the host updates the model.

**Example output:**

Healthcare-adjacent SaaS, EU customers: Could not use closed-weight US-hosted APIs for the customer-data path. Built on Mistral La Plateforme (EU region) for the prod calls, kept Claude for the eval rig. Internal compliance team signed off in week 2 because the audit trail (model card + EU hosting) was clean.

**Pitfalls:**

- **Self-hosting before you have to**: Renting a GPU box for one app is rarely a good trade. Hosted Llama (Groq, Together) covers most cases at lower cost. Self-host only when audit / latency / cost-at-scale forces it.
- **Treating 'open weights' as a feature without a use case**: Open weights only matter if you'd actually self-host or audit. If you'll never inspect them, the closed-weight APIs are usually still the right call. Be honest about which camp you're in.
- **Forgetting the eval drift**: Hosted open-weight models update quietly. The monthly eval is the cheapest way to catch a regression before a customer does.

---

## Goal: Ship code with AI

### Antigravity (Google) (Google's AI IDE: Gemini 2M-context inside the editor)

URL: https://aistack.sh/stack/ship-code-antigravity
Category: Code · Workflow · Level: Intermediate · Updated: 2026-05-05

Antigravity is Google's answer to Cursor. The differentiator: Gemini's 2M-token context window means the IDE can hold the entire repo in its working memory. Best fit when your codebase is big and your platform is already Google.

Title: Google's IDE on top of Gemini's 2M-context. Especially clean if your stack is GCP, Firebase, Android, or Workspace.

**Tools in this stack:**

- Antigravity — Plan + write + iterate. IDE-shaped surface backed by Gemini. Built-in planner agent + editor agent + review pass. Best when the repo is large enough that smaller-context IDEs lose the thread.
  - Alternatives: Cursor, Claude Code, Codex
- Claude — Cross-vendor diff review. Antigravity's built-in review pass uses Gemini — same model that wrote the code. A second opinion from Claude (different lineage, different priors) catches what Gemini's self-review wouldn't. Paste the diff + ticket; ask for a structured review.
  - Alternatives: CodeRabbit

**Cost (monthly USD):**

- Solo OSS, preview tier: $0/mo
- Solo dev, full IDE features: $40/mo
- Team on Google Cloud: $200/mo

**Workflow:**

1. **Open the repo in Antigravity** (Antigravity): Same as VS Code. The first index pass takes a minute or two for a large repo; subsequent prompts use the cached context.
2. **Hand the planner the ticket** (Antigravity): Antigravity's planner agent reads the whole repo. Generates a plan that respects existing patterns.
3. **Editor agent applies the plan** (Antigravity): Multi-file edits with diffs you accept hunk-by-hunk. The 2M context means it rarely loses the thread mid-refactor.
4. **Cross-vendor review with Claude** (Claude): Push the branch and grab the diff. Paste it into Claude with the ticket; ask for a structured review. The cross-vendor angle catches Gemini-flavored blind spots before a human reviewer does.
5. **Apply review fixes** (Antigravity): Bounce Claude's findings back to Antigravity's editor agent. The whole-repo context makes review-fix patches especially clean.

**Example output:**

Backend team on GCP: Ran a 4-week refactor (replacing an internal pubsub layer) with Antigravity as the primary IDE. The 2M-context kept the agent on-track across ~30 files; previous attempt with a smaller-context IDE drifted by file 12.

**Pitfalls:**

- **Preview-tier limits**: Antigravity is in preview. Quotas, model availability, and pricing may shift before GA. Don't bet a deadline on a specific feature without a fallback IDE.
- **Whole-repo context can leak secrets**: If your repo has committed .env or service-account keys, the agent will happily quote them in chat. Audit git history before pointing the agent at the repo.

**Proof artifacts:**

- [booststash — three weeks of daily Antigravity use](https://www.booststash.com/google-antigravity-review-i-tested-googles-ai-ide-for-3-weeks/) — Independent reviewer, not a Google employee. Three weeks of daily-driver use with screenshots of real IDE output and called-out bugs. Antigravity is preview-tier so most write-ups are still vendor-authored — this one isn't.

---

### Claude Code only (Pure-CLI: scope, plan, write, fix all inside Claude Code)

URL: https://aistack.sh/stack/ship-code-claude-code
Category: Code · Workflow · Level: Intermediate · Updated: 2026-05-05

If you live in the terminal, you don't need a chat tab next to your editor. Claude Code can plan a change, edit the files, run the tests, and stage the commit — all in one session. CodeRabbit reviews the PR; you bounce findings back to the same Claude Code session for fixes.

Title: One surface, end to end. Claude Code does the planning AND the implementation; CodeRabbit reviews.

**Tools in this stack:**

- Claude Code — Plan + write + fix. Give it the ticket, it produces a plan, asks clarifying questions, then implements. Skills encode your repo conventions; hooks gate dangerous commands. Subagents handle reviews and verifications without polluting the main thread.
  - Alternatives: Codex, Cursor
- CodeRabbit — Automated PR review. Same role as the recommended stack. The CLI-only loop pairs especially well with CodeRabbit because the agent can pipe `gh pr view` straight into its own context to fix findings without you copying anything.
  - Alternatives: Claude

**Cost (monthly USD):**

- Solo OSS maintainer: $0/mo
- Indie dev, daily PRs: $32/mo
- Heavy agentic use, multiple parallel sessions: $160/mo

**Workflow:**

1. **Open a session** (Claude Code): `cd repo && claude`. Hand it the ticket as the first message. Let it ask the clarifying questions a chat tab would have surfaced.
2. **Open the PR** (Claude Code): Use Claude Code's `/pr` workflow (or `gh pr create`). It writes the PR description from the diff. CodeRabbit posts its review within ~60s.
3. **Triage CodeRabbit, in-session** (Claude Code): Pipe the review into the same Claude Code session: `gh pr view --json comments | claude`. The agent sees the findings, the diff, and the codebase together — no copy-paste.
4. **Re-review and merge** (CodeRabbit): Push fixes; CodeRabbit re-reviews. Merge once it's quiet and the human review is approved.

**Example output:**

Solo founder, comparison week: Ran the same 8 tickets through this stack and the chat-led recommended stack. Time-to-merge dropped from 90 to 55 min/PR average. The trade-off: 2 of the 8 tickets needed a chat-tab debate before the agent could plan well — for those, the chat-led stack still won.

**Pitfalls:**

- **No external rubber duck**: When you genuinely don't know what to build, a CLI agent that's already implementing is the wrong tool. Drop back to a chat tab for the open-ended part, then come back to Claude Code to ship.
- **Yolo mode on a real repo**: Claude Code can run shell commands. Configure permissions and hooks before you hand it a repo with deploy credentials, migrations, or a production .env in scope.

**Proof artifacts:**

- [aistack.sh — this site](https://aistack.sh) — Built mostly inside Claude Code, terminal end-to-end. 28+ passes, 40+ PRs, the full SDLC inside one CLI session per ticket.
- [Source repo](https://github.com/theomarsoliman/aistacksh) — AGENTS.md at the repo root encodes the conventions Claude Code reads on each session. Open it to see what 'authoritative repo conventions' looks like in practice.
- [PR #29 · launch prep (sitemap + robots + 404 + error boundary)](https://github.com/theomarsoliman/aistacksh/pull/29) — Four loosely-related concerns shipped in one focused PR. Demonstrates the 'plan-and-implement in one CLI session' loop without a separate chat tab.
- [PR #39 · slim /pricing under 2 MB Googlebot limit](https://github.com/theomarsoliman/aistacksh/pull/39) — Constrained, debug-flavored work — the kind of ticket that reads like 'figure out why X is too big and fix it'. Claude Code triaging and patching in the same session.

---

### Codex (CLI + cloud) (OpenAI Codex: terminal CLI for active work, cloud lane for autonomous PRs)

URL: https://aistack.sh/stack/ship-code-codex
Category: Code · Workflow · Level: Intermediate · Updated: 2026-05-05

Codex CLI replaces the recommended stack's Claude Code. Then the differentiator: Codex Cloud takes a ticket from the queue, opens a PR autonomously, and pings you when it's ready for review. Pair it with Claude reading the diff and the human reviewer mostly approves clean diffs.

Title: OpenAI's Codex on two surfaces: CLI in the terminal for hands-on work, Cloud for fire-and-forget PRs.

**Tools in this stack:**

- Codex — CLI for active work + Cloud for backlog PRs. Codex CLI handles the same role Claude Code does: terminal agent with file system access. Codex Cloud is the ChatGPT-bundled cloud lane that takes a task and opens a PR while you do something else. Use both: hands-on for the work you're thinking about, cloud for the cleanup tickets in the queue.
  - Alternatives: Claude Code, Cursor
- Claude — Cross-vendor diff review. Cloud PRs especially want a second model reading the diff. Claude is a different lineage than Codex, so it catches things Codex's self-review wouldn't (different priors on idiomatic code, different blind spots). Paste the PR diff + the ticket, ask for a structured review.
  - Alternatives: CodeRabbit

**Cost (monthly USD):**

- Solo OSS maintainer: $0/mo
- Indie dev, daily PRs, ChatGPT Plus + Claude Pro: $40/mo
- Cloud-leaning team, ChatGPT Pro plans: $320/mo

**Workflow:**

1. **Decide: hands-on or backlog?**: If you're already thinking about the change, run the CLI. If it's a clear ticket from the backlog (refactor, dep bump, doc update), send it to Cloud and move on.
2. **Hands-on with Codex CLI** (Codex): `codex` in your repo. Same shape as Claude Code: scope, plan, implement, run tests.
3. **Backlog tickets via Codex Cloud** (Codex): From ChatGPT, hand off the ticket with the repo selected. Cloud reads the repo, plans, opens a draft PR, and pings you. Read the PR description first; reject and retry if scope drifted.
4. **Cross-vendor review with Claude** (Claude): Paste the PR diff + the original ticket into Claude. Ask for a structured review (coverage, edge cases, regressions, naming, idiomatic-code drift). For Cloud PRs especially, this catches drift the autonomous lane introduced. A different model lineage notices things the same vendor's self-review wouldn't.
5. **Bounce findings back to Codex** (Codex): For CLI work: paste Claude's findings into the same Codex session. For Cloud PRs: kick the same task back via the Cloud `Address review feedback` action with the findings as the new prompt.

