Local-first knowledge graph
Shared context for your team and your agents.
Ingest a document once. Everyone on the team, and every AI agent they use, can traverse it, with the sources attached.
Free during beta · no invitation needed · your knowledge, connected
Local checks for sensitive data. Your choice of AI provider. Privacy, explained →
Organize
Organic Free-floating - the physics decides
Brain Your second brain, literally
Galaxy Spiral arms by type, hubs at the core
Orbits Most-connected at the center
Jarvis Sir, the vault is ready for review
Timeline When each piece of knowledge arrived
Grid A tidy census, one shelf per type
Nodes stay draggable - let go and they drift home.
Person 7
Organization 3
Project 3
Meeting 4
Decision 4
Concept 4
Repository 2
Conversation 1
Source 4
Agent 1
Skill 1
×
0 connections · created
vault/ .md
Graph chat asking about
What is this connected to?
Who is involved?
Where did this come from?
✓ claims · verified · est. $0.004
Demo answers are composed from the graph's own links and cite the files they touch.
Swipe to scroll the page. Use two fingers to zoom or move the graph. Tap an entity to explore.
Pause tour − +
Explore an entity Choose an entity
How it works
Your knowledge, connected in three steps.
01
Drop in what you already have
PDFs, decks, spreadsheets, email, Drive, Gmail, GitHub repos, exported chats. Screened for sensitive data before any model reads it.
Plus any PDF, Office, or text file, a pasted link, or a local git repo.
All sources ↓
02
An agent builds the graph, once
People, projects, decisions, meetings, and concepts each become one markdown file. Relationships are plain links. Every fact keeps its source.
wrote people/ada-patel.md linked → projects/relay.md updated people/README.md
03
Answers for people. Context for agents.
Teammates chat with the graph and get cited answers. Coding agents and assistants read the same folder over CLI or MCP. Nobody ingests the file twice.
✓ 3 claims · 3 verified
Sources & file types
Connect a source, then start a sync when you want to bring in new changes.
7 connected · 3 on the roadmap · 13 file types
Read-only. Pick folders or your whole Drive; only new or changed files sync.
Read-only. A metadata scan and a short interview set the scope before anything is read.
Browse your projects, private ones included, and choose which to track.
gitlab.com or self-hosted. Tracking a project clones and distills its code.
Any local path or clone URL, no account needed. Later syncs distill only new commits.
Imports your bookmarks via the official API, one distilled document each.
Import exported ChatGPT, Claude, or Gemini conversations: split locally, previewed, screened.
Channel and thread content, once this ships.
Pages and databases, once this ships.
Meetings and attendees, once this ships.
File types Anything with text ingests. The type is sniffed from the bytes, so a renamed file still routes correctly. Up to 25 MB each.
Shared context
Build the context once. Share it with the team and every agent.
One vault for you, your team, and your agents. Add knowledge once. Share it everywhere.
Every answer links back to its sources.
Hover the highlighted names →
What changed across Relay, and who owns the next steps?
Read projects/relay.md
Read people/ada-patel.md
Read meetings/relay-release-readiness-review.md
Relay is ready for a focused partner review. Relay keeps the first rollout bounded. Ada Patel coordinates the outcome and Julian Morgan owns delivery.
The standing decision is Keep account context traceable ; the next checkpoint is the release readiness review .
✓ 2 claims · 2 verified · $0.014
Context graph 14 of 35 files
Early results
OpenContext reduces token costs, API calls, and tool calls.
20.3% fewer model API responses. 16.4% fewer tool calls.
Preliminary results. Results may vary.
Based on preliminary testing on one coding task, codebase, and model. Cost and token figures average three paired comparisons; API response and tool-call figures average the two pairs with recorded counts. Costs combine recorded and transcript-estimated usage and exclude graph creation and maintenance. Quality was mixed across these preliminary comparisons.
Read the full analysis →
−20.6%
mean agent-run cost
$2.34 vs $2.94 across these three pairs
−33%
cache-read tokens
Mean reduction across the same three comparisons
Team token budget / month
$500,000
Potential savings / month
$100,000
Read the full analysisIllustrates a 20% reduction in your team’s token budget, not a measured team-wide result. Excludes graph creation, maintenance, and human time.
Try it on your own questions
Stop asking around. Ask the graph.
The question you would have posted in Slack today. Type it, open the workspace, and it is the first thing you ask.
Get started with the local beta
Who owns Relay's next release?
What changed this week?
Which decisions mention a customer?
What is our auth security stance?
Answers cite the files behind them
Browser workspace, no installation
Free during beta
Your teammates and agents read the same graph