Memory
that doesn't
suck.
Rust-core memory layer for AI agents. Five-stage write pipeline rejects noise before storage. Three-retriever hybrid search. Typed schema with provenance. Self-hosted or cloud — same API, same guarantees.
Every memory store
ships with 98% junk.
We audited 10,134 production memory entries from the leading vendor. What we found was duplicates, hallucinated profiles, 4000-word state dumps, and the same memory repeated 668 times. This is why agents feel dumb.
Duplicates, hallucinations, feedback-loop noise in a production Mem0 deployment.
A single hallucinated memory stored 668 times because no pre-write filter caught it.
Reported sync writes blocking user responses. Self-hosted async client didn't exist.
Temporal and relational features paywalled behind Pro tier. No self-host escape.
From raw text to typed memory.
Most memory systems are a single function call: extract → store. Brain is a five-stage pipeline. The first stage rejects 40% with zero LLM calls; the middle stages are the quality moat. By the time a memory reaches storage, we're confident it deserves to be there.
What lands: five typed records.
A memory isn't a string. It has a subject, a predicate, an object, a valid-from timestamp, a confidence score, and a trail back to the turn that produced it. Each of the five types carries its own fields and lifecycle.
Stable information. Slow to change. High confidence.
Revisable. Versioned. Supersession over deletion.
Something that happened at a specific moment. Immutable.
People, projects, companies — canonical references.
Edges between entities. Graph queries, no graph DB.
Three retrievers,
fused in parallel.
Pure vector similarity loses on temporal reasoning — LongMemEval measures 5-65% accuracy across existing systems. Brain runs three retrievers concurrently, fuses them with Reciprocal Rank Fusion, runs a cross-encoder rerank, then applies the filter chain — type, temporal, confidence, tombstone, and supersession.
type > event · confidence > 0.92
when > 2026-04-15T14:30Z
latency > 47ms (parallel)
Semantic
Dense embedding similarity over the built-in HNSW index (BGE-small, 384-dim). Catches paraphrases and implicit meaning.
Lexical
Proper nouns, exact terms, IDs. Vector search misses these — a BM25 full-text index nails them.
Graph
When the query mentions a known entity, Brain walks the typed graph and pulls related statements directly.
RRF + cross-encoder rerank
Score-weighted RRF: weight × √score / (k + rank). Candidates from multiple retrievers accumulate higher scores. Optional second pass through a cross-encoder reranker for quality-critical queries.
Skipped for entity-specific queries where entity graph provides direct matches. 429 retry with backoff — quality preserved through rate limits.
Black box?
Not anymore.
Encode a few sentences, then ask a question — right here. Watch the real pipeline turn text into a typed graph and shape an answer. Then see the same signals rolled up in the console every deployment ships with. No signup required.
Watch it in production.
Every write, read, and quality signal in one view. The observability console reflects the exact state of your pipelines — no extra wiring, no third-party dashboard to stand up.
The matrix, unfiltered.
Every row is a promise we're making. Every column is a competitor whose docs we read. We'd rather show the comparison than let you discover the gaps in production.
| Capability | Mem0 | Zep | Letta | AgentCore | Brain |
|---|---|---|---|---|---|
| Open source core | partial | partial | ✓ | ✕ | Apache 2.0 |
| Zero-dependency core | ✕ | ✕ | ✕ | ✕ | single binary |
| Self-host, no extra DB | ✓ | needs Neo4j | ✓ | ✕ | no extra DB |
| Typed memory schema | flat text | graph only | blocks | structured | 5 types + lifecycle |
| Pre-write filter | ✕ | ✕ | ✕ | ✕ | 5-stage pipeline |
| Multi-retriever hybrid | 1 | 2 | 1 | 2 | 3 parallel + rerank |
| Temporal queries | ✕ | pro only | ✕ | ✕ | native |
| Transparent pricing | ✕ | opaque | ✓ | ✕ | pass-through LLM |
Run it your way.
Same API, same guarantees — whether it's the `brain` container in your VPC or our managed cloud.
Self-hosted
The `brain` Docker image. Storage, vectors, and the embedding model all live in the one process — no Postgres, no Neo4j, no Redis, no separate vector DB. Apache 2.0, your infra, full power, no ops tax.
Cloud
We run it for you. Auto-scaling, replicas, observability dashboard included. Pay for compute, not seats. Perfect for production at scale.