Persistent AI Memory Evidence Brief

Purpose

This brief separates the different AI memory claims on MikeKappel.com so reviewers can see what is proven, what is architectural context, and what remains deliberately bounded.

Safe claim

Michael's portfolio demonstrates persistent AI memory architecture as a software-governance pattern: compact local startup memory, explicit project/file handoff, durable curated notes, machine-readable discovery, and review-gated promotion from raw research into public evidence.

Memory lanes

• Local active memory: .uai/short-term-memory.uai, .uai/project-handoff.uai, .uai/file-handoff.uai, constraints, progress, and startup packets preserve task continuity for future work.
• Durable curated memory: /docs/ Architecture Notes hold reviewed context that can be linked from public pages.
• Shared runtime memory: Multi-Agent Memory demonstrates a separate coordination lane for external agents and reviewed durable memory.
• Research recall: /reports/ retains raw report material with explicit status labels and search-only/noindex handling where required.
• Machine discovery: llms.txt, llms-full.txt, .well-known/ai-agent.json, public route JSON, and evidence JSON expose the review order.

Best proof routes

• /ai-memory/
• /ai-memory-handoff/
• /docs/uai-memory-handoff-map/
• /docs/uai-memory-package-current-alignment/
• /docs/long-term-memory-map/
• /case-studies/multi-agent-memory/

Supporting report archive context

Representative background reports include:

• /reports/uaix-ai-memory-handoff/ai-memory-wizard-audit-plan/
• /reports/uaix-ai-memory-handoff/spec-wizard-and-docs-changes-for-long-term-memory/
• /reports/uaix-ai-memory-handoff/uaix-project-handoff-and-canonical-ai-memory-assessment/
• /reports/ai-wikis-agentic-web/local-active-memory-plus-hosted-matm-memory/
• /reports/uaix-ai-memory-handoff/project-handoff-system-and-canonical-ai-memory-uaix-and-llmwikis/

These reports help explain the design vocabulary. They do not prove automatic truth, automatic synchronization, credential authority, or autonomous write permission.

Claim boundary

Persistent AI memory here means governed continuity and reviewable context. It does not mean an AI memory system is always correct, automatically accepted, certification-bearing, or authorized to mutate private systems without human review.