Learning from use
Proposes a governed learning loop that turns user corrections and usage signals into improvements across knowledge, search, instructions, memory and model training, with evaluation and human approval before release.
Read the paperRESEARCH / ENTERPRISE AI
Memory that respects people and policy
A memory architecture that separates personal preferences from organisational policy, resolves conflicting information by type and scope, and links authoritative knowledge to current source documents rather than storing stale copies.
Read on this pageAt a glance
An employee's preferred answer format and a company's approval policy are both useful context, but they should not have equal authority. This technical report describes N-Tier Memory, a way for enterprise assistants to remember information while respecting who owns it and what kind of information it is.
The design separates personal preferences, shared procedures, policy and other memory types. It uses different rules to resolve conflicts and points authoritative memory back to current source documents. The report also describes the implementation and defects found during review. For enterprise buyers and builders, it explains why useful memory needs more than personalisation: it needs permission-aware retrieval, clear precedence and a way to avoid silently treating an old policy value as current knowledge.
What the paper covers
Memory is classified by both its owner and the kind of information it represents.
Personal preferences and organisational policy follow different conflict-resolution rules.
Authoritative memory refers back to versioned source documents rather than copying policy values.
Research context
The implementation and verification described are from a single-tenant deployment. The baseline assumes a simple organisational hierarchy; overlapping groups, automatic promotion to shared memory, a full administrative interface and other extensions remain outside the documented scope.
THE ORIGINAL WHITEPAPER
Download PDF Read it here.
The complete paper, with its original figures, tables and references.
44 pages · Read without leaving this pageFROM RESEARCH TO YOUR WORK
Discuss this research Bring your questions.
Discuss the approach, evidence and relevance to your team with the people building it.