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
| Atlaso | Bourdon | |
|---|---|---|
| Where memory lives | Atlaso’s hosted platform | Plain YAML on your own disk |
| Custody | Account with a provider | Files you can grep, git, and take with you |
| Data egress | Context flows through their service by design | Nothing leaves your machine unless you explicitly enable federation |
| Format | Internal to the platform | Published schema, Apache-2.0 wire format anyone can implement |
| Runtime model | Ambient context surfacing | Recognition-first timing: ms-scale recall, background hydration (L0-L6) |
| Integrations | Claude Code, Cursor, Codex, ChatGPT | 16 shipped participants + anything that speaks MCP |
| Source | Closed service | Source-available (BUSL-1.1 engine, Apache-2.0 interop surface) |
| Price | Free tier | Free, 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
- You want zero infrastructure: nothing installed, nothing on disk, someone else operating the whole thing.
- You work primarily in hosted chat tools and want memory to follow you across devices without running anything locally.
- You’d rather have a managed product than a file and a runtime.
When Bourdon is the better fit
- Your working context is sensitive enough that “nothing leaves the machine by default” is a requirement, not a preference.
- You want the memory itself to be inspectable and portable: greppable files under version control, not rows in someone else’s database.
- You run multiple local coding agents over MCP and want them reading and writing one shared library with a published contract.
- You care about response timing: recognition answers in milliseconds from a hot cache while deeper context hydrates in the background.
Try it
pip install bourdon && bourdon setup
# or the MCP server via npm:
npx @getbourdon/mcp-server --library ~/agent-libraryStart at the landing page, or read the recognition-first thesis to see the runtime argument in full.