**Example output:**

Two-engineer startup, mixed-mode week: Half the PRs (the ones requiring product judgment) went through Codex CLI. The other half (dep updates, type-error cleanups, doc touch-ups) ran on Codex Cloud while the engineers worked on real product. The Cloud lane shipped 22 PRs in one week with 3 needing a retry for scope drift.

**Pitfalls:**

- **Codex Cloud scope drift**: Autonomous lanes occasionally rewrite the world. Always read the PR description before the diff; if the description says 'also refactored…', kill the PR and retry with a tighter ticket.
- **Cloud-task budget burn**: Cloud tasks are metered separately from the chat subscription. A team that sends 30 cloud tasks/day will exceed the bundled budget within a week. Track it.
- **Conflating CLI tells with Cloud tells**: Codex CLI is supervised; Cloud isn't. The same prompt works very differently in the two lanes. Don't copy a CLI prompt into Cloud and expect the same result.

**Proof artifacts:**

- [OpenAI Codex CLI — live demo (6:20)](https://www.youtube.com/watch?v=FUq9qRwrDrI) — Recorded by OpenAI on the OpenAI YouTube channel. Vendor-authored, but a real terminal session against a real codebase — useful for seeing the actual prompt-to-diff loop before committing to the workflow.

---

### Cursor (IDE-first) (AI inside the editor, not in a chat tab or terminal)

URL: https://aistack.sh/stack/ship-code-cursor
Category: Code · Workflow · Level: Beginner · Updated: 2026-05-05

If you'd rather see the diff as you go than read a terminal log, Cursor is the answer. Cmd-K for inline edits, agent mode for multi-file changes, Bugbot reviews the PR. The editor stays the source of truth all the way through.

Title: Cursor as the primary surface. Inline edits, agent mode for multi-file work, Bugbot for review.

**Tools in this stack:**

- Cursor — Plan + write + review. Cmd-K for surgical edits, Cmd-L for chat with file context, Composer (agent mode) for multi-file work, Bugbot for PR review on push. One tool for the whole loop.
  - Alternatives: Claude Code, Codex

**Cost (monthly USD):**

- Solo OSS, Hobby tier: $0/mo
- Indie dev, daily PRs: $20/mo
- 5-dev team, Business plan: $200/mo

**Workflow:**

1. **Plan in Cursor's chat panel** (Cursor): Cmd-L. Add the ticket and the most-relevant 2 to 3 files via @-mentions. Get the plan there; you stay in the editor.
2. **Implement with Composer** (Cursor): Switch to agent mode (Composer). Hand it the plan + the files. Cursor proposes a multi-file diff inline; accept or reject hunks visually.
3. **Inline iterate with Cmd-K** (Cursor): For tight edits inside one file, skip the agent — Cmd-K is faster. The editor stays in flow.
4. **Open the PR** (Cursor): Push the branch. Bugbot posts its review on the PR automatically — same surface as a human reviewer, so teammates see the findings too.
5. **Patch findings inline** (Cursor): Bring the Bugbot comments back into Cursor's chat panel; Composer turns them into a patch you can review hunk-by-hunk.

**Example output:**

Frontend-heavy team, comparison week: On UI work, the IDE-first loop won: changes are visual, the editor is where you naturally iterate. On backend refactors that span 10+ files, the team preferred Claude Code because Composer's hunk-by-hunk acceptance got tedious. Mixed-mode users tended to keep both subscriptions active.

**Pitfalls:**

- **Composer's tab cost on big diffs**: On a 12-file diff, accepting hunks one at a time in the editor takes longer than reading a CLI agent's summary. Pick the right surface for the size of the change.
- **Bugbot is Cursor-only**: If your team also reviews from the GitHub web UI without Cursor open, Bugbot's comments are still visible (they post to the PR), but the inline IDE features are not. Make sure the PR-level review is the source of truth, not the in-editor markers.

**Proof artifacts:**

- [madewithlove — testing Bugbot on real PRs](https://madewithlove.com/blog/automatic-pull-request-reviewing-with-cursors-bugbot/) — Independent dev consultancy (not Anysphere). Includes screenshots of actual Bugbot comments on their own PRs and an honest tradeoffs section — they later switched away. Good for setting expectations before adopting.

---

### Windsurf (Cascade) (Codeium's IDE with the Cascade autonomous-edit agent)

URL: https://aistack.sh/stack/ship-code-windsurf
Category: Code · Workflow · Level: Intermediate · Updated: 2026-05-05

IDE-first, agent inside the editor, Cascade reviews the diff before push. Different agent than Cursor: Cascade is more autonomous on multi-step tasks. The trade-off vs Cursor: less hunk-by-hunk granularity, more 'kick it off and review what it did'.

Title: Windsurf's Cascade agent plans + executes multi-step changes in the IDE. Fewer per-turn cycles on big diffs than Composer.

**Tools in this stack:**

- Windsurf — Plan + write + review. Cascade plans across files and runs the plan with less hand-holding than Composer. After the agent finishes, ask it to review its own diff before you push — it's the same model that wrote the code, but the second pass catches its own omissions.
  - Alternatives: Cursor, Claude Code

**Cost (monthly USD):**

- Solo OSS: $0/mo
- Solo, daily PRs: $15/mo
- 5-dev team: $175/mo

**Workflow:**

1. **Open in Windsurf** (Windsurf): VS Code fork — keymap and extensions carry over.
2. **Cascade for multi-step work** (Windsurf): Hand it the plan from Claude (or describe the change directly). Cascade plans + executes; you review the result, not each step.
3. **Self-review before push** (Windsurf): In Cascade, ask 'review the diff for issues before I push'. The agent rereads its own work and flags missed edge cases. Catches ~70% of what an external reviewer would. Push, then open the PR.

**Example output:**

Backend dev, weekly comparison: Ran the same refactor tickets through Windsurf and Cursor. Windsurf shipped ~20% faster on the multi-file work because the hunk-by-hunk overhead in Composer slowed each step. Lost a percentage point on accuracy on tighter edits — Cursor's per-hunk control caught a regression Cascade missed.

**Pitfalls:**

- **Cascade's autonomy = harder rollbacks**: If Cascade does 12 things and 1 is wrong, undoing is harder than Cursor's per-hunk acceptance. Use git branches per Cascade run as a safety net.
- **Smaller community than Cursor**: Fewer Stack Overflow answers when something breaks. Worth knowing before committing the whole team.

**Proof artifacts:**

- [Mews engineering — full-stack app with Cascade](https://developers.mews.com/how-windsurf-builds-full-stack-apps/) — Mews is a real hospitality SaaS (not a Codeium customer story). Their engineering blog walks through shipping a feature with Windsurf + Cascade end-to-end, with code snippets and friction notes.

---

### Ship code with AI (recommended)

URL: https://aistack.sh/stack/ship-code-with-ai
Category: Code · Workflow · Level: Beginner · Updated: 2026-05-05

The full SDLC with three Anthropic-and-friends tools. Claude (chat) scopes the change so the diff stays small. Claude Code, the terminal agent, writes and edits across files without copy-pasting context. CodeRabbit reviews every PR, Claude turns its findings into patches, and the human reviewer only sees clean diffs.

Title: Plan with Claude, hand off to Claude Code in the terminal for autonomous edits, let CodeRabbit review, fix what it finds, ship.

**Tools in this stack:**

- Claude — Scope + plan + fix. Use the chat surface for the parts that need human judgment: scoping, debating tradeoffs, turning a CodeRabbit comment into a sharp patch. Long context holds the whole plan + relevant files.
  - Alternatives: ChatGPT, Gemini
- Claude Code — Autonomous multi-file edits. Once the plan is locked, hand the implementation to Claude Code. It runs in your repo, reads the file system, executes commands, and edits files in place — no per-turn paste loop. Skills + hooks let you encode local conventions.
  - Alternatives: Codex, Cursor
- CodeRabbit — Automated PR review. Runs on every push without configuration. Reads the diff in context of the whole repo, leaves inline comments tagged by severity, and ships a 1-paragraph PR summary. Free tier covers OSS; $12/dev/mo for private repos.
  - Alternatives: Claude

**Cost (monthly USD):**

- Solo OSS maintainer: $0/mo
- Indie team, 1 to 3 devs, private repo: $32/mo
- 5-dev startup, daily PRs, Max plans: $160/mo

**Workflow:**

1. **Scope the change with Claude** (Claude): Drop the ticket or 1-paragraph problem statement into Claude. Get back a plan: files to touch, edge cases, the smallest viable diff.
2. **Hand the plan to Claude Code** (Claude Code): Drop into your repo, run `claude`, and feed it the plan from step 1. Claude Code reads the relevant files itself, makes edits in place, runs your tests, and iterates on failures. No copy-paste loop.
3. **Open the PR** (CodeRabbit): Push your branch, open the PR. CodeRabbit posts a 1-paragraph summary plus inline comments within ~60 seconds. Findings come tagged: bug, security, performance, style.
4. **Triage the findings** (CodeRabbit): Skim CodeRabbit's summary first; it surfaces the high-signal issues. Resolve trivial style nits in bulk. The bug, security, and perf comments earn a real fix.
5. **Apply fixes with Claude** (Claude): For each non-trivial finding, paste the CodeRabbit comment plus the file region into Claude. Get the patch back, test it locally, push.
6. **Re-review and merge** (CodeRabbit): Push the fixes; CodeRabbit re-reviews automatically. If it goes quiet, the human reviewer takes over with a clean diff to look at.

**Example output:**

Solo founder, 18-PR week: From ticket to merged PR averaged 90 minutes (down from a half-day) because Claude scoped the work, Claude Code did the multi-file edits without a paste loop, and CodeRabbit caught the issues before the human reviewer. CodeRabbit alone caught 4 real bugs that week (1 SQL injection, 2 unhandled promise rejections, 1 N+1 query).

**Pitfalls:**

- **Skipping the plan step**: Jumping straight into Claude Code without the chat-tab plan makes the agent over-scope. Always do the plan first; the diff comes out 30 to 50 percent smaller and the PR is reviewable.
- **Letting Claude Code run unsupervised on first use**: It will happily edit 14 files on a refactor that should have touched 3. Watch the first few diffs go by; tighten the prompt or use hooks to scope its access until you trust the loop in this repo.
- **Treating every CodeRabbit nit as a blocker**: CodeRabbit will leave 30 comments on a 400-line PR. Resolve the trivial style ones in bulk; only spend cycles on bug, security, and perf tags.
- **Skipping the human review**: AI review catches mechanical issues, not product or architecture mistakes. It complements the human reviewer, never replaces them.
- **Auto-applying suggested fixes blindly**: CodeRabbit's suggested patches are usually right but occasionally drop a guard or miss a side effect. Read the diff before clicking commit.

