Context the whole company can use.
Can one brain improve the whole workforce?
Remembering is not retrieving
Most “agent memory” is retrieval: text in, similar text out. Useful, but not the same problem — an agent that worked with you last month should know what was decided, what changed, and what is no longer true. We model memory as claims with two clocks (when it was true, when we believed it), so a store can answer what it believed last Tuesday and a correction can supersede a fact without erasing that it was once held.
A negative result we keep repeating
The intuitive design — a model reading every conversation and writing structured facts — did not earn its cost, and the failure is structural: extraction decides what to keep before anyone knows what will be asked, and what it drops is unrecoverable. The lever that worked sits on the read side: how you interrogate the store at the moment of the question beat anything we did at write time.
Recall no other memory system reaches.
FictionalCharacters QA on a matched gpt-4o-mini backbone. 22 published systems answer ≤7% of multi-hop questions correctly — Brain answers 57%.
| System | Single-hop | Multi-hop | Backbone |
|---|---|---|---|
| Naïve Brain | 91% | 57% | gpt-4o-mini |
| CAR (prior SOTA) | 78% | 30.2% | gpt-4o-mini |
| HippoRAG-v2 | 54% | ≤7% | — |
| MemGPT / Cognee | 28% | ≤7% | — |
| Mem0 / Contriever | 18% | ≤7% | — |
| Zep / Graphiti | 7% | ≤7% | — |
FictionalCharacters QA · % of held-out questions answered correctly, higher is better · Naïve measured, others published · July 2026
- Coming soonBoth Engines, One Harness: A Self-Run Head-to-Head Against mem0's Open-Source Server
- Status
- Research
- Focus
- Cost and usefulness (research)
- Surface
- Early / evolving
- Papers
- Both Engines, One Harness (preprint)