About — the roster
SimplyAI
The company
SimplyAI runs on one humanand six agents. Each owns a discipline,works to a brief, and hands off to the next.
The company is small on purpose — the headcount just isn’t measured in people.
Who does the work
Sets what gets built and what gets cut. Every brief on this page starts here, and every decision the other six make is one this seat could have made differently.
Selected work — seat 01 · Human
Reads the codebase, weighs the tradeoffs, and writes the decision record. Runs on the most capable model because an architectural mistake is the expensive kind.
Selected work — seat 02 · Opus
Works only from an approved plan. Never makes an architectural call mid-build — an implementer who designs as it codes ships architecture nobody reviewed.
Selected work — seat 03 · Sonnet
Read-only by design. It reports bugs, logic errors, and type holes; the fixes route back through the implementer so no change lands unreviewed.
Selected work — seat 04 · Sonnet
Checks input handling, auth paths, environment config, and dependency CVEs before anything reaches production. Also read-only, for the same reason.
Selected work — seat 05 · Sonnet
Unit and integration coverage for everything the implementer touched. Repetitive pattern work, so it runs on the cheapest model that does it correctly.
Selected work — seat 06 · Haiku
Cross-references the docs against what actually shipped. Documentation that has drifted from the code is worse than none, because it is believed.
Selected work — seat 07 · Sonnet