**Proof artifacts:**

- [aistack.sh — this site](https://aistack.sh) — 51 tools, 42 stacks, real backend (Better Auth + Supabase + PostHog). 28+ shipping passes built end-to-end with this stack: Claude scoped each pass, Claude Code did the multi-file edits, CodeRabbit summarized every PR.
- [Source repo](https://github.com/theomarsoliman/aistacksh) — All 40 PRs public. Read CHANGELOG.md for the per-pass record of what each AI tool actually contributed.
- [PR #34 · programmatic-SEO scaffold](https://github.com/theomarsoliman/aistacksh/pull/34) — Multi-file change (combinatorial routes, JSON-LD, RSS). CodeRabbit summary in the PR body, line comments on the diff. Representative of how a single chat-scoped + CC-implemented + CR-reviewed PR moves through this loop.
- [PR #23 · launch blockers (admin role check + tooling)](https://github.com/theomarsoliman/aistacksh/pull/23) — Substantial, security-touching diff. Good example of a CodeRabbit comment turning into a Claude-authored patch (the chat-tab fix loop).

---

## Goal: Short-form distribution (TikTok, Reels, Shorts)

### AI-native posting (Ghost Feed handles content + scheduling in one. Set-and-forget.)

URL: https://aistack.sh/stack/short-form-ai-native
Category: Marketing · Distribution · Level: Beginner · Updated: 2026-05-05

When the goal is 'never miss a posting day' more than 'maximize each post', Ghost Feed collapses the whole pipeline into one tool. You give it a brand, it gives you a queue. The trade-off vs the recommended stack: less per-post control, more reliability of cadence.

Title: Ghost Feed writes the captions, generates the visuals, and posts on a fixed cadence. The 'I just want a feed running' option.

**Tools in this stack:**

- Ghost Feed — AI content + scheduler. Pulls trends, drafts in your voice, generates visuals, queues posts. The 'one-tool' answer when you don't have time for the recommended pipeline.
  - Alternatives: Post Bridge, GenViral
- CapCut — Optional edit polish. Use only when Ghost Feed's auto-generated video needs a manual touch (rare on the Pro tier, more common on Starter).

**Cost (monthly USD):**

- 1 brand, 1 to 3 posts/day: $29/mo
- Higher cadence + more polish: $79/mo
- Multi-brand: $237/mo

**Workflow:**

1. **Brand setup** (Ghost Feed): One-time: train Ghost Feed on your brand voice (10 to 20 example posts), niche, and visual style.
2. **Approve the daily queue** (Ghost Feed): Ghost Feed proposes 3 to 5 posts each morning; you approve / reject / edit. The 5-min review replaces the hour of writing + posting.
3. **Polish in CapCut (optional)** (CapCut): Some auto-generated videos need a beat re-cut or a typography tweak. Open in CapCut, fix in 2 to 3 min, re-upload.
4. **Read the weekly recap** (Ghost Feed): Ghost Feed's weekly recap surfaces the 2 to 3 posts that drove most reach. Copy that pattern into the next week's queue.

**Example output:**

Side-project founder, no time: Couldn't keep up with the cross-post-by-hand cadence. Switched to Ghost Feed at $29/mo. Posts went from 3/week to daily without adding hours. Reach dropped slightly per post (no per-platform tailoring) but total reach grew because the queue actually fired.

**Pitfalls:**

- **Voice drift to generic AI**: Ghost Feed's voice match is good, not perfect. After 30 days, re-train on your most recent 10 posts. Skipping this is how AI accounts start sounding like every other AI account.
- **Auto-generated visuals = trend-late**: Templated visuals don't catch this week's specific format. If you care about being on-trend, pair Ghost Feed with manual CapCut edits for the 1-2 weekly posts that should ride a current trend.

---

### Cross-platform short-form (recommended)

URL: https://aistack.sh/stack/short-form-cross-platform
Category: Marketing · Distribution · Level: Beginner · Updated: 2026-05-05

Most creators waste an hour a day cross-posting by hand. Post Bridge replaces that hour with one upload form. CapCut handles the edit, Claude tailors captions per platform (algorithms reward different tones), Post Bridge schedules the lot.

Title: Edit one vertical video, write captions with Claude, ship it to TikTok, Reels, Shorts, X, LinkedIn, and Threads via Post Bridge.

**Tools in this stack:**

- CapCut — Edit. 9:16 vertical, captions, beat-aligned cuts, music duck. Free tier covers most short-form needs; the Pro features matter mostly for batch export.
  - Alternatives: Descript
- Claude — Per-platform caption tailoring. TikTok captions ride a different algorithm than YouTube Shorts descriptions. Claude rewrites a single source caption into 6 platform-shaped variants in one prompt.
  - Alternatives: ChatGPT
- Post Bridge — Schedule + ship. Native API integrations with the major short-form platforms. Reliable; the most-loved part by users is the unified analytics that show which platform actually drives views per video.
  - Alternatives: Postiz, Ghost Feed

**Cost (monthly USD):**

- 1 video/day, 1 brand: $29/mo
- Daily across 3 brands: $49/mo
- Agency, 10 brands: $133/mo

**Workflow:**

1. **Edit in CapCut** (CapCut): 9:16, captions, music duck. Export at 1080×1920, H.264, 10–12 Mbps.
2. **Caption pass with Claude** (Claude): One Claude turn per video → 6 platform-shaped caption variants.
3. **Schedule in Post Bridge** (Post Bridge): One upload, attach the 6 captions, set posting times per platform (IG Reels: 7pm local; YT Shorts: morning; TikTok: anytime; X: 9am).
4. **Read the analytics** (Post Bridge): Post Bridge unified view tells you which platform drove the views. Double down where the watch time is. Don't chase platforms with low retention.

**Example output:**

Solo creator, niche education account: Switched from posting-by-hand to this stack. Daily output stayed constant; time spent on distribution dropped from 50 to 12 min/day. Discovered (via Post Bridge analytics) that 65% of views came from YT Shorts, not TikTok — reallocated thumbnail attention accordingly.

**Pitfalls:**

- **Identical captions across platforms**: Each platform's algorithm rewards different things. The whole point of the Claude step is per-platform variants. Skipping this kills reach by ~30% based on creator surveys.
- **Posting at the same time everywhere**: TikTok prime time is not LinkedIn prime time. Use Post Bridge's per-platform scheduling, not the 'post now to all' button.

---

### Self-hosted scheduler (Self-host on a $5 VPS. Open source. Own your data.)

URL: https://aistack.sh/stack/short-form-self-hosted
Category: Marketing · Distribution · Level: Intermediate · Updated: 2026-05-05

Same edit + caption pipeline as the recommended stack. Different distribution surface: Postiz, MIT-licensed, self-hosted in Docker. Pays off if you have ops chops, hate per-seat pricing, or run an agency where 10+ team members would balloon a SaaS bill.

Title: Postiz on a $5 VPS replaces the SaaS scheduler. Same edit + caption flow; you keep the data and the bill stays at the cost of a Hetzner box.

**Tools in this stack:**

- CapCut — Edit. Same role as the recommended stack.
  - Alternatives: Descript
- Claude — Per-platform caption tailoring. Same prompt as the recommended stack works as-is.
  - Alternatives: ChatGPT
- Postiz — Self-hosted scheduler. MIT-licensed Docker container. Run it on a $5/mo Hetzner box, point your team at it, never pay per-seat. Covers the same major platforms as Post Bridge.
  - Alternatives: Post Bridge

**Cost (monthly USD):**

- Self-host, solo: $25/mo
- Self-host, small team (no per-seat): $25/mo
- Hosted Postiz (skip ops): $49/mo

**Workflow:**

1. **Edit + caption (same as recommended)** (Claude): CapCut edit + Claude per-platform captions. Use the same prompts.
2. **Spin up Postiz** (Postiz): One-time: rent a $5 Hetzner box, install Docker, `docker compose up` from the Postiz repo. ~30 minutes for someone who's done this before, longer if it's new.
3. **Connect your accounts** (Postiz): OAuth each platform from the Postiz UI. Set up posting schedule. Invite teammates without paying per seat.
4. **Schedule + ship** (Postiz): Same outputs, different button. The ops cost is real but predictable; the per-seat ceiling is gone.

**Example output:**

5-person micro-agency: Moved from a $99/seat SaaS to self-hosted Postiz on a $5 VPS. Saved $400/mo across the team. Spent ~2 hours/month on routine maintenance (dependency updates, reconnecting OAuth tokens). Net win once the team crossed 3 members.

**Pitfalls:**

- **Underestimating ops time**: OAuth tokens expire. Platforms change APIs. You'll pay 1-2 hours/month even when 'nothing is wrong'. Worth it at team scale; not worth it solo unless you enjoy the work.
- **Mixing up the brand-color palette in scheduling**: Postiz is open source — UI changes between versions. Lock the version in your docker-compose to avoid surprise breakage on a release morning.

---

### Viral content engine (GenViral mines trends, Post Bridge distributes. Optimized for breakout reach.)

URL: https://aistack.sh/stack/viral-content-engine
Category: Marketing · Distribution · Level: Intermediate · Updated: 2026-05-05

Different mode than the other three variants. Where the recommended stack is daily posting at consistent quality, this one is for creators chasing breakout videos: ride a trending audio, post fast, multi-platform. Low-effort baseline content stays elsewhere; this engine fires when you want to swing.

Title: GenViral mines trending audios + formats, generates on-trend videos, and Post Bridge ships them to every platform — built for the swing-for-the-fences play.

