About

About Zander

I’m Zander, an engineer and product builder. I tend to think about product, infrastructure, and business constraints as parts of the same system. At Flowith, that means my work often crosses conventional functions because the interesting problems rarely respect those boundaries.

01 / Practice

What I build

Part of my work goes deep inside agent systems; another part sits around them: model platforms, reliability, cost, and product infrastructure. Together they answer the same question: how does a new capability become a product people can depend on under real constraints?

02 / Definition

How I think about the agent harness

I think of an agent harness as an operating system for an agent. It coordinates runtime, context engineering, memory, evaluation, and tool orchestration—and defines how those systems work together in real tasks. Its job is not to prescribe every step, but to create the conditions for reliable execution, recovery, and improvement.

03 / Questions

What I keep returning to

I’m less interested in what AI can do than in why it works reliably: how an agent earns room to reason, how context survives a long task, how state becomes durable, how a system recovers from failure, and how evaluation changes what it does next.

Outside product work, I study markets as another system of incentives, narratives, and decisions under uncertainty. Writing is how I make those ideas inspectable.

Earlier visual notes

Two maps from an earlier version of this site. The visual language changed; the underlying questions did not.

A hand-drawn map connecting AI systems, agents, memory, decisions, markets, and writing.
Systems for thought, agents, markets, and myself.
A hand-drawn night desk map connecting market context, risk, watchlists, and a trading decision loop.
Markets as a practice of decisions under uncertainty.