Voice
How you sound, written as claims — and every claim quotes the sentence of yours that proves it. A claim with no quote behind it is dropped, not softened.
Check 11 of 14.
What happens
How you sound, proposed by a language model from your own copy — and then checked against it. At tier >= 2, the social leg's own YouTube-transcript quotes ride along as pass-through evidence, never re-analyzed by the model, and render even when the AI read itself did not run.
Calls it makes
OpenAI POST v1/chat/completions — gpt-5-mini, response_format json_schema with strict: true, token-cappedNo call of its own for video quotes — reads the social leg's own v1/youtube/video/transcript results (see the social leg's docs), tier >= 2 only, $0 extra
The facts a reader needs
| Language model | THIS IS ONE OF THE TWO. Every claim the model returns must quote the captured copy; a claim whose quote cannot be found is dropped before render. A failed call still writes its real token cost to the ledger — never zero, never omitted. Video-transcript quotes are NOT sent to this model at all — they are pass-through evidence, not a claim it made. |
| Cost | $0.002 per pull ($0.25 per 1M input tokens, $2.00 per 1M output). Video quotes add no model cost — their cost is the social leg's own YouTube-transcript credits. |
| Depends on | Reads the website leg's captures. |
| When it is absent | If every claim fails its quote check, the section reports that it produced none. A brand with no OPENAI_API_KEY still gets its own video quotes if tier >= 2 fetched them — the two evidence sources do not gate each other. |
Who supplies it
- ScrapeCreators — Social profiles and posts across seven platforms, the brand's own video transcripts, organic mentions, and the ad-transparency archives.
- OpenAI — The fallback door for four of the five model calls — the voice read, the confirm-step adjudicator, the quality review and the confirm-step suggestions — on a deployment that holds no Gateway key. The fifth, the AI-visibility grid, needs the Gateway and is parked in the alpha.
- Vercel AI Gateway — The door every model call goes through when AI_GATEWAY_API_KEY is set — the voice read, the confirm-step adjudicator, the quality review, the confirm-step suggestions and the AI-visibility grid. Unset it and the first four fall back to calling OpenAI directly, unchanged; the grid needs the Gateway and is parked in the alpha.
Typography
The typefaces actually declared and loaded, split into headings and body — resolved through CSS variables, so you get font names and never variable names.
AI visibility
What ChatGPT and Gemini say about you when someone asks, which sources they lean on to say it, and how often you are mentioned in AI answers at all.