**Tools in this stack:**

- GenViral — Trend mining + on-trend video. Surfaces trending audios + formats from TikTok / Reels in your niche; generates videos that fit. The 'don't miss a wave' tool.
  - Alternatives: Ghost Feed
- CapCut — Polish + caption. GenViral output is 80% there; CapCut closes the last 20% (custom text, font choices, brand watermark).
- Post Bridge — Multi-platform ship. Speed matters on trends — hours, not days. Post Bridge's one-upload flow gets the video out across platforms before the trend cools.
  - Alternatives: Postiz

**Cost (monthly USD):**

- Solo, swinging weekly: $48/mo
- Solo, daily swings: $68/mo
- Agency, multi-brand: $198/mo

**Workflow:**

1. **Morning trend pass** (GenViral): GenViral: pull this morning's trending audios + formats for your niche. Pick 1 to 3 worth riding.
2. **Generate the video** (GenViral): GenViral generates an on-trend video from your brand assets + the trending format. Output is usable as-is for fast turnarounds.
3. **Polish in CapCut** (CapCut): Custom text, brand watermark, last typography pass. 4 to 6 minutes per video. Skip if the trend is decaying fast.
4. **Ship via Post Bridge** (Post Bridge): Same upload flow as the recommended stack. Post within 24 hours of the trend; the curve falls fast.

**Example output:**

Niche creator, breakout video: Hit a 4M-view TikTok by riding a trending audio that GenViral surfaced 6 hours after it took off. Post Bridge shipped to all platforms in 12 min. The single video drove 18k followers in 48 hours — more than the previous 3 months combined.

**Pitfalls:**

- **Riding every trend kills the brand**: Most trends don't fit your niche. Pick 1 in 5. Riding everything makes the account feel rented.
- **Over-polishing kills speed**: Trends decay in days, sometimes hours. The 'good enough at hour 6' beats the 'perfect at hour 30'. Lean toward shipping.
- **Treating this as your baseline**: This engine is the swing-bat, not the daily-driver. Pair it with the recommended stack (or AI-native one) for the daily cadence.

---

## Goal: Solo podcast

### Solo podcast (recommended)

URL: https://aistack.sh/stack/solo-podcast
Category: Audio · Podcast · Level: Beginner · Updated: 2026-05-05

Riverside records studio quality, Descript edits with the transcript, Castmagic produces show notes + clips, Opus auto-clips for shorts.

Title: Record on Sunday, ship on Monday. Solo podcast pipeline with no editor required.

**Tools in this stack:**

- Riverside — Recording. Local-track recording = no internet wobble. Auto noise removal.
  - Alternatives: Descript
- Descript — Editing. Edit the transcript, the audio follows. Strip 'um's in two clicks.
  - Alternatives: Riverside
- Castmagic — Show notes + posts. Drop the audio, get titles, show notes, timestamps, social posts back in 5 minutes.
- Opus Clip — Shorts pipeline. Long episode → 6–10 short clips with viral scores and captions.
  - Alternatives: Castmagic

**Cost (monthly USD):**

- 1 episode / week: $55/mo
- 2 / week + clips: $87/mo
- Daily, multi-show: $210/mo

**Workflow:**

1. **Record** (Riverside): Riverside, 60-min target. Track-locally, magic edit later.
2. **Edit by transcript** (Descript): Descript: strip filler words, tighten beats. ~30 min for an hour.
3. **Notes + posts** (Castmagic): Castmagic generates titles, show notes, X thread, LinkedIn post.
4. **Auto-clip** (Opus Clip): Opus picks the 6 best moments, captions them, exports vertical.
5. **Publish** (Castmagic): Upload audio to your host, video to YT, clips to TikTok/Reels.

**Example output:**

Solo founder, 4 months in: 1,200 listeners/episode, 24 sponsor inbounds from clips alone. Total weekly time: 3.5 hours from record to publish.

**Pitfalls:**

- **Episode length drifts**: Without a guest, 60 min stretches to 90. Force a hard outline; cut to 35–45.
- **Clip captions wrong**: Always re-check Opus captions; one mis-spelled name kills shareability.

---

### Solo podcast (lean) (All-in-one with Descript. Skip Riverside and Castmagic.)

URL: https://aistack.sh/stack/solo-podcast-lean
Category: Audio · Podcast · Level: Beginner · Updated: 2026-05-05

Descript records, edits by transcript, and exports both audio and video. Castmagic's show-notes job becomes a Claude prompt against the Descript transcript. Opus Clip handles shorts.

Title: Solo podcast in one app. Record, edit, transcribe, and clip in Descript; auto-clip in Opus.

**Tools in this stack:**

- Descript — Record, edit, transcribe. Replaces Riverside (recording) and the editing tool in one. Studio Sound covers what Riverside's local-track usually fixes.
  - Alternatives: Riverside
- Opus Clip — Auto-clips for shorts. Free tier covers 1 hour/month of long-form to clips. Drop your Descript export, get 6 to 10 vertical clips with captions.
  - Alternatives: Castmagic

**Cost (monthly USD):**

- 1 episode/wk, free clips: $16/mo
- 2/wk + clips: $35/mo
- Daily, multi-show: $110/mo

**Workflow:**

1. **Record in Descript** (Descript): Built-in recording with Studio Sound. No second tool. 60-min target.
2. **Edit by transcript** (Descript): Strip filler words automatically; tighten beats by editing words. ~25 minutes for an hour of audio.
3. **Show notes from transcript**: Drop the Descript transcript into Claude. The Castmagic prompt works as-is.
4. **Auto-clip** (Opus Clip): Export the Descript video. Opus picks the 6 best moments, captions them, exports vertical.
5. **Publish** (Descript): Audio to your podcast host (Transistor, Buzzsprout). Video to YouTube. Clips to TikTok/Reels.

**Example output:**

Solo founder, month 2: Same publish cadence as the recommended stack at $35/mo vs $87/mo. The trade-off: ~10 minutes more editing per episode, no studio-grade local-track recording.

**Pitfalls:**

- **Descript Studio Sound != local-track**: If you record in a noisy room or with a guest on shaky internet, Riverside's local recording wins. Solo + decent room, Descript is enough.
- **Opus free tier caps fast**: 1 hour/month of long-form is one episode. If you ship weekly, you'll need Starter ($19) by week 4.

---

## Goal: TikTok character account

### TikTok character (recommended)

URL: https://aistack.sh/stack/tiktok-character
Category: Video · Creator · Level: Intermediate · Updated: 2026-05-05

ElevenLabs gives you a consistent voice; Pika + Runway hold a consistent look across episodes; CapCut ships them.

Title: Build a recurring TikTok character. Same voice, same look, daily uploads, no on-camera time.

**Tools in this stack:**

- Claude — Character bible + script. Long context lets you keep the persona, voice, and arc consistent over weeks.
  - Alternatives: ChatGPT
- ElevenLabs — Character voice. One cloned/marketplace voice = recognizable in 0.3s of audio. Worth it.
- Pika — Visual style. Pick a Pika style preset and lock it. Runway is too photoreal for cartoon characters.
  - Alternatives: Runway
- CapCut — Edit. Vertical-first, captions, fast.

**Cost (monthly USD):**

- 1 ep / day: $50/mo
- 1 ep + 3 shorts / day: $92/mo
- Multi-character agency: $310/mo

**Workflow:**

1. **Lock the bible** (Claude): Claude project: persona, voice, visual style, recurring beats.
2. **Daily script** (Claude): Pull the bible + today's news; Claude writes 60s of dialogue.
3. **Voice + visuals** (Pika): ElevenLabs renders the line; Pika renders the scene; mash in CapCut.
4. **Ship** (CapCut): Upload at 6pm local. CapCut auto-suggests caption variants.

**Example output:**

Character account, 90k followers in 4 months: Brand: a sarcastic raccoon who reviews tech. Daily 60s episodes. Two sponsorships in month 4 at $1,200 each.

**Pitfalls:**

- **Visual drift**: Pika style preset only carries you so far; lock a seed + a reference image, re-render if it slips.
- **Voice sounding generic on day 30**: Re-clone every 60 days as ElevenLabs models update; pin a sample line for QA.

---

### TikTok character (audio + text) (No AI video. Voice + typography + a single recurring image.)

URL: https://aistack.sh/stack/tiktok-character-audio
Category: Video · Creator · Level: Beginner · Updated: 2026-05-05

AI video is the slowest, most expensive, most failure-prone part of the recommended stack. Cut it. Use one consistent character image (commission once or generate once) plus ElevenLabs voice and beat-cut typography in CapCut. Ship 3x faster, half the cost.

Title: Recurring character without generating any video. Voice carries; CapCut's typography carries the rest.

**Tools in this stack:**

- Claude — Character bible + script. Same as the recommended stack. The character lives in the voice and the writing, not the visuals.
  - Alternatives: ChatGPT
- ElevenLabs — Character voice. Even more important here than in the Pika version: the voice IS the character. One marketplace voice or clone, locked.
- CapCut — Typography + edit. Animated text on the character image carries the visual. CapCut's keyframe text is the bottleneck-free way to do this.
  - Alternatives: Descript

**Cost (monthly USD):**

- 1 ep/day: $22/mo
- Daily + 3 shorts: $42/mo
- Multi-character agency: $210/mo

**Workflow:**

1. **Lock the bible (no visual section)** (Claude): Same prompt as the recommended stack, minus 'visual style'. Replace it with a single character image and a typography lock-up.
2. **Daily script** (Claude): Same script structure as the Pika version, but every line gets a typography cue instead of a B-ROLL cue.
3. **Voice** (ElevenLabs): ElevenLabs renders the script as one audio file. Same voice, same settings, every episode.
4. **Typography pass** (CapCut): CapCut: drop voice on track 1, character image as background. Punch text per [TEXT] cue, beat-aligned, with 2 fonts max from the lock-up.
5. **Ship** (CapCut): Upload at 6pm local. Whole loop ships in 25 minutes once the bible exists.

**Example output:**

Audio-first character account, 4 months: Same niche as the Pika version (sarcastic tech raccoon). 110k followers vs 90k for the AI-video version, mostly because they shipped 1.7x more episodes (no Pika render bottleneck). Visual drift was a non-problem.

**Pitfalls:**

- **Visuals get static**: One character image only works if the typography pulls weight. Spend the time on the typography lock-up; it's the difference between 'cheap' and 'identifiable'.
- **Voice fatigue**: Audio-first means the voice is everything. Re-clone every 60 days; pin a sample line for QA. Same as the Pika version, but more load-bearing here.

