SOUVENANCE VS MEM0
They solve different problems, and both answers are right. Mem0 is a memory layer for developers building AI applications — an excellent one, near the top of the LoCoMo benchmark. Souvenance is a finished product for one person's life: capture, media, sleep cycles and governance included.
| Mem0 | Souvenance | |
|---|---|---|
| You are… | a developer shipping an app with memory | a person who wants to own their memory |
| Shape | SDK / API (managed platform or OSS library) | Complete self-hosted system: MCP server, iOS capture, pipelines, web timeline |
| Capture | You build it | Built in: share-sheet shortcut, daily voice log, perceptors, word-aligned transcription, vision, EXIF |
| Consolidation | Extraction pipeline per exchange | Sleep cycles: nightly consolidation, weekly REM recombination, active forgetting, reconsolidation |
| Time | Facts with update semantics | Bi-temporal graph, personal landmarks, prospective memory, humanised time framing computed in code |
| Originals | Out of scope | Voice, photo, video, PDF kept forever, replayable through signed links |
| Governance | Out of scope | Executable constitution, audit log, ratified persona, testament clauses |
| Substrate | Managed platform, or OSS you assemble | One PostgreSQL on your VPS; SQL dump = export |
USE MEM0 IF…
You're building a product and need user-level memory as a component. It's the right tool for that job — and Souvenance's own extraction approach cites Mem0's work as a reference. Good components deserve credit.
USE SOUVENANCE IF…
The memory in question is yours. You want the whole loop — effortless capture, nightly consolidation, "what did I think about X last year?" — running on your own hardware, governed by rules you can read in a YAML file, with your actual voice preserved for the replay.
A memory layer is a component you build with. Souvenance is an apparatus you live with. Choosing between them is choosing which side of that line you're standing on.
Related: what you can actually do with it · vs ChatGPT memory · how to own your AI memory