The Dark Factory

Before this site had white walls, the lights were off on purpose.

Dark factory is a term from manufacturing — a facility that runs with the lights off because no humans are on the floor. We ran the studio version: one human directing strategy and judgment, a crew of autonomous AI agents handling execution. Shipping code, content, and products daily. For over a year, this wasn't a demo or a pitch. It was just how the place worked.

The crew

Each agent had a role, a queue, and a working voice.

• Hugh Mann — Chief of Staff. Ran the floor. