← Compare

Bourdon vs Atlaso

Atlaso and Bourdon describe the same problem in nearly the same words: you tell one AI tool something, and the next tool has never heard of it. Both give Claude Code, Codex, Cursor, and other assistants a memory that persists across sessions and across tools. The resemblance ends at the architecture, and the difference is easy to state: Atlaso is a service you connect to. Bourdon is a file you own.

At a glance

AtlasoBourdon
Where memory livesAtlaso’s hosted platformPlain YAML on your own disk
CustodyAccount with a providerFiles you can grep, git, and take with you
Data egressContext flows through their service by designNothing leaves your machine unless you explicitly enable federation
FormatInternal to the platformPublished schema, Apache-2.0 wire format anyone can implement
Runtime modelAmbient context surfacingRecognition-first timing: ms-scale recall, background hydration (L0-L6)
IntegrationsClaude Code, Cursor, Codex, ChatGPT16 shipped participants + anything that speaks MCP
SourceClosed serviceSource-available (BUSL-1.1 engine, Apache-2.0 interop surface)
PriceFree tierFree, self-host forever

Atlaso claims are drawn from its public Product Hunt listing as of 2026-08-04. If Atlaso’s architecture changes, we’ll update this page; corrections welcome at [email protected].

The custody question

A memory layer ends up holding your most valuable working context: what you’re building, the decisions you’ve made, how you like to work. The central question to ask of any tool in this category is what happens to that context if the tool goes away, or if you simply want to leave.

With a hosted layer, your context lives in the provider’s database and your leverage is an export button. With Bourdon, there is nothing to export: the library is already a directory of plain YAML on your own machine, in a schema published under Apache-2.0. You can read it without Bourdon installed. That is not a feature we added; it is the design premise.

When Atlaso is the better fit

When Bourdon is the better fit

Try it

pip install bourdon && bourdon setup
# or the MCP server via npm:
npx @getbourdon/mcp-server --library ~/agent-library

Start at the landing page, or read the recognition-first thesis to see the runtime argument in full.