Memory Infrastructure
Between you
and AI.
A database-backed context layer that every AI you use can read and
write.
Store once. Retrieve anywhere. Own your memory.
The Problem
Your AI starts from zero.
ChatGPT doesn't know what you told Claude.
Claude doesn't know your project constraints.
Cursor doesn't know your decisions.
You waste time re-explaining context. Vendors benefit from memory lock-in. You lose compounding intelligence.
The Solution
A shared memory layer.
Membrix sits between you and every AI system. It captures thoughts, decisions, constraints, and context — converts them into embeddings — and stores them in a standard PostgreSQL database.
How It Works
Three steps. One substrate.
Capture
Send a thought from any connected interface. The processing layer generates embeddings and extracts structured metadata.
Store
Raw text, vector embedding, and structured metadata are committed to your self-hosted PostgreSQL database.
Retrieve
Agents query via Model Context Protocol. Semantic search, recent logs, pattern analysis — one memory substrate, every AI.
Real Impact
The compounding mind.
Membrix is a compounding asset. Every captured thought strengthens future AI interactions. Here's what that looks like in practice.
The Handoff
Design → Development
The Scenario
Alex, a product designer, spends a week iterating on a new checkout flow with Claude. They establish specific constraints: "Error states must be inline, never modals, and always use the brand's neutral-600 color for text."
Without Membrix
When handover time arrives, the developer starts a new Cursor session. They ask the AI to build the checkout, but it generates default browser modals and wrong colors. The developer has to manually re-read Figma comments and re-prompt the AI.
With Membrix
Alex tells Claude, "Record these design constraints for the checkout flow." Later, the developer opens Cursor and types "Build the checkout component based on Alex's recent design rules." Cursor's MCP agent queries Membrix, retrieves the exact inline-error and color constraints, and generates perfect code on the first try.
Why Membrix
Infrastructure, not an app.
Infrastructure, not an app
Database-backed. Protocol-compatible. No UI dependencies.
Deployable in 45 min
Docker Compose up. Schema applied. MCP online.
Completely self-hosted
No SaaS. No vendor lock-in. Your server, your data.
Costs cents per month
Runs on minimal resources. Free-tier friendly.
You own the database
Standard PostgreSQL. Export, migrate, or query directly.
MCP-compatible
Any authorized AI agent reads and writes. Claude, Cursor, ChatGPT.
For Developers
Minimal surface area.
Composable architecture. No proprietary storage. That's it.
Vision
Browsers standardized documents.
Stripe standardized payments.
Membrix standardizes memory.
Between you and AI.