---

## Goal: UGC ad creative

### UGC ad creative (recommended)

URL: https://aistack.sh/stack/ugc-ads
Category: Marketing · Ads · Level: Intermediate · Updated: 2026-05-05

Claude writes 30 ad-script variations, ElevenLabs voices them, Pika generates the b-roll, CapCut edits and adds captions.

Title: AI-generated UGC ad creative (scripts, voices, b-roll) at 1/10 the cost of a real creator.

**Tools in this stack:**

- Claude — Script variations. Best at ad-copy hooks; can generate 30 variants from 1 brief without all sounding the same.
  - Alternatives: ChatGPT
- ElevenLabs — Voiceover. Pick from the marketplace voices for variety; pay only for characters used.
- Pika — B-roll. Pika's stylized motion is more 'ad-shaped' than Runway's photoreal.
  - Alternatives: Runway
- CapCut — Edit + caption. Built for vertical ad formats. AI captions are good enough to ship.

**Cost (monthly USD):**

- 10 ads/wk: $24/mo
- 30 ads/wk: $70/mo
- Agency · 100 ads/wk: $240/mo

**Workflow:**

1. **Brief to 30 hooks** (Claude): Claude turns one product brief into 30 hook variants across 3 angles.
2. **Voice** (ElevenLabs): ElevenLabs marketplace voice per persona. Render at 192kbps.
3. **B-roll** (Pika): Pika generates 4–6s clips per beat. Stylized > photoreal for ads.
4. **Cut** (CapCut): CapCut vertical 9:16, captions, music duck.

**Example output:**

DTC supplements brand: 30 AI-UGC ads tested in week 1. 4 winners; CPA dropped 38% vs the agency-shot UGC pack at $5k.

**Pitfalls:**

- **Voices sound the same across ads**: Force diversity by mapping persona → voice up front; never re-use the same voice across ads in the same creative test.
- **Compliance flags on health/finance**: Meta auto-flags AI-voiced health ads. Disclose 'created with AI' if you're in a regulated vertical.

---

### UGC ads (CapCut-only) (Single-tool: Claude writes hooks, CapCut handles voice, b-roll, edit)

URL: https://aistack.sh/stack/ugc-ads-budget
Category: Marketing · Ads · Level: Beginner · Updated: 2026-05-05

When you're testing whether ads work at all, the full pipeline is overkill. Claude writes 30 hooks; CapCut's AI voice library, stock footage, and auto-captions do everything else. Validate the angle, then graduate to ElevenLabs + Pika for the winners.

Title: AI UGC ads at <$25/mo. CapCut's built-in AI voice + stock library replaces ElevenLabs and Pika.

**Tools in this stack:**

- Claude — Script variations. Same hook generation as the recommended stack. The output is text — same prompt, same quality at the free or $20 tier.
  - Alternatives: ChatGPT
- CapCut — Voice + b-roll + cut + caption. CapCut now ships AI text-to-speech with passable voices, a built-in stock library, auto-captions, and beat-aligned cuts. One tool replaces ElevenLabs + Pika for testing.
  - Alternatives: Descript

**Cost (monthly USD):**

- 10 ads/wk, free CapCut: $0/mo
- 30 ads/wk, Pro features: $24/mo
- Brand testing 100/wk: $45/mo

**Workflow:**

1. **30 hooks, 3 angles** (Claude): Same prompt as the recommended stack. The bottleneck is the hooks, not the tools.
2. **Voice in CapCut** (CapCut): Open CapCut, paste the hook, pick a voice from the AI voice library. Render at default settings.
3. **B-roll from CapCut stock** (CapCut): Search the built-in stock library by keyword from the hook. 4 to 6 clips per ad. Stock variety is shallow per niche; rotate aggressively.
4. **Cut + caption + export** (CapCut): 9:16, auto-captions, beat-align cuts, music duck. Export 1080p. Whole loop is 6 to 8 minutes per ad once you have the hooks.

**Example output:**

DTC founder, week 1 testing: 30 ads shipped at $0 spend on tooling. 3 angles tested across hooks. The 2 winners moved to the full pipeline (ElevenLabs + Pika) for production-quality scaling. The other 28 cost nothing to throw away.

**Pitfalls:**

- **CapCut voices have the AI tell**: Built-in voices are good enough for testing, not for production. Once an angle wins, re-render with ElevenLabs before scaling spend.
- **Stock library repetition**: After 10 to 15 ads in one niche, the same clips repeat. That's the signal to graduate to Pika or Runway for the winners.
- **Treating the test as the final cut**: Some founders ship the budget version to broad audiences and wonder why CTR is half what testing predicted. Test in cold audiences, scale only with the upgraded version.

---

---

# Tool-vs-tool comparisons

44 head-to-head pages. Each pair has a per-page markdown alternate at `<url>/llms.txt` with structured pricing, FAQ, and stack-usage data.

## Android Studio vs Xcode

URL: https://aistack.sh/compare/android-studio-vs-xcode
Markdown: https://aistack.sh/compare/android-studio-vs-xcode/llms.txt
Android Studio (Free · Play developer account $25 one-time, 1 stack) vs Xcode (Free · Apple Developer Program $99/year for App Store, 1 stack). Both in the AI IDEs category.

## Antigravity vs Claude Code

URL: https://aistack.sh/compare/antigravity-vs-claude-code
Markdown: https://aistack.sh/compare/antigravity-vs-claude-code/llms.txt
Antigravity (Free · $20/mo AI Pro · $249.99/mo AI Ultra, 1 stack) vs Claude Code (Bundled with Claude Pro $20/mo · Max $100-200/mo · API metered, 2 stacks). Both in the AI IDEs category.

## Antigravity vs Codex

URL: https://aistack.sh/compare/antigravity-vs-codex
Markdown: https://aistack.sh/compare/antigravity-vs-codex/llms.txt
Antigravity (Free · $20/mo AI Pro · $249.99/mo AI Ultra, 1 stack) vs Codex (Bundled with ChatGPT Plus $20/mo · ChatGPT Pro $200/mo · API metered, 1 stack). Both in the AI IDEs category.

## Antigravity vs Cursor

URL: https://aistack.sh/compare/antigravity-vs-cursor
Markdown: https://aistack.sh/compare/antigravity-vs-cursor/llms.txt
Antigravity (Free · $20/mo AI Pro · $249.99/mo AI Ultra, 1 stack) vs Cursor (Free Hobby · $20/mo Pro · $60/mo Pro+ · $200/mo Ultra · $40/seat Business, 5 stacks). Both in the AI IDEs category.

## Apollo vs Clay

URL: https://aistack.sh/compare/apollo-vs-clay
Markdown: https://aistack.sh/compare/apollo-vs-clay/llms.txt
Apollo (Free · $49/seat/mo Basic · $79/seat/mo Professional (annual billing), 2 stacks) vs Clay ($185/mo Launch · $495/mo Growth, 1 stack). Both in the Outbound and inbox category.

## Beehiiv vs Substack

URL: https://aistack.sh/compare/beehiiv-vs-substack
Markdown: https://aistack.sh/compare/beehiiv-vs-substack/llms.txt
Beehiiv (Free up to 2,500 · $43/mo Scale, 1 stack) vs Substack (Free to publish · 10% of paid subs, 1 stack). Both in the Social and publishing category.

## Canva vs Figma

URL: https://aistack.sh/compare/canva-vs-figma
Markdown: https://aistack.sh/compare/canva-vs-figma/llms.txt
Canva (Free · $15/mo Pro, 2 stacks) vs Figma (Free · Pro from $12-$20 per seat/mo (seat-based), 2 stacks). Both in the Image and design category.

## CapCut vs Descript

URL: https://aistack.sh/compare/capcut-vs-descript
Markdown: https://aistack.sh/compare/capcut-vs-descript/llms.txt
CapCut (Free · $9.99/mo Standard · $19.99/mo Pro, 11 stacks) vs Descript ($24/mo Hobbyist · $35/mo Creator · $65/mo Business, 3 stacks). Both in the Editing category.

## Castmagic vs Opus Clip

URL: https://aistack.sh/compare/castmagic-vs-opus-clip
Markdown: https://aistack.sh/compare/castmagic-vs-opus-clip/llms.txt
Castmagic ($29/mo Hobby · $99/mo Starter, 1 stack) vs Opus Clip (Free · $15/mo Starter · $29/mo Pro, 2 stacks). Both in the Social and publishing category.

## ChatGPT vs Claude

URL: https://aistack.sh/compare/chatgpt-vs-claude
Markdown: https://aistack.sh/compare/chatgpt-vs-claude/llms.txt
ChatGPT ($20/mo Plus · API $2.50/M tokens, 1 stack) vs Claude ($20/mo Pro · Sonnet API $3/$15 per M tokens (input/output), 33 stacks). Both in the LLMs category.

## ChatGPT vs DeepSeek

URL: https://aistack.sh/compare/chatgpt-vs-deepseek
Markdown: https://aistack.sh/compare/chatgpt-vs-deepseek/llms.txt
ChatGPT ($20/mo Plus · API $2.50/M tokens, 1 stack) vs DeepSeek (Free chat · V4 API ~$0.30/$0.50 per M tokens (input/output), 1 stack). Both in the LLMs category.

## ChatGPT vs Gemini

URL: https://aistack.sh/compare/chatgpt-vs-gemini
Markdown: https://aistack.sh/compare/chatgpt-vs-gemini/llms.txt
ChatGPT ($20/mo Plus · API $2.50/M tokens, 1 stack) vs Gemini ($19.99/mo Advanced · API tier-based, 1 stack). Both in the LLMs category.

## ChatGPT vs Grok

URL: https://aistack.sh/compare/chatgpt-vs-grok
Markdown: https://aistack.sh/compare/chatgpt-vs-grok/llms.txt
ChatGPT ($20/mo Plus · API $2.50/M tokens, 1 stack) vs Grok ($30/mo SuperGrok · $40/mo X Premium+ (X bundled), 1 stack). Both in the LLMs category.

## ChatGPT vs Meta AI

URL: https://aistack.sh/compare/chatgpt-vs-meta-ai
Markdown: https://aistack.sh/compare/chatgpt-vs-meta-ai/llms.txt
ChatGPT ($20/mo Plus · API $2.50/M tokens, 1 stack) vs Meta AI (Free chat · Open weights · API via Groq/Together metered, 1 stack). Both in the LLMs category.

