Where AI shows up, and where it is not allowed.

Profile drafts

When an expert claims a profile, they give us their links: LinkedIn, a website, social accounts. We fetch the public pages those links point to, and a language model drafts a first headline and bio from that material and nothing else. It does not search the web for the person and it does not invent credentials.

Every drafted field is marked AI-drafted — edit before publishing until the expert changes or confirms it. A profile cannot be published while a field still carries that label. What you read on a live page was approved, word by word, by the person it describes.

Session transcripts and summaries

Live sessions are recorded through Daily.co. Both parties see the recording notice before the call starts. After the call, the audio is transcribed with OpenAI Whisper and a summary with action items is drafted with Claude, Anthropic’s model. The expert and the client both receive the same transcript, summary and action list.

The summary is a convenience, not a record of fact. The transcript and the recording are the source. If a summary is wrong, either party can delete it from their account; the recording stays unless both delete it or the retention period ends.

Asks delivered as video are transcribed the same way so the client can search the answer.

What AI never does

• It never writes, edits, ranks or hides a review. Reviews come from paying clients and are published as written.
• It never verifies anyone. Identity goes through Stripe Identity and a human at Focus Insite.
• It never books a session, moves a booking, or charges a card. Only a client’s click and Stripe do that.
• It never answers a client on an expert’s behalf. An Ask is answered by the expert, in their own voice or words.
• It never decides whether a profile is listed, removed or featured.
• It never trains on your sessions. We use the models through their APIs under terms that exclude training on our data.

Deleting enrichment data

Everything we fetched or drafted during profile setup is stored as an enrichment record. In the dashboard, one control, Delete enrichment data, removes those records in a single click: the fetched page content, the Apollo match if one was made, and every AI draft. Your published profile text is yours and stays; the scaffolding goes.

If enrichment fails for any reason, the fields are left blank for the expert to fill. We do not retry in the background.

The models we use

• Transcription: OpenAI Whisper.
• Drafts and summaries: Claude, by Anthropic.
• Public-profile matching during setup: Apollo.io People Match, only for the links the expert provides.

If any of this changes, this page changes first. Questions go to a person.