Hebbian

For builders

Skip building the memory layer.

Give your product, your systems or your agents a memory that stays current, through our API and MCP connector.

Works inside the AI tool you already use, or over MCP and the API.

A normal week

A week of building on it.

An example week at Touchgrass Software, a made-up software company that connects its own systems and agents to its company memory.

  1. Monday

    Data in

    The engineers send call notes and documents in through the capture API. Hebbian files them where they belong.

  2. Tuesday

    Their own systems

    An internal tool reads the company memory through the REST API before it drafts a reply to a contractor.

  3. Wednesday

    Their agents

    A reporting agent they built reads from the memory and writes to it, with its own identity and grant.

  4. Thursday

    Learning from use

    The notes their people rely on stay close at hand. What nobody needs stops getting in the way.

  5. Friday

    Usage insight

    They see which knowledge the company relied on this week, as totals, never per person.

How you use it

Your tools, your code, one engine.

Connect Hebbian to the AI tool you already use, then work there. People ask from Claude Desktop, Claude Code or Cursor. Your product, your systems and your agents use the REST API or the open-source MCP connector. The Hebbian app is for setup, admin and reading what is in memory.

claude_desktop_config.json
{  "mcpServers": {    "hebbian": {      "command": "npx",      "args": ["-y", "@hebbianvault/mcp"],      "env": { "HEBBIAN_API_TOKEN": "your_token" }    }  }}

Five ways in.

In your product

Add company memory to the software you sell. Each customer's data stays in its own tenant, kept apart server-side. We are building this with our first design partners, so talk to us.

In your own systems

Connect your internal tools and workflows to your company memory through the REST API.

For your agents

Give the agents you build a memory to read from and write to, each with its own identity and grant.

Data in

Send notes, documents and events in through the capture API. Hebbian files them where they belong.

Usage insight

See which knowledge your organisation relies on, as totals, never per person.

Yours and the company’s

The same ownership, through the API.

Each customer’s data stays in its own tenant, kept apart server-side. The company memory belongs to the company. Each person’s Professional memory and Private memory belong to that person, and nothing moves into the company’s unless they share it. An agent you build gets only the grant you give it.

How it holds
Each tenantKept apart, enforced server-side
The company memoryBelongs to the company
Professional and Private memoryBelongs to each person
An agent you buildIts own identity and grant

Trust

Where it runs.

Hosting

Hosted in the EU, with primary storage in Frankfurt.Hosted in the EU, with primary storage in Frankfurt. Our US region is not open yet and opens on request. Until we move your data there, it is hosted in the EU. Contact us to set a date. Some processing runs outside the EU. The sub-processor list names each provider.

Tenant isolation

Each tenant kept apart, enforced server-side.

We run it

We run the engine for you, so you build on it without operating it.

Trust center →

Tell us what you are building.

Sign up and create a token to try the API on your own memory. For a pilot, talk to us. Every engagement is quoted.

  1. Talk to usWhat you are building, and for whom.
  2. A pilotOne use, one tenant.
  3. Roll outMore tenants or more uses on the same engine.
Ask Hebbian