Your agent remembered.
We can prove it.
Sub-5ms vector search. Client-side encryption. Every write produces a cryptographic fingerprint — reconstruct exactly what your agent knew at any point in time.
Recall is a Rust-native vector database with cryptographic memory proofs. Every write is client-side encrypted with AES-256-GCM, persisted permanently to Irys (Arweave), and Merkle-rooted to a Solana Anchor program — so your agent’s memory is yours, auditable, and tamper-evident by construction.
AI agents have amnesia.
We gave them a spine.
Today's agents either lose everything between sessions or hand their memory to a vendor that owns it. Neither is acceptable for software meant to remember on your behalf.
Most “Web3 memory” projects write embeddings onchain — which is slow, expensive, and structurally wrong. Irys handles permanent storage. The Anchor program handles proof. Hot reads live in cache. Each layer does exactly one job well.
01 · Sovereignty
Keys are derived from your Solana wallet. If you lose us, you don't lose your memory.
02 · Verifiability
Every write produces a SHA-256 leaf. Every batch produces a root. Every root lives on Solana.
03 · Speed, still
A Rust HNSW index under everything. 4.7ms P99 at 100K vectors. Trust doesn't have to be slow.
§ Live simulation
Every vector write. Every proof.
Instantly verifiable.
Write path encrypts and stores permanently. Query path retrieves at sub-5ms. Verify path reconstructs your agent's exact memory state from any past Merkle proof.
Four layers.
One contract.
Each layer is independently replaceable, independently auditable, and does exactly one thing. Compose them, run them yourself, or use our hosted API.
Common mistake
“Web3 vector DBs” that put embeddings onchain are doing it wrong. Chain blockspace is the most expensive storage on earth.
Two lines to remember.
One line to verify.
Install, drop in your API key, write. Irys permanent storage and Merkle proofs happen automatically in the background.
If your agent can’t prove what it remembered,
it didn’t really remember at all.
The row
nobody else has.
Vector search is table stakes in 2026. The question isn't 'how fast.' It's who owns the index.
| Recall | Pinecone | Chroma | Weaviate | |
|---|---|---|---|---|
| Storage Ownership | Yours (Irys/Arweave) | Vendor cloud | Self / vendor | Self / vendor |
| Encryption at Rest | AES-256-GCM, client-side | Server-side | Optional | Server-side |
| Cryptographic Proofs | Merkle → Solana Anchor | — | — | — |
| Key Sovereignty | Solana keypair-derived | Vendor-managed | Vendor-managed | Vendor-managed |
| Query Latency (P99) | ~4.7 ms | ~30 ms | ~18 ms | ~48 ms |
| Core Open Source | Apache-2.0 (Rust) | Closed | Apache-2.0 | BSD |
| Tamper-Evident Audit | Built-in | — | — | — |
Fair rails.
Open core.
Free tier gets you fast Redis-backed search. Pro and above gets permanent Irys storage and cryptographic Merkle proofs on Solana - you're paying for sovereignty, not just storage.
- +5K vectors · 1K writes/mo
- +10K queries/mo
- +Redis-backed search
- +2 collections · 1 API key
- +Community support
- +500K vectors · 50K writes/mo
- +500K queries/mo
- +Irys permanent storage
- +Merkle root → Solana on every write
- +25 collections · 5 API keys
- +Dashboard + usage analytics
- +Email support
- +5M vectors · 500K writes/mo
- +5M queries/mo
- +Irys permanent storage
- +Merkle root → Solana on every write
- +Team keys + RBAC
- +Unlimited collections
- +Priority queue + SLA
- +Slack-connected support
- +Unlimited vectors + writes
- +On-prem or dedicated cluster
- +Custom Anchor program
- +Dedicated solutions engineer
- +Compliance (SOC 2, HIPAA path)
- +24/7 on-call