Technology
3 synchronized stories carry this tag.
ContextWeave reconstructs multi-month workflows from 14 participants into 1,005 executable tasks, including 568 core evaluations; the strongest memory raised workspace quality from 68.08 to 78.20 and preference alignment from 41.50 to 70.60.
Why it matters: Memory should be judged by whether it improves future work and preferences, not only whether an isolated fact can be retrieved.
ANCHOR evaluated 2,008 synthetic conversations across 27 personas, nine interaction schedules, three memory settings, and four models. Mean trajectory accuracy was 44.4%, user-state recall stayed near four-choice chance, and no context or memory setup consistently preserved continuity.
Why it matters: An assistant can sound coherent in each reply while gradually losing role boundaries, values, and consequential facts about the user.
Cloudflare previewed a memory service that extracts durable information from conversations, stores it outside the context window and retrieves relevant items for later sessions.
Why it matters: External memory can control context growth and preserve useful facts, but it creates a sensitive data store whose writes and recalls require policy.