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.
Same hook generation as the recommended stack. The output is text — same prompt, same quality at the free or $20 tier.
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.
Fast mode for Claude Opus 4.8 is being deprecated and will be removed on November 30, 2026. Developers using fast mode with Opus 4.8 must migrate to Claude Opus 5.5 to continue using the feature.
Anthropic released the Usage and Cost Analytics API in beta, enabling organizations to monitor token usage and costs across Claude API consumption with filtering and grouping by dimensions like model, workspace, and API key. The API provides /v1/organizations/analytics/usage and /v1/organizations/analytics/cost endpoints. Legacy usage reporting endpoints remain supported.
The maximum request size for the Messages API and Token Counting API doubled from 32 MB to 128 MB on the first-party Claude API, allowing developers to process significantly larger payloads in single requests. This change takes effect immediately and applies to both APIs.
The Claude Tag API is now available in beta for Claude Enterprise organizations, providing endpoints to read and configure Claude Tag settings on Slack channels and workspaces with Access bundle support. The API requires Admin API keys with read:claude_tag or write:claude_tag scope. This feature is exclusive to Claude Enterprise.
Same prompt as the recommended stack. The bottleneck is the hooks, not the tools.
Product brief:
"""
{{paste product brief: what it is, who it's for, what makes it different, 1 customer quote}}
"""
Generate 30 short-form ad hooks for vertical video ads, split evenly across 3 angles:
1. **Pain hook** — open with the audience's problem in their words.
2. **Solution hook** — open with the surprising mechanic or result.
3. **Social proof hook** — open with a specific outcome from a named (or anonymized) user.
For each hook:
- Format: "{HOOK LINE} (angle: pain/solution/proof)"
- 8 to 14 words, spoken-word friendly, hard consonants up front
- No "Are you tired of", no "Imagine if", no "Stop scrolling"
- One concrete noun or number per hook
Output as a numbered list 1 to 30.Open CapCut, paste the hook, pick a voice from the AI voice library. Render at default settings.
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.
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.
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.
Built-in voices are good enough for testing, not for production. Once an angle wins, re-render with ElevenLabs before scaling spend.
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.
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.
What changed in AI tools and the stacks that ship. One email a week.