Walrus, the decentralized storage protocol built on the Sui blockchain, just launched something designed to solve this. The MemWal SDK is a developer toolkit that gives AI agents persistent, encrypted memory stored on Walrus’ decentralized infrastructure, complete with semantic search, so agents can actually retrieve what they’ve learned.
What MemWal actually does
The SDK stores encrypted memories on the Walrus network and adds semantic search to them. That means agents don’t just dump information in a blob of text. They can intelligently query their own memory and retrieve relevant context based on meaning rather than exact keyword matches.
Abinhav Garg, Group Product Manager at Mysten Labs (the team behind both Sui and Walrus), formulated the core value proposition around openness. The framework allows memory to reside at a data layer that is both open and verifiable, and not dependent on the infrastructure or whims of a single AI provider.
The Sui blockchain governs ownership and access control. So users, not companies, decide who can read, write, or share an agent’s reminders.
The integration story and the developer game
Walrus didn’t launch MemWal in a vacuum. The SDK comes with integrations for Vercel AI SDK, along with plugins for frameworks called OpenClaw and NemoClaw. Quickstart guides and extensive documentation are available, and the team actively solicits developer feedback via GitHub. The SDK is currently in beta.
Why decentralized memory is important for AI
MemWal’s thesis is that agent memory should be a user-owned infrastructure. The four pillars the team emphasizes are verifiability, availability, portability and shareability.
Verifiability means you can confirm that the memory has not been tampered with. Availability means that it continues even if a specific provider goes offline. Portability means it moves between models and suppliers. Shareability means that multiple agents or multiple users can work together via shared memory pools.
This builds on the work that Walrus and Mysten Labs have been developing since at least March 2025, when the first foundations for this type of decentralized storage infrastructure were laid. The MemWal SDK represents the most explicit step yet to position that infrastructure as an AI-specific tool.