Now your
agents know
what you know.
Groundframe is the context layer for AI coding agents. It syncs your meetings, Slack, and iMessage into .context/ files inside your repos — so your agents read what you know. No briefing. No copy-paste.
The problem
Agents can't search for what
they don't know exists.
Without Groundframe
With Groundframe
“The unknown unknowns problem is unsolvable with pull-based retrieval. You can't ask for context you don't know you need.”
— Groundframe thesis
The architecture case
Push beats pull.
An agent can only pull context it knows to look for. It can't ask for the meeting it doesn't know happened.
Fully managed context engineering
Always briefed.
Never stale.
Every meeting, message, and decision — understood, organized, and ready for your agents.
Fully managed
Connect your sources and Groundframe handles everything else — classification, curation, freshness, delivery. Your agents are always up to date without any manual work.
Decision extraction
Groundframe pulls decisions out of meetings and Slack threads and maintains them as a living log. Your agent knows what was decided, when, and why — even if it happened last quarter.
Agent-agnostic delivery
Works with Claude Code, Cursor, Copilot, Windsurf, and any agent that reads files. Not locked to one ecosystem.
Multi-project scoping
One meeting can populate context for three repos simultaneously. Context classified once, delivered everywhere it's relevant.
Knowledge that compounds
After six months, your agents have context that would take days to reconstruct manually. The knowledge base gets richer as you work.
Zero git footprint
.context/ is gitignored by default. No meeting notes in version control. No merge conflicts on context files. No sensitive business context in your commit history. Your repo stays clean.
Team mode
Your team's context,
in everyone's repo.
Create a team, invite your teammates, and mark a project as shared. Every teammate's context that classifies into that project is pooled and synced to everyone — attributed by who it came from.
Personal context stays personal. Two cofounders on the same product each get the other's relevant meeting notes and Slack automatically — without sharing anything private.
How it works
Three commands.
Your agents know the rest.
Install & init
Install the CLI and run init in your repo. Groundframe reads your README, package.json, and git remote to understand the project — no manual configuration.
Connect your sources
Connect Granola for meeting notes, Slack for decisions and threads, iMessage for the conversations that live on your Mac. Groundframe handles OAuth and encrypts your credentials. Your keys never appear in logs.
Your agent knows it
Run watch and compiled context materializes in your repo — summaries, decisions, structured knowledge built from raw transcripts. Your agent reads it naturally. No tool calls, no briefing required.
Sources
Connect your
whole work context.
Granola
MeetingsAI meeting notes, synced automatically. Decisions and action items routed to the right project.
Slack
Channels + Slack ConnectPick the channels that matter — internal and Slack Connect. Threads become structured context.
iMessage
Local · PrivateReads a local copy of your Messages on your Mac. Stays on your machine and private by default.
Manual paste
AnythingPaste a note, an email, a doc. Claude classifies it and files it where it belongs.
More sources on the way. Tell us what you need.
Early access
Ground your agents.
Connect a source, link a repo, and your agents read what you know.
Stay in the loop
Not ready yet?
Drop your email and we'll keep you posted on new sources, team features, and what's next.