## ChatGPT vs Mistral

URL: https://aistack.sh/compare/chatgpt-vs-mistral
Markdown: https://aistack.sh/compare/chatgpt-vs-mistral/llms.txt
ChatGPT ($20/mo Plus · API $2.50/M tokens, 1 stack) vs Mistral (Free chat · $14.99/mo Pro · API metered, 2 stacks). Both in the LLMs category.

## Claude Code vs Codex

URL: https://aistack.sh/compare/claude-code-vs-codex
Markdown: https://aistack.sh/compare/claude-code-vs-codex/llms.txt
Claude Code (Bundled with Claude Pro $20/mo · Max $100-200/mo · API metered, 2 stacks) vs Codex (Bundled with ChatGPT Plus $20/mo · ChatGPT Pro $200/mo · API metered, 1 stack). Both in the Coding agents category.

## Claude Code vs Cursor

URL: https://aistack.sh/compare/claude-code-vs-cursor
Markdown: https://aistack.sh/compare/claude-code-vs-cursor/llms.txt
Claude Code (Bundled with Claude Pro $20/mo · Max $100-200/mo · API metered, 2 stacks) vs Cursor (Free Hobby · $20/mo Pro · $60/mo Pro+ · $200/mo Ultra · $40/seat Business, 5 stacks). Both in the Coding agents category.

## Claude Code vs Windsurf

URL: https://aistack.sh/compare/claude-code-vs-windsurf
Markdown: https://aistack.sh/compare/claude-code-vs-windsurf/llms.txt
Claude Code (Bundled with Claude Pro $20/mo · Max $100-200/mo · API metered, 2 stacks) vs Windsurf (Free · $20/mo Pro · $40/seat Teams · $200/mo Max · $60/seat Enterprise, 1 stack). Both in the Coding agents category.

## Claude vs CodeRabbit

URL: https://aistack.sh/compare/claude-vs-coderabbit
Markdown: https://aistack.sh/compare/claude-vs-coderabbit/llms.txt
Claude ($20/mo Pro · Sonnet API $3/$15 per M tokens (input/output), 33 stacks) vs CodeRabbit (Free for OSS · $12/mo per dev Lite · ~$24/mo per dev Pro, 2 stacks). Both in the LLMs category.

## Claude vs DeepSeek

URL: https://aistack.sh/compare/claude-vs-deepseek
Markdown: https://aistack.sh/compare/claude-vs-deepseek/llms.txt
Claude ($20/mo Pro · Sonnet API $3/$15 per M tokens (input/output), 33 stacks) vs DeepSeek (Free chat · V4 API ~$0.30/$0.50 per M tokens (input/output), 1 stack). Both in the LLMs category.

## Claude vs Gemini

URL: https://aistack.sh/compare/claude-vs-gemini
Markdown: https://aistack.sh/compare/claude-vs-gemini/llms.txt
Claude ($20/mo Pro · Sonnet API $3/$15 per M tokens (input/output), 33 stacks) vs Gemini ($19.99/mo Advanced · API tier-based, 1 stack). Both in the LLMs category.

## Claude vs Grok

URL: https://aistack.sh/compare/claude-vs-grok
Markdown: https://aistack.sh/compare/claude-vs-grok/llms.txt
Claude ($20/mo Pro · Sonnet API $3/$15 per M tokens (input/output), 33 stacks) vs Grok ($30/mo SuperGrok · $40/mo X Premium+ (X bundled), 1 stack). Both in the LLMs category.

## Claude vs Meta AI

URL: https://aistack.sh/compare/claude-vs-meta-ai
Markdown: https://aistack.sh/compare/claude-vs-meta-ai/llms.txt
Claude ($20/mo Pro · Sonnet API $3/$15 per M tokens (input/output), 33 stacks) vs Meta AI (Free chat · Open weights · API via Groq/Together metered, 1 stack). Both in the LLMs category.

## Claude vs Mistral

URL: https://aistack.sh/compare/claude-vs-mistral
Markdown: https://aistack.sh/compare/claude-vs-mistral/llms.txt
Claude ($20/mo Pro · Sonnet API $3/$15 per M tokens (input/output), 33 stacks) vs Mistral (Free chat · $14.99/mo Pro · API metered, 2 stacks). Both in the LLMs category.

## Claude vs Perplexity

URL: https://aistack.sh/compare/claude-vs-perplexity
Markdown: https://aistack.sh/compare/claude-vs-perplexity/llms.txt
Claude ($20/mo Pro · Sonnet API $3/$15 per M tokens (input/output), 33 stacks) vs Perplexity ($20/mo Pro · API tier-based, 3 stacks). Both in the LLMs category.

## Claude vs Qwen

URL: https://aistack.sh/compare/claude-vs-qwen
Markdown: https://aistack.sh/compare/claude-vs-qwen/llms.txt
Claude ($20/mo Pro · Sonnet API $3/$15 per M tokens (input/output), 33 stacks) vs Qwen (Free chat · API metered via DashScope, 1 stack). Both in the LLMs category.

## Codex vs Cursor

URL: https://aistack.sh/compare/codex-vs-cursor
Markdown: https://aistack.sh/compare/codex-vs-cursor/llms.txt
Codex (Bundled with ChatGPT Plus $20/mo · ChatGPT Pro $200/mo · API metered, 1 stack) vs Cursor (Free Hobby · $20/mo Pro · $60/mo Pro+ · $200/mo Ultra · $40/seat Business, 5 stacks). Both in the Coding agents category.

## Cursor vs Windsurf

URL: https://aistack.sh/compare/cursor-vs-windsurf
Markdown: https://aistack.sh/compare/cursor-vs-windsurf/llms.txt
Cursor (Free Hobby · $20/mo Pro · $60/mo Pro+ · $200/mo Ultra · $40/seat Business, 5 stacks) vs Windsurf (Free · $20/mo Pro · $40/seat Teams · $200/mo Max · $60/seat Enterprise, 1 stack). Both in the AI IDEs category.

## DeepSeek vs Qwen

URL: https://aistack.sh/compare/deepseek-vs-qwen
Markdown: https://aistack.sh/compare/deepseek-vs-qwen/llms.txt
DeepSeek (Free chat · V4 API ~$0.30/$0.50 per M tokens (input/output), 1 stack) vs Qwen (Free chat · API metered via DashScope, 1 stack). Both in the LLMs category.

## Descript vs Riverside

URL: https://aistack.sh/compare/descript-vs-riverside
Markdown: https://aistack.sh/compare/descript-vs-riverside/llms.txt
Descript ($24/mo Hobbyist · $35/mo Creator · $65/mo Business, 3 stacks) vs Riverside ($19/mo Standard · $29/mo Pro, 1 stack). Both in the Editing category.

## Feedly vs Perplexity

URL: https://aistack.sh/compare/feedly-vs-perplexity
Markdown: https://aistack.sh/compare/feedly-vs-perplexity/llms.txt
Feedly (Free · $6.99/mo Pro · $12.99/mo Pro+ (AI), 1 stack) vs Perplexity ($20/mo Pro · API tier-based, 3 stacks). Both in the Research category.

## Gemini vs Grok

URL: https://aistack.sh/compare/gemini-vs-grok
Markdown: https://aistack.sh/compare/gemini-vs-grok/llms.txt
Gemini ($19.99/mo Advanced · API tier-based, 1 stack) vs Grok ($30/mo SuperGrok · $40/mo X Premium+ (X bundled), 1 stack). Both in the LLMs category.

## Gemini vs Meta AI

URL: https://aistack.sh/compare/gemini-vs-meta-ai
Markdown: https://aistack.sh/compare/gemini-vs-meta-ai/llms.txt
Gemini ($19.99/mo Advanced · API tier-based, 1 stack) vs Meta AI (Free chat · Open weights · API via Groq/Together metered, 1 stack). Both in the LLMs category.

## Gemini vs Perplexity

URL: https://aistack.sh/compare/gemini-vs-perplexity
Markdown: https://aistack.sh/compare/gemini-vs-perplexity/llms.txt
Gemini ($19.99/mo Advanced · API tier-based, 1 stack) vs Perplexity ($20/mo Pro · API tier-based, 3 stacks). Both in the LLMs category.

## GenViral vs Ghost Feed

URL: https://aistack.sh/compare/genviral-vs-ghostfeed
Markdown: https://aistack.sh/compare/genviral-vs-ghostfeed/llms.txt
GenViral ($29/mo Creator · $49/mo Professional · $99/mo Business, 1 stack) vs Ghost Feed ($29/mo Starter · $79/mo Pro, 1 stack). Both in the AI video category.

## Ghost Feed vs Post Bridge

URL: https://aistack.sh/compare/ghostfeed-vs-post-bridge
Markdown: https://aistack.sh/compare/ghostfeed-vs-post-bridge/llms.txt
Ghost Feed ($29/mo Starter · $79/mo Pro, 1 stack) vs Post Bridge ($9/mo Starter · $18/mo Creator · $27/mo Pro, 2 stacks). Both in the AI video category.

## Ghost Feed vs Postiz

URL: https://aistack.sh/compare/ghostfeed-vs-postiz
Markdown: https://aistack.sh/compare/ghostfeed-vs-postiz/llms.txt
Ghost Feed ($29/mo Starter · $79/mo Pro, 1 stack) vs Postiz (Free self-hosted · $29/mo Standard · $39/mo Team · $49/mo Pro · $99/mo Ultimate (cloud), 1 stack). Both in the AI video category.

## Lovable vs Replit

URL: https://aistack.sh/compare/lovable-vs-replit
Markdown: https://aistack.sh/compare/lovable-vs-replit/llms.txt
Lovable (Free 5 credits/day · $25/mo Pro · $50/mo Business, 1 stack) vs Replit (Free Starter · $25/mo Core · $100/mo Pro, 1 stack). Both in the App builders category.

## Lovable vs v0

URL: https://aistack.sh/compare/lovable-vs-v0
Markdown: https://aistack.sh/compare/lovable-vs-v0/llms.txt
Lovable (Free 5 credits/day · $25/mo Pro · $50/mo Business, 1 stack) vs v0 (Free · $20/mo Premium · $30/user/mo Team · $100/user/mo Business, 1 stack). Both in the App builders category.

## Pika vs Runway

URL: https://aistack.sh/compare/pika-vs-runway
Markdown: https://aistack.sh/compare/pika-vs-runway/llms.txt
Pika (Free · $10/mo Standard, 2 stacks) vs Runway (Free · $12/mo Standard · $28/mo Pro, 1 stack). Both in the AI video category.

