AI chat over your history

What it is

AI chat over your history turns your entire RecordX library into something you can talk to. Instead of opening one meeting and asking about it, you open a workspace-level chat and ask questions that cut across every meeting you have ever recorded, uploaded, or joined with RecordX. The chat pulls the relevant transcripts, synthesizes an answer, and cites every meeting and timestamp it drew from.

Per-meeting chat is great when you know which conversation the answer is in. Chat over history is what you reach for when you do not. A rep asks "what did the buyer at Acme say about our onboarding last month" and the chat surfaces the exact quote from the Aug 21 call, plus the follow-up on Sep 04. A founder asks "which meetings mentioned the Frankfurt datacenter" and gets a short list back with links. A support lead asks "summarize what our customers said about latency across all Q3 support calls" and gets a synthesis grounded in dozens of grounded excerpts.

It is a semantic search plus summarization layer on top of your workspace. Every answer is grounded in the transcripts it cites. If the answer is not in your history, the model says so instead of guessing. You can filter by date range, participant, and tag to narrow the corpus before you even ask.

Try chat over history free

How it works

  1. Every meeting is indexed. As soon as a meeting is transcribed, its content is added to a searchable index in your workspace. There is no separate indexing step to run.
  2. Open the workspace chat. From the app sidebar. It is not scoped to any single meeting.
  3. Optionally filter. Date range, participants, tags. The filter narrows the corpus the model considers.
  4. Ask a question. In any language. Semantic search finds the relevant meetings; the model synthesizes an answer from the transcripts it retrieved.
  5. Read the answer with citations. Every claim in the reply is grounded in a specific meeting title and timestamp, and each citation is a click back into the source.

Example prompts

These are the kinds of questions chat over history is built for.

The chat always cites its sources. If a claim in the answer is not covered by any meeting in the retrieved set, the model returns "not found in your history" instead of inventing one. That negative answer is often as useful as a positive one.

Where it lives

Chat over your history is inside the RecordX app, on every surface. Web on macOS and Windows, mobile on iOS and Android. The workspace chat is one panel that spans every meeting you have access to.

The same search is also exposed through the RecordX MCP server, which lets external MCP clients such as Claude Desktop or Cursor run the same search_meetings and fetch_transcript tools against your workspace. Same corpus, same grounding, different client. If you prefer per-meeting scope, AI chat per meeting is the panel that lives inside a single meeting page.

Privacy and retention

Every search runs inside your workspace. Chat requests do not train foundation models. Source audio is deleted after processing; transcripts stay in your account for one year and can be deleted at any time. All data is encrypted in transit and at rest. See the privacy policy.

On team plans, the search corpus is limited to the meetings your account has access to. If a meeting is restricted to a specific person or team, it is not searchable from your account. Deleting a meeting removes it from history immediately.

Pricing

Free tier includes a limited number of history queries per month. Pro raises the cap. Business removes the cap for the whole team and adds per-seat billing. See the current numbers on the pricing page.

Getting started

Sign up, record or upload a few meetings, then open the workspace chat from the sidebar. The index builds automatically as new meetings land. There is nothing to configure.

Start for free

FAQ

How is this different from per-meeting chat

Per-meeting chat scopes every answer to a single conversation. Chat over history searches and synthesizes across every meeting in your workspace, and each answer cites the specific meetings and timestamps it pulled from. If you already know which meeting the answer is in, use AI chat per meeting.

How far back does it search

As far back as your retention window. Transcripts stay in your account for one year by default and are all searchable. If you delete a meeting, it drops out of history immediately.

Can I filter to just my meetings vs the whole workspace

Yes. You can scope a query to only the meetings you attended, a specific participant, a date range, or a tag. On team plans, the workspace scope is limited to meetings you have access to.

What language does it work in

Ask in any language and answers come back in the language of the question. Meetings in mixed languages are searched and quoted in their original language.

Does it hallucinate or cite sources

Every answer is grounded in transcript excerpts, and each excerpt is cited by meeting title and timestamp. If the answer is not in your history, the model says so instead of guessing.

What does it run on

It runs on Amazon Bedrock, RecordX's standard model host. Your transcripts stay in your workspace and are not used to train foundation models.

Start for free