threadfork

Private meeting memory you control.

A local-first AI notetaker for consultants, founders, and operators who need editable transcripts, speaker-aware notes, and searchable meeting memory that stays on their Mac.

Watch demo

Platform standup

Jan 8 · 38 min · 3 speakers

Transcript

Speaker-aware capture

Live
Sarah Chen10:12

I think we should go with v2 prefix for all new endpoints.

Mike Rodriguez10:18

Agreed. Let’s document the migration path before Friday.

Priya Shah10:24

I’ll update the client SDK checklist once we decide.

Memory

Extracted as you work

Decision

Use v2 prefix for new API endpoints

Action item

Document migration path before Friday

Open question

Which endpoints need backwards-compatible support?

Search memory

api versioning decision

API decision found

Platform standup · 94% match

Local

Sarah recommended v2 prefixes for new endpoints; Mike agreed and asked to document the migration path.

3 speakers linked to meeting context
Stored locally on your Mac

In the app

See threadfork do the work

Live capture, clean summaries, search, and speaker-aware notes shown through real interface moments.

Demo

From transcript to searchable memory.

A 39-second walkthrough: transcript and summary generation, entity mentions, and the knowledge graph in action.

Across meetings

Your meetings become a map you can search.

Threadfork links people, projects, decisions, and terms across calls so meeting memory becomes cumulative instead of disposable.

Mentions become nodes automatically.

Related meetings stay connected to the source.

Search can follow concepts, not just keywords.

Meeting memory graph

People, decisions, projects, and terms stay connected.

Graph

Auto-linked from meeting context

Person

Sarah Chen

Decision

API versioning

Project

Client SDK

Team

Backend Team

Meeting

Platform standup

Search graph

api versioning

Selected node

API versioning

Mentioned across decisions, SDK work, and launch planning. Every edge keeps a source meeting.

Platform standup

94%

Sarah proposes the v2 endpoint prefix.

Client SDK review

87%

Migration checklist needs backwards compatibility.

Q1 launch planning

82%

Docs and customer comms depend on API naming.

Private by default

All the useful AI, none of the cloud leakage.

Keep the local-first promise front and center: no meeting bot, no silent upload, and no internet requirement once the models are installed.

On-device processing

Transcription, extraction, and indexing stay on your Mac.

No bot joins the call

threadfork listens from your computer audio instead of inviting a participant.

Works offline

Once installed, you can keep recording and searching without a connection.

You choose exports

Nothing leaves your device unless you decide to share or export it.

Private local workspace

Capture, process, and search without sending content out.

Input

Your Mac captures the call

Zoom / Meet / Teams

No meeting bot required

In-person or phone

Recorded from your device

On-device

AI runs inside your workspace

Mac local storage
Yours
Capture

Meeting audio

System audio and mic input stay on your Mac.

Process

Local AI models

Transcript, summary, entities, and actions run on-device.

Remember

Private index

Search and graph context are written to local storage.

Encrypted local files, user-controlled exports

Output

Useful memory, not uploads

Cloud path

Blocked by design

Meeting content is processed locally instead of being uploaded for transcription or AI.

Transcript.md

Speaker-aware notes

Entities.json

People, projects, terms

Graph index

Searchable memory

No silent uploadNo meeting botOffline after installExports only when you choose

Try it on your own calls

Ready to keep the meeting and lose the busywork?

Download threadfork for macOS, see the product on your own conversations, and keep every note local.

Works offline after model download. Requires macOS 14+ and Apple Silicon.