## Post Bridge vs Postiz

URL: https://aistack.sh/compare/post-bridge-vs-postiz
Markdown: https://aistack.sh/compare/post-bridge-vs-postiz/llms.txt
Post Bridge ($9/mo Starter · $18/mo Creator · $27/mo Pro, 2 stacks) vs Postiz (Free self-hosted · $29/mo Standard · $39/mo Team · $49/mo Pro · $99/mo Ultimate (cloud), 1 stack). Both in the Social and publishing category.

## Replit vs v0

URL: https://aistack.sh/compare/replit-vs-v0
Markdown: https://aistack.sh/compare/replit-vs-v0/llms.txt
Replit (Free Starter · $25/mo Core · $100/mo Pro, 1 stack) vs v0 (Free · $20/mo Premium · $30/user/mo Team · $100/user/mo Business, 1 stack). Both in the AI IDEs category.

## Suno vs Udio

URL: https://aistack.sh/compare/suno-vs-udio
Markdown: https://aistack.sh/compare/suno-vs-udio/llms.txt
Suno (Free · $10/mo Pro, 1 stack) vs Udio (Free · $10/mo Standard · $30/mo Pro, 2 stacks). Both in the Music category.

## Taplio vs Typefully

URL: https://aistack.sh/compare/taplio-vs-typefully
Markdown: https://aistack.sh/compare/taplio-vs-typefully/llms.txt
Taplio ($39/mo Starter · $65/mo Standard · $199/mo Pro, 1 stack) vs Typefully (Free · $12.50/mo Pro, 3 stacks). Both in the Social and publishing category.

---

# Tools

## AI IDEs

### Android Studio

URL: https://aistack.sh/tool/android-studio
Homepage: https://developer.android.com/studio
Pricing: Free · Play developer account $25 one-time · Updated: 2026-05-05

JetBrains-based IDE with Gemini in Android Studio for code completion, error explanations, and unit-test generation. Where Android apps build, profile, and sign.

Alternatives: Xcode

### Antigravity

URL: https://aistack.sh/tool/antigravity
Homepage: https://antigravity.google
Pricing: Free · $20/mo AI Pro · $249.99/mo AI Ultra · Updated: 2026-05-05

Google's answer to Cursor. Built on top of Gemini's 2M-token context, ships with a planner, an editor agent, and built-in code review. Strongest fit when your stack is already Google-heavy (GCP, Firebase, Android, Workspace integrations).

Alternatives: Cursor, Claude Code, Codex

### Cursor

URL: https://aistack.sh/tool/cursor
Homepage: https://cursor.com
Pricing: Free Hobby · $20/mo Pro · $60/mo Pro+ · $200/mo Ultra · $40/seat Business · Updated: 2026-05-05

VS Code fork with inline AI, agent mode for multi-file changes, and a built-in Bugbot reviewer. The default if you want the AI inside the editor instead of in a terminal or chat tab.

Alternatives: Claude Code, Codex

### Expo

URL: https://aistack.sh/tool/expo
Homepage: https://expo.dev
Pricing: Free · $19/mo Starter · $199/mo Production · Enterprise from $1,999/mo · Updated: 2026-05-05

The default starting point for React Native apps. Expo Router for navigation, EAS Build for CI signing, EAS Update for over-the-air JS bundles, and an SDK that smooths native module pain.

### Replit

URL: https://aistack.sh/tool/replit
Homepage: https://replit.com
Pricing: Free Starter · $25/mo Core · $100/mo Pro · Updated: 2026-05-05

Browser-based IDE with the Replit Agent that scaffolds, edits, and runs apps from prompts. Built-in deployment, hosting, and a database. Closest competitor to Lovable for the prompt-to-app use case.

Alternatives: Lovable, v0

### Windsurf

URL: https://aistack.sh/tool/windsurf
Homepage: https://windsurf.com
Pricing: Free · $20/mo Pro · $40/seat Teams · $200/mo Max · $60/seat Enterprise · Updated: 2026-05-05

VS Code fork with Cascade — an agent mode that plans + executes multi-step changes. Strong refactor performance, fewer per-turn cycles than Cursor's Composer on large diffs.

Alternatives: Cursor, Claude Code

### Xcode

URL: https://aistack.sh/tool/xcode
Homepage: https://developer.apple.com/xcode/
Pricing: Free · Apple Developer Program $99/year for App Store · Updated: 2026-05-05

Where iOS apps actually compile, sign, and ship. Pair it with an AI agent for the editing — Xcode is the build/run/profile/distribute surface, not the writing surface.

Alternatives: Android Studio

## Outbound and inbox

### Apollo

URL: https://aistack.sh/tool/apollo
Homepage: https://apollo.io
Pricing: Free · $49/seat/mo Basic · $79/seat/mo Professional (annual billing) · Updated: 2026-05-06

275M contacts, sequencing, and dialer. Where most cold outbound stacks source leads.

Alternatives: Clay

### Clay

URL: https://aistack.sh/tool/clay
Homepage: https://clay.com
Pricing: $185/mo Launch · $495/mo Growth · Updated: 2026-05-06

Pull from 50+ data providers, then call an LLM per row to write personalized outreach.

Alternatives: Apollo

### Gmail

URL: https://aistack.sh/tool/gmail
Homepage: https://mail.google.com
Pricing: Free · $7/user/mo Workspace Starter (annual) · Updated: 2026-05-06

Where everything starts. Most stacks bolt assistants onto Gmail rather than replace it.

### HubSpot

URL: https://aistack.sh/tool/hubspot
Homepage: https://hubspot.com
Pricing: Free · $20/mo/seat Starter · Updated: 2026-05-05

Pipeline, calendar, sequences. Free tier covers small teams.

Alternatives: pipedrive

### Smartlead

URL: https://aistack.sh/tool/smartlead
Homepage: https://smartlead.ai
Pricing: $39/mo Basic · $94/mo Pro · Updated: 2026-05-05

Unlimited mailboxes, deliverability, and warmup. The send engine for serious outbound.

Alternatives: instantly

### Superhuman

URL: https://aistack.sh/tool/superhuman
Homepage: https://superhuman.com
Pricing: $30/mo Starter · $40/mo Business · Updated: 2026-05-06

Splits, snippets, scheduled send, and AI replies. Gets a power user to inbox-zero fast.

Alternatives: shortwave

## Social and publishing

### Beehiiv

URL: https://aistack.sh/tool/beehiiv
Homepage: https://beehiiv.com
Pricing: Free up to 2,500 · $43/mo Scale · Updated: 2026-05-06

Built by ex-Morning Brew. Free up to 2,500 subs, then tiered. Strong analytics and a built-in ad network.

Alternatives: Substack

### Castmagic

URL: https://aistack.sh/tool/castmagic
Homepage: https://castmagic.io
Pricing: $29/mo Hobby · $99/mo Starter · Updated: 2026-05-06

Drop in an episode, get titles, show notes, timestamps, social posts, and clips.

Alternatives: castaudio

### Opus Clip

URL: https://aistack.sh/tool/opus-clip
Homepage: https://opus.pro
Pricing: Free · $15/mo Starter · $29/mo Pro · Updated: 2026-05-06

Long-form to TikTok/Shorts/Reels in one click. Picks viral-shaped moments and adds captions.

Alternatives: Castmagic

### Post Bridge

URL: https://aistack.sh/tool/post-bridge
Homepage: https://www.post-bridge.com/
Pricing: $9/mo Starter · $18/mo Creator · $27/mo Pro · Updated: 2026-05-06

Cross-platform short-form scheduler with native API integrations. Upload one vertical video, ship it to TikTok, Instagram Reels, YouTube Shorts, X, LinkedIn, Threads, Bluesky in one click. Where most short-form creators move once Buffer can't keep up with the platforms.

Alternatives: Postiz, Ghost Feed

### Postiz

URL: https://aistack.sh/tool/postiz
Homepage: https://postiz.com/
Pricing: Free self-hosted · $29/mo Standard · $39/mo Team · $49/mo Pro · $99/mo Ultimate (cloud) · Updated: 2026-05-06

MIT-licensed social media scheduling tool. Self-host on a VPS or take the hosted plan. Covers TikTok, IG, YT Shorts, X, LinkedIn, Reddit, Mastodon, Bluesky. AI integrations for caption + image generation.

Alternatives: Post Bridge, Ghost Feed

### Substack

URL: https://aistack.sh/tool/substack
Homepage: https://substack.com
Pricing: Free to publish · 10% of paid subs · Updated: 2026-05-05

Free to publish. 10% of paid subscriptions, 0% on free. Notes feed and recommendations push your work to readers already on Substack — the trade-off vs Beehiiv is network reach over ad and referral monetization.

Alternatives: Beehiiv

### Taplio

URL: https://aistack.sh/tool/taplio
Homepage: https://taplio.com
Pricing: $39/mo Starter · $65/mo Standard · $199/mo Pro · Updated: 2026-05-06

All-in-one LinkedIn tooling: hooks, scheduling, analytics, viral post search. Where most ghostwriting agencies live day-to-day.

Alternatives: buffer

### Typefully

URL: https://aistack.sh/tool/typefully
Homepage: https://typefully.com
Pricing: Free · $12.50/mo Pro · Updated: 2026-05-06

Distraction-free editor for threads + scheduling + analytics.

Alternatives: Taplio

## Image and design

### Canva

URL: https://aistack.sh/tool/canva
Homepage: https://canva.com
Pricing: Free · $15/mo Pro · Updated: 2026-05-05

The default for carousels, post images, and quick brand assets. Templates carry most of the weight.

Alternatives: Figma

### Figma

URL: https://aistack.sh/tool/figma
Homepage: https://figma.com
Pricing: Free · Pro from $12-$20 per seat/mo (seat-based) · Updated: 2026-05-06

When the asset is more than a square image. UI mocks, brand systems, complex compositions.

Alternatives: Canva

### Midjourney

URL: https://aistack.sh/tool/midjourney
Homepage: https://midjourney.com
Pricing: $10/mo Basic · $30/mo Standard · $60/mo Pro · $120/mo Mega · Updated: 2026-05-06

Frontier-quality image generation with the most consistent house style. Best when you need brand-grade thumbnails, ad creative, or marketing visuals. Discord-first historically; the web UI now matches the bot for serious work.

