The First Notetaker for Squads & Agents.
Perfect memory of every team meeting and customer call.
Made for squads that ship, fast.
You don't need more notes. You need answers.
Your meetings are objects now, not transcripts. Ask in your own words from Claude or Cursor and get a sourced answer, not a wall of text to re-read.
- Onboarding stalled without bulk inviteVP Eng · discovery · Mar 4
- Wanted SSO before rolloutIT lead · Feb 19
- Seat pricing was unclearChampion · Feb 11
- "Too expensive per seat"7 calls
- "No annual discount"4 calls
- "Hard to justify vs free tools"3 calls
- Acme — "Procurement needs SOC 2"demo · Mon
- Globex — "Price vs the free tier"discovery · Wed
- Initech — "No Salesforce sync"eval · Thu
- Ship the experimentation framework behind a flagdecision
- Maya: wire the flag by Thursdayaction item
- Analytics backfill is the risk to watchrisk
It's querying, not guessing.
Squad Notetaker saves you tokens. Lots of them.
You only pull what you ask for, so a decision is a few hundred tokens, not a 16k transcript your agent re-reads and re-extracts every call. And every meeting is kept in two ready versions, full and redacted, so you feed a clean, right-sized one once.
agent → get_session(id)
{
"type": "internal",
"decisions": [
"Ship the experimentation
framework behind a flag
this sprint"
],
"action_items": [
"Maya: wire the flag by Thursday"
],
"takeaways": [
"Analytics backfill could slip
the rollout a week"
]
}agent → get_transcript(redacted)
[ greeting + weekend chat · 0:45 ]
[00:14] Maya: put the experimentation framework behind a flag this sprint
[00:31] Sam: the backfill could slip the rollout a week
[ off-topic: office logistics · 1:10 ]
[00:48] Maya: I'll own the flag, due Thursday
Two kinds of meetings. Two kinds of records. One notetaker.
Generic notetakers treat a retro and a customer call as the same summary. Fragment types each as what it is, so each is queryable on its own terms.
Why a squad picks Fragment
Frequently asked
You don't have to switch. If your current notetaker works for you, keep it. Fragment can hook into many existing notetakers and turn their output into typed records, so you get the structure without changing tools. If your vendor isn't supported yet, contact us and we will consider adding it. And if there is a feature you are missing, we would genuinely like to hear what it is.
No, by design. Your pipeline or agent does the routing. It pulls the typed action items and decisions over MCP and pushes them wherever you want. We would rather be the best source in your stack than another workflow tool fighting to own it.
Theirs returns transcripts and summaries, text your agent has to re-extract on every call. Fragment extracts structure once, at analysis time. Discovery calls become typed feedback objects (pains, requests, objections) the MCP returns directly; sprint planning and retros become decisions, action items, and takeaways your agent pulls from the session. Fewer tokens, a repeatable shape, no re-extraction.
Yes. Your sprint planning produces decisions and action items. Your discovery and user-testing calls produce customer evidence. Squad Notetaker extracts each as what it actually is, so you get one notetaker for both kinds of meeting.
Sprint planning lands as decisions, action items, and risks. Discovery calls land as pains, requests, and themes.
Start free, no card. Record your first meetings and connect your agent on the free tier. For squad-scale usage we will shape pricing around how your team actually meets, so book a demo or just start now and we will sort the plan when you outgrow free.
Sign up free, then add the Fragment MCP server to Claude or Cursor over npm or OAuth, about two minutes. Record a meeting and pull typed records straight from your agent. No export, no copy-paste.
Stop scrolling transcripts. Ask your meetings instead.
Start free, no card. Connect Claude or Cursor in two minutes.