Introducing: Walrus Memory. 🦭 A portable memory layer that lets your AI agents carry context across every app you run them in. No more starting from zero. No more being locked into one platform.…

Today, we're going after one of AI's most important unsolved problems. Introducing: Walrus Memory. 🦭 A portable memory layer that lets your AI agents carry context across every app you run them in. No more starting from zero. No more being locked into one platform. Portable, verifiable, and fully under your control. Take your agent's memory anywhere:

🎤 Walrus asked builders at Sui Live whether agent memory should travel with users across platforms and providers. Every answer was yes. Shared context across tools, fewer tokens burned re-explaining yourself, agents that pick up where the last one left off. Memory that lives wherever you signed up first isn't really yours. Data matters. So does who owns it.

We asked builders at Sui Live what would happen if their AI agent lost all of its memory tomorrow. Most said it would be brutal. A few mention having backups. Everyone agreed that starting from zero would be the worst outcome. Agent memory needs to be persistent and portable so that it moves with the user from agent to agent. That's what Walrus is for. 😎

I posted about @WalrusProtocol 2 weeks back. While drafting that, I came across a research paper they dropped in April. Finally sat down with it. TL;DR ↓ • Every decentralized storage network has the same problem: How do you keep data safe without making 25 copies of everything? • Arweave: Just make 25 copies. Safe, expensive. • Storj/Sia: chop files into pieces, 3 copies worth. Cheap, but if a node dies you have to download the whole file to rebuild its piece. Brutal at scale. • Walrus: 4.5 copies worth, but when a node dies, the replacement only downloads a tiny fraction of the file to rebuild what was lost. • Imagine the file as a grid instead of a list. Each node holds one row and one column. Lose a node and the new node grabs one cell from each surviving node, enough to mathematically reconstruct the missing row and column. When storage nodes swap in or out, most protocols pause writes during the handoff. Walrus doesn't. New writes go to the new nodes, old reads stay with the old. Every file is tagged with when it was written so clients know where to look. Decentralized storage used to force tradeoffs of low cost, fast recovery, OR no downtime. Walrus’ architecture somehow managed to get all three done. Disclosure: I’m holding $WAL

Most wearables sell your data back to you. @CUDISwellness flipped the model. Every heart rate and sleep score from 30k+ rings lives on Walrus. Users own the data, choose who sees it, and earn from sharing it. See how 🦭👇

We asked builders at Sui Live if they'd trust AI agents with their money and got a nearly 50/50 split. "I trust agents with my money rn." "I don't trust them at all." "Not yet, but soon." @SuiNetwork believes that the agentic economy will only grow from here. The real question is what closes the trust gap between now and mainstream adoption. It likely happens when agents can be verified, actions traced, and data proven. That is exactly where Walrus fits in. 🦭

📣 @GDanezis: The most valuable logic shouldn’t be the most vulnerable. “The skills file within a hedge fund of how to be a trader is probably going to be their secret sauce. They will not want that set of skills files and strategies to just go and live on some server, and they will not also want to lose it.”

“The difference between an agent and a chatbot is quite literally memory.” Chatbots are stateless. Agents are not. They need long-term context, persistent across days and tools. MemWal solves this 🙏🦭