## Editing

### CapCut

URL: https://aistack.sh/tool/capcut
Homepage: https://capcut.com
Pricing: Free · $9.99/mo Standard · $19.99/mo Pro · Updated: 2026-05-06

The default editor for short-form creators. AI captions, auto-cut, voice generation, templates.

Alternatives: Descript

### Descript

URL: https://aistack.sh/tool/descript
Homepage: https://descript.com
Pricing: $24/mo Hobbyist · $35/mo Creator · $65/mo Business · Updated: 2026-05-06

Cuts, fillers, and overdubs by editing words. Studio Sound, eye contact correction, AI voices.

Alternatives: Riverside

### Riverside

URL: https://aistack.sh/tool/riverside
Homepage: https://riverside.fm
Pricing: $19/mo Standard · $29/mo Pro · Updated: 2026-05-06

Local-track recording for podcasts and interviews. AI editor cleans and chops episodes.

Alternatives: Descript

## LLMs

### ChatGPT

URL: https://aistack.sh/tool/chatgpt
Homepage: https://chatgpt.com
Pricing: $20/mo Plus · API $2.50/M tokens · Updated: 2026-05-05

The default LLM for most workflows. Strong tool use, image generation, and a huge ecosystem of GPTs.

Alternatives: Claude, Gemini

### Claude

URL: https://aistack.sh/tool/claude
Homepage: https://claude.ai
Pricing: $20/mo Pro · Sonnet API $3/$15 per M tokens (input/output) · Updated: 2026-05-05

Anthropic's Claude is the LLM most ghostwriters and editors reach for when voice and tone matter. Strong at long context, structured output, and editorial-grade rewrites.

Alternatives: ChatGPT, Gemini

### DeepSeek

URL: https://aistack.sh/tool/deepseek
Homepage: https://chat.deepseek.com
Pricing: Free chat · V4 API ~$0.30/$0.50 per M tokens (input/output) · Updated: 2026-05-05

DeepSeek-V3 / R1 deliver near-frontier reasoning and code at a fraction of OpenAI / Anthropic API rates. Free chat tier, very cheap API. The default 'cheap LLM' for cost-sensitive automations.

Alternatives: Claude, ChatGPT

### Gemini

URL: https://aistack.sh/tool/gemini
Homepage: https://gemini.google.com
Pricing: $19.99/mo Advanced · API tier-based · Updated: 2026-05-05

Strong on multimodal input and 2M+ token context. Best when you need to feed a whole codebase, video, or a stack of PDFs.

Alternatives: Claude, ChatGPT

### Grok

URL: https://aistack.sh/tool/grok
Homepage: https://grok.com
Pricing: $30/mo SuperGrok · $40/mo X Premium+ (X bundled) · Updated: 2026-05-05

Bundled with X Premium+ and SuperGrok subscriptions. Real-time access to X, image generation (Aurora), and a less-filtered voice. The default for X-power-users and anyone who wants live-search grounding.

Alternatives: Claude, ChatGPT, Gemini

### Meta AI

URL: https://aistack.sh/tool/meta-ai
Homepage: https://meta.ai
Pricing: Free chat · Open weights · API via Groq/Together metered · Updated: 2026-05-05

The Llama family powers Meta AI in WhatsApp, Instagram, and meta.ai. Open-weight models also dominate the open-source LLM ecosystem on Groq, Together, and self-hosted Ollama. Best when you want frontier-quality output with weights you can inspect and run yourself.

Alternatives: Claude, ChatGPT, Gemini

### Mistral

URL: https://aistack.sh/tool/mistral
Homepage: https://mistral.ai
Pricing: Free chat · $14.99/mo Pro · API metered · Updated: 2026-05-05

Le Chat is the consumer surface; Mistral Large, Codestral, and the open-weight family (Mixtral, Mistral Small) cover most use cases. Strong code performance and EU data residency are the standout buys.

Alternatives: Claude, ChatGPT

### Qwen

URL: https://aistack.sh/tool/qwen
Homepage: https://chat.qwen.ai
Pricing: Free chat · API metered via DashScope · Updated: 2026-05-05

Qwen 3 family (open-weight) and the Qwen Chat surface. Strong multilingual coverage (especially Chinese), good code performance, and very competitive API pricing via DashScope.

Alternatives: DeepSeek, Claude

## Coding agents

### Claude Code

URL: https://aistack.sh/tool/claude-code
Homepage: https://claude.com/code
Pricing: Bundled with Claude Pro $20/mo · Max $100-200/mo · API metered · Updated: 2026-05-05

CLI agent that reads your repo, runs commands, and edits files autonomously. Long-running multi-file refactors without copy-pasting context every turn. Skills, hooks, and MCP servers extend it. Bundled with Claude Pro/Max; metered API access for heavy use.

Alternatives: Codex, Cursor

### CodeRabbit

URL: https://aistack.sh/tool/coderabbit
Homepage: https://coderabbit.link/aistacksh
Pricing: Free for OSS · $12/mo per dev Lite · ~$24/mo per dev Pro · Updated: 2026-05-05

GitHub-native AI reviewer that reads diffs, leaves line-by-line comments, and suggests fixes. Catches bugs, security issues, and style drift before a human looks at the PR.

Alternatives: Claude

### Codex

URL: https://aistack.sh/tool/codex
Homepage: https://openai.com/codex
Pricing: Bundled with ChatGPT Plus $20/mo · ChatGPT Pro $200/mo · API metered · Updated: 2026-05-05

Two surfaces from one product. Codex CLI runs in your terminal with file system access. Codex Cloud takes a ticket and opens the PR autonomously. Bundled with ChatGPT Plus and Pro; the cloud lane bills as a separate task budget.

Alternatives: Claude Code, Cursor

### Manus

URL: https://aistack.sh/tool/manus
Homepage: https://manus.im
Pricing: Free credits · paid plans from ~$20/mo · see manus.im/pricing · Updated: 2026-05-05

Send it a goal (research a market, draft an analysis, run a small project) and it plans, executes, and returns a deliverable. Closer to a virtual analyst than a chatbot. The right pick for high-context, multi-step work you'd rather not babysit.

## Voice

### ElevenLabs

URL: https://aistack.sh/tool/elevenlabs
Homepage: https://elevenlabs.io
Pricing: $5/mo Starter · $22/mo Creator · Updated: 2026-05-05

The voiceover engine behind most faceless YouTube and audiobook stacks. Voice cloning, multilingual, low-latency streaming.

Alternatives: openai-voice

## Research

### Feedly

URL: https://aistack.sh/tool/feedly
Homepage: https://feedly.com
Pricing: Free · $6.99/mo Pro · $12.99/mo Pro+ (AI) · Updated: 2026-05-06

Train Leo to surface stories worth reading. The research mouth of most newsletter stacks.

Alternatives: Perplexity

### Granola

URL: https://aistack.sh/tool/granola
Homepage: https://granola.ai
Pricing: Free · $18/mo Individual · $14/seat/mo Business · Updated: 2026-05-06

Records calls in the background and produces structured notes. No bot in the meeting.

Alternatives: fireflies

### Perplexity

URL: https://aistack.sh/tool/perplexity
Homepage: https://perplexity.ai
Pricing: $20/mo Pro · API tier-based · Updated: 2026-05-05

Search that returns synthesized answers with sources. The research front end behind a lot of newsletter and content stacks.

Alternatives: Claude, Gemini

## AI video

### GenViral

URL: https://aistack.sh/tool/genviral
Homepage: https://www.genviral.io/
Pricing: $29/mo Creator · $49/mo Professional · $99/mo Business · Updated: 2026-05-06

Mines trending audios + formats from TikTok and Reels, then generates on-trend short-form video tailored to your niche. The viral-content engine that pairs with a scheduler like Post Bridge for the distribution lane.

Alternatives: Ghost Feed

### Ghost Feed

URL: https://aistack.sh/tool/ghostfeed
Homepage: https://ghostfeed.ai/
Pricing: $29/mo Starter · $79/mo Pro · Updated: 2026-05-05

An AI video content creation platform that takes you from idea to posted reel without manual editing. Pulls trends, generates short-form video (visuals + captions + voice), then schedules across TikTok and Instagram on a cadence you set. The 'one tool from idea to feed' lane for creators who want to post daily without a content habit.

Alternatives: Post Bridge, GenViral

### Pika

URL: https://aistack.sh/tool/pika
Homepage: https://pika.art
Pricing: Free · $10/mo Standard · Updated: 2026-05-05

Quick generative b-roll. Good at stylized motion when you don't need photoreal.

Alternatives: Runway

### Runway

URL: https://aistack.sh/tool/runway
Homepage: https://runwayml.com
Pricing: Free · $12/mo Standard · $28/mo Pro · Updated: 2026-05-06

Gen-3 text-to-video, image-to-video, and an editor with green-screen, inpainting, and motion brush.

Alternatives: Pika

## App builders

### Lovable

URL: https://aistack.sh/tool/lovable
Homepage: https://lovable.dev
Pricing: Free 5 credits/day · $25/mo Pro · $50/mo Business · Updated: 2026-05-05

AI app builder that handles UI, backend, database, and deployment from natural-language prompts. The right pick when you want a working product, not a component. Strong with Supabase + Stripe out of the box.

Alternatives: v0, Replit

### v0

URL: https://aistack.sh/tool/v0
Homepage: https://v0.dev
Pricing: Free · $20/mo Premium · $30/user/mo Team · $100/user/mo Business · Updated: 2026-05-05

Started as a UI generator, now a full-stack assistant: design, iterate, and ship web apps end to end with Git integration and a built-in editor. Ships to a Vercel preview URL without leaving the chat.

Alternatives: Lovable, bolt

## Music

### Suno

URL: https://aistack.sh/tool/suno
Homepage: https://suno.com
Pricing: Free · $10/mo Pro · Updated: 2026-05-06

Generates full songs (vocals + instruments) from a text prompt. Strong on hooks and stems.

Alternatives: Udio

### Udio

URL: https://aistack.sh/tool/udio
Homepage: https://udio.com
Pricing: Free · $10/mo Standard · $30/mo Pro · Updated: 2026-05-06

Often picked when you want a cleaner mix or different vocal styles than Suno.

Alternatives: Suno