MNEME gives AI agents persistent, verifiable memory. Agents store decisions, link outcomes, and recall context across sessions backed by 0G Storage and on-chain attestation.
Every AI agent today starts from zero. The moment a session ends, everything it learned, every decision it made, every outcome it observed disappears. There is no record of why it did what it did, and no way to prove it.
MNEME solves this by giving agents a structured memory store that persists across sessions. When an agent makes a decision, it stores that memory. When an outcome follows, it gets linked back to the original decision, forming a causal chain that shows exactly how one action led to another. Over time, an agent builds up a body of experience it can actually search and reason from.
Every memory is embedded using OpenAI and stored in a vector database so agents can retrieve relevant past context using natural language queries rather than exact lookups. The full memory content is uploaded to 0G decentralized storage, and each memory is cryptographically signed and attested on-chain via a deployed smart contract on 0G Chain. This means the memory cannot be altered or denied. Anyone can verify what the agent knew and when it knew it.
Mneme also integrates 0G Storage for permanent content persistence, 0G Chain for on-chain attestations through the MnemeAttestation contract, and exposes a TypeScript SDK so any agent can connect with three lines of code. A dashboard lets operators create agents, browse their memory history, inspect causal chains, and run semantic searches across everything an agent has ever stored.