I build things that work.

Most marketing teams scale by adding people. As ours was shrinking, I scaled it by building a system.

We were already losing people on the marketing team. Rather than backfill the roles, I built an AI system to absorb them — what began as a way to keep the work flowing became a full production pipeline: autonomous agents, orchestrated through a custom Anthropic API integration, that take on the work of an entire content function and run it on a monthly cadence behind a single human approval gate.

Three input streams — live market signals, brand knowledge encoded as Claude Skills, and real search performance — feed a central orchestrator that assembles shared context, enforces one consistent brand voice, and dispatches structured briefs to two specialist agents: a Social agent and an SEO agent, each on its own schedule. Everything routes to a human for sign-off before it publishes — the speed of automation without giving up editorial control.

The same output that used to require a seven-person function, produced by two agents running monthly and autonomously — removing roughly $520K in annual overhead. Not a prototype or a proof of concept. A live system doing real work.