Cursor knows my repo. Claude knows my architecture decisions. Now they both know, without me retyping it.
⌘Engineering · 4 tools synced
Memory Infrastructure For AI
One memory
for every AI
you use.
Empirical is the personal memory layer that sits between you and every model. Save once. Your preferences, projects, voice, and decisions. Then recall them in ChatGPT, Claude, Cursor, and anything else you open tomorrow.
Start tracking right away:
Works with every major AIYou own your memoryNo vendor lock-in
Memories
—
Queries
—
Synced to
7 tools
Last saved
just now
Speaks MCP. Works wherever you do.
ClaudeChatGPTCursorWindsurfClaude CodeDiscordCopilot CLICodex CLI
The problem worth solving
You've explained your project to ChatGPT this week. And to Claude. And to Cursor. Empirical is one memory layer, so you only ever tell it once.
From the founders. Why we are building this.
How it works
A second brain that any model can read.
Capture context once. Empirical structures it, ranks it, and serves the right slice to whichever AI you are talking to. Automatically, on the fly.
Capture
notes • chats • docs
Index
semantic + episodic
Recall
contextual
Sync
any AI surface
empirical · core
01
Capture once
Drop notes, paste chats, or let Empirical watch the tools you already use. It pulls the signal: projects, preferences, decisions, voice. Without the noise.
MCP server, REST API, and CLI. Your data stays secure and exportable.
02
Structure & index
Memories are organized into a personal graph (semantic, episodic, and procedural), so the right slice is always one query away.
Auto-deduplicated, versioned, and ranked by recency × relevance × intent.
03
Recall, in context
When you talk to any model, Empirical injects exactly what is relevant. No copy-pasting backstory, no re-explaining yourself for the hundredth time.
Triggered server-side via MCP, or directly through the Empirical CLI.
04
Own it. Forever.
Export, delete, or move your memory between vendors any time. It is yours. Not OpenAI's. Not Anthropic's. Switch models without losing yourself.
Stored securely, never sold or shared. Open formats. SOC 2 in progress.
Built for the way you actually work
Designed around your context, not theirs.
Universal recall
One memory across supported surfaces: ChatGPT, Claude, Cursor, Copilot, Codex, and your terminal. All read the same context graph.
// works everywhereRead →
Scoped memory
Separate work from personal. Switch projects with one toggle. The right model gets the right slice. Nothing more.
// privacy by designRead →
Portable forever
Export everything as JSON. Stored securely, never sold or shared. No vendor can lock you in.
// you own itRead →
Time-aware
Empirical understands that the you from six months ago is different from the you today. Recent context wins by default.
// episodic memoryRead →
Audit every recall
See exactly what was injected into every conversation. Approve, redact, or rewrite. Memory you can inspect is memory you can trust.
// transparent by defaultRead →
Developer-grade API
MCP server, REST API, and CLI. Bring Empirical into supported agents and pipelines with the documented OAuth flows.
// hackableRead →
Who uses Empirical
For people who live across their tools.
Empirical gave me a separate memory layer when ChatGPT's built-in memory became unreliable. For long-running continuity work, that extra memory has been more than useful—it has been a blessing.
✓Thomas L. · Long-running continuity work
My notes, papers, and chat history are one searchable brain. I ask it questions a year later. It answers.
⟡PhD candidate · 2k+ memories
Product roadmap
What's Next for Empirical? You decide.
Vote on the product ideas that should move first, or submit the next workflow you want Empirical to handle.
View all ideas →No ideas submitted yet. Be the first to post one.
From the build
Latest build updates
LatestSep 3, 2026product
Blueprints now interview you before they install anything
A canned memory graph made your assistant sound like a template. The new guided setups interview you, branch on what you say, and turn the transcript into real memories your assistant uses from the first question you ask.
Read post →
Jul 30, 2026product
The Empirical CLI is becoming the front door
Installing Empirical should feel like choosing where you want memory to work, not like learning the internals of an integration stack.
Read post
Jul 30, 2026product
Empirical became a set of tools, not just one plugin
The memory layer is getting smaller, clearer, and more useful: one core skill, plus optional tools for work history, hard-earned lessons, and project follow-ups.
Read post
From the blog
Ideas for a better memory layer
LatestAug 25, 2026agents
Multi-agent workflows have an amnesia problem
Every agent in a pipeline starts from zero unless something manually threads context between them. Salesforce measured what that actually costs, and it is not small.
Read post →
Jun 27, 2026second brain
Your AI needs your second brain
If you already spent years capturing ideas, decisions, research, and notes, you should not have to re-explain all of it every time you open a new AI chat.
Read post
Jun 25, 2026vibe coding
Your vibe-coded app got serious faster than expected
The dangerous part was not that AI wrote code quickly. It was how easy it became to forget which shortcuts were only safe while the app was still a toy.
Read post
Ready when you are
Stop re-explaining yourself to every model. Start owning your memory.
Free to start · No credit card · Export any time