Agent memory, confidential computing, and the economics of separating storage from compute.
An agent in a Kubernetes Agent Sandbox proves who it is with its own token, gets a Recall key for itself alone, and resumes after the sandbox is thrown away.
On 267 user-fact questions from LongMemEval, Recall 0.19 with an embedding model answered 92.9% with Claude Sonnet 5.5, against 88.0% for the whole 104,000-token history in context, on about 22,000 tokens. Keyword search answered 89.9%.
When an agent's pod dies, the next process picks up from records in your own bucket instead of a snapshot of the whole machine.
On one machine and one million vectors, Polign answered faster at the same recall, and a second agent can share the data.
The index stays in one bucket; only the machine in front of it changes how many searches a second you get.
Typed memory with fixed correction rules, on a database small enough for edge hardware.