Notifications
Clear all
Topic starter
19/08/2026 3:59 pm
TL;DR: The current wave of “memory” features spans four distinct architectures, but most products still amount to retrieval, compaction, consolidation, or filesystem-shaped harnesses rather than true persistent memory, according to Arize’s analysis. The gap matters because the missing primitive is still multi-evidence reconstruction, contradiction handling, and temporal supersession, and those failures shape how agents retain context and act over time.
NHIMG editorial — based on content published by Arize: Memory is still a missing primitive: Cataloguing what the field is actually shipping
By the numbers:
- HydraDB shipped a memory layer with a 90.79% LongMemEval-S claim.
- Anthropic’s compact-2026-01-12 beta header triggers at 150K tokens by default.
- GitHub Copilot Memory turned on by default with a 28-day TTL.
Questions worth separating out
Q: How should security teams govern memory in AI agents?
A: Treat memory as a privileged state store, not a convenience feature.
Q: Why does summarised agent memory drift over time?
A: Summarised memory drifts because each compression step rewrites prior context into a new approximation.
Q: How do security teams know if AI memory controls are working?
A: They know the controls are working when they can prove which entries were written, why they were accepted, who can reuse them, and whether retrieved context is checked before action.
Practitioner guidance
- Classify memory as governed state Map every persistent AI memory store to an owner, purpose, retention period, and access model.
- Separate source history from consolidated memory Keep immutable session transcripts alongside any summarised or consolidated memory artifact so teams can compare versions, investigate drift, and roll back bad consolidation decisions.
- Limit write access to memory artifacts Restrict which services, agents, or operators can modify long-lived memory.
What's in the full article
Arize’s full article covers the architectural detail this post intentionally leaves at the classification level:
- The four-bucket taxonomy with product-by-product placement and the reasoning behind each classification
- The benchmark and mechanism discussion behind LongMemEval, LoCoMo, ConvoMem, and MemBench comparisons
- The contrast between preserved memory stores and in-place rewriting in cross-session systems
- The specific engineering patterns behind filesystem-shaped memory, including LargeJson and PlanMessage
👉 Read Arize’s analysis of AI memory architectures and what is actually shipping →
AI memory systems are shipping, but what are they really doing?
Explore further
19/08/2026 4:36 pm
Memory is becoming an identity surface, not just a product feature. Once an agent can retain user context across sessions, the system is no longer stateless automation. That means write access to memory becomes as sensitive as read access to secrets or tokens, especially when memory can influence future decisions. For IAM and NHI programmes, the governance question is who can mutate the retained state and whether that state is lifecycle-managed like any other privileged asset.
A question worth separating out:
Q: What is the difference between retrieval memory and cross-session consolidation?
A: Retrieval memory fetches relevant stored context at runtime, while cross-session consolidation rewrites or reorganises past context into a new persistent artifact. Retrieval mostly affects what gets surfaced; consolidation changes what the system will remember next time. The second carries more truth-drift risk and demands stronger version control.
👉 Read our full editorial: Memory is missing in AI systems: what shipped this week
