Skip to content
Conversations

Meetings and evidence

Every claim carries the sentence somebody actually said.

A meeting record holds one conversation: the date, the type, the attendees, the agenda and the transcript. An evidence record holds one claim type, one theme and the exact quote from a linked transcript. The save is refused if that quote is not in the transcript character for character, so no count in Growth-OS exists that cannot be walked back to a sentence.

4claim types behind every observation7themes the explorer clusters by0quotes accepted without a transcript match

In the app

What it looks like

Captures of Growth-OS itself, rendered with sample data rather than our own records.

An evidence record with the verified quote from its linked transcript.
An evidence record with the verified quote from its linked transcript.
A meeting record with its pasted transcript.
A meeting record with its pasted transcript.
The evidence explorer clustering observations by theme.
The evidence explorer clustering observations by theme.
One observation, its claim type and the account behind it.
One observation, its claim type and the account behind it.

How it works

From an empty record to something you can act on, in the order you do it.

  1. 1

    Save the meeting before it happens

    Date, type, attendees, agenda and the outcome you want. The preparation page pulls linked accounts, earlier conversations, evidence and battlecards.

  2. 2

    Paste the transcript

    Growth-OS does not record calls and does not transcribe audio. A person brings the text.

  3. 3

    Extract or write the observation

    The model proposes a quote and a character span. The server verifies that span against the transcript independently and stores nothing it cannot verify.

  4. 4

    Approve it

    A person accepts, or rejects with a reason. Approved observations cluster in the explorer by theme or claim type over seven, thirty or ninety days, against the window before it.

The problem this solves

What goes wrong when this work lives in spreadsheets, documents and one person's memory.

A number nobody can trace

Somebody says customers keep asking for this. Nobody can produce the sentence, so the room argues about memory instead of the claim.

Paraphrase drift

The quote gets tidied once, then again, and by the third retelling it means something else. A character comparison stops that on the first edit.

The insight that died in a thread

The best thing said on a call ends up in a channel nobody searches. Here it lives on the account, with the transcript behind it.

Highlights

Why it matters

The quote is checked, not trusted

The server compares the span against the transcript itself. A retyped or tidied quote fails at save time rather than entering the count.

AI proposes, a person approves

Meeting intelligence prepares observations as pending proposals. Approval writes them through the same service as a hand-written record, and rejecting one needs a reason.

Observations are dated by the conversation

An observation takes the date of its latest linked meeting, never the day the row was created, so a transcript entered six weeks late still counts in the week it happened.

Honest limits

What this does and does not promise

  • Growth-OS records and transcribes nothing. The claim type, theme and summary are generated and then reviewed, only the quote is verified, and nothing judges whether a claim is true or representative.

  • The explorer counts every evidence record whatever its status, so a draft nobody has checked is already in the totals.

Questions

Frequently asked questions

Do this work in Meetings and evidence

The page opens on this project's records. The in-app guide describes every field.