Tom Reinholt
Principal engineer · Iowa City
Bio
I'm Tom. I'm forty-nine, I live in Iowa City, and I've been shipping software for a living for over two decades.
I started my career in safety-critical embedded systems — the kind of work where regulatory standards like DO-178C and IEC 61508 set the floor for what "done" means. That world taught me the difference between "tested" and "trusted," and I've never quite let go of the distinction.
The years since have moved through platform engineering across a handful of industries — insurance, ag-tech, healthcare integrations — and for the last several years I've been a principal engineer at a developer-tooling company, building agent-driven CI/CD and workflow systems. Go for the orchestration layer, TypeScript for the developer-facing surfaces, PostgreSQL for everything that matters.
I write here because the gap between what vendors say about agentic systems and what we observe running them in production has gotten wide enough to be worth talking about. I'm not a journalist. I cite primary sources, I try to acknowledge when I'm uncertain, and I try to be wrong in public so that people can correct me.
About Arctovec
Arctovec covers five areas: the gap between agentic AI pilots and production, the security model agents are quietly breaking, the shadow AI question every governance team is now confronting, AI-powered vulnerability discovery, and multi-agent system architecture.
The editorial principles are simple. Cite primary sources. Name names. Don't bury the lede. Numbers beat adjectives. Acknowledge counter-evidence. Recommend only tools I've actually used. The name comes from "arc to vector" — the idea that the messy, curved reality of these systems can be reduced to something you can reason about, if you're willing to do the work.
The site has no tracking beyond privacy-respecting analytics, no newsletter (yet), no popups, and no AI-generated body copy. The hero illustrations are AI-generated and prompted by me; everything else is written by a human in Iowa.