Memory and governance are enterprise AI's defining problems in 2026. Organizations scaling AI hit the same wall: dozens of agents touch the same customers, deals, and accounts, but none of them share memory, enforce schemas, or follow the same organizational rules. I call the root cause the memory governance gap.
I'm the co-founder and CTO of Personize.ai, where I built the memory and governance layer for AI agents. The work sits at three intersections: autonomous agents running inside business workflows, agents operating over large and messy databases, and organizations running many agents from different vendors in the hands of employees with different knowledge and authority. The architecture spans dual memory stores, governance routing by authority rather than semantic similarity, and extraction pipelines that turn raw interactions into structured entity knowledge. It runs in production across 60+ enterprise customers in multiple industries.
My bet on the product side was to build for AI agents as the primary consumer. Developers, leaders, and business users are all increasingly working through agents and running more autonomous ones, so every surface we ship assumes an agent is on the other end: MCP servers, Skills, a CLI, open-source sample projects that encode best practices, and documentation written to be read by machines as much as by people. Memory, coordination, and access to complex data are treated as primitives, not as afterthoughts layered on top of a human-first CRM. That bet has driven every architecture decision since.
Two decades architecting AI products. I've built AI and data systems across GIS, ML, search, NLP, chatbots, knowledge systems, LLMs, and autonomous agents, from SMB to Fortune 500, with end-to-end ownership from novel architecture through enterprise deployment. I've worn most of the hats a technical founder wears: working directly with customers to figure out what problem is actually worth solving, designing the experiments and evaluation frameworks that tell you whether a change matters, and pushing on architecture where the off-the-shelf answer isn't good enough.
What I write here are the decisions behind the architecture, and the observations that didn't fit inside a paper or a product announcement. Most of it I had to figure out before I could find it written down anywhere.
I live and work in Vancouver.