Memory has two layers: 1. How an agent selects, organizes, and retrieves what it learns 2. Where that knowledge persists, and who controls it The first makes memory useful. The second makes it durable, portable, and verifiable Walrus Memory is infra for the second ✔️ https://x.com/kimblgn/status/2119788084501873056 Quoted @kimblgn: There’s no standard for agent memory yet, and Anthropic’s post is upfront about that. What it does have is a pattern that keeps repeating across very different tests: 1. Purpose >> structure: Rules about what to record made the strongest models worse. A clear purpose worked better. 2. Quality beats quantity: Every note the agent saves is something it may have to sort through later. A useful memory should save the agent more work than it creates. 3. The record is the foundation: Underneath the notes is the full log of what happened. Wrong notes got fixed by checking them against it. 4. Memory needs upkeep: Notes go stale and some are wrong from the start. A regular pass that checks them against the log did more than writing better notes in the moment. None of these principles are tied to a particular model, even if how well a model uses memory clearly is. Which is why I’d keep memory separate from any one model or platform. That’s the best way to make sure that the next model can build on the work you’ve already done.(Historical earnings: For fiscal 2026FY (period ended 2026-09-03), Micron reported basic EPS of 75.38, diluted EPS of 74.33, net income of USD 84.969 billion, and revenue of USD 133.188 billion. Consensus expectations: For the upcoming 2026Q4 event, consensus expects EPS of 32.5625; no revenue estimate is available.)