The Blog

Operator notes

Field notes on running AI coding agents: gates, context economics, verification, and the craft of making agent work repeatable.

July 2026
A prompt a small model can follow is a contract, not a wish

Task, constraints, output format, temperature: the four-part prompt contract that makes 3B to 13B local models reliable in agent loops, and the order to tune in when output is bad.

July 2026
Client billing data doesn't belong in someone else's cloud

Client names, rates, hours, and invoice history are confidential business data. Why billing records belong in files you own, shared one document at a time.

July 2026
Why I Am Building Zedia Labs

A one person software shop building AI agent tools, field manuals, and services in the open. The belief system behind it, from the founder who still has the problem.

July 2026
I built a mission control for my AI agents

Agents that wake up briefed, a board that limits work in progress, criteria-gated loops, and offline local models that ship real work.

July 2026
Agent Loops with Local AI — build reliable agents on your own hardware

The Plan → Act → Observe → Reflect pattern that turns a 7B model on your laptop into a worker that finishes jobs — and why local models need loops the most.

July 2026
Your freelance business shouldn't live in someone else's subscription

What local-first means for your income, client, and invoice records — and the sixty-day math that ends the SaaS-dashboard discussion.

July 2026
Benchmark runs are data. Decisions are the product.

Why benchmark results rot in folders, and the three artifacts a workflow should produce instead: comparisons, trends, verdicts.

July 2026
Introducing the Agent Harness Kit — the operating layer your coding agent is missing

Three interactive field manuals for operating AI coding agents: the free doctrine manual, a harness generator you own, and an operator course. Now live.

July 2026
The gate matrix: the four-word safety system every AI agent needs

AUTO, NOTIFY, APPROVE, FORBID — how to decide what your agent may do on its own, and why one table prevents most agent disasters.

July 2026
Your agent's biggest bill isn't the model — it's the context you re-send

Prompt caching is the single biggest cost lever in agent loops. Stable prefixes, just-in-time retrieval, and hard budget stops.

July 2026
An agent that can't verify its work isn't autonomous — it's just fast

The generator–verifier gap, the test-editing incident every operator eventually meets, and machine-checkable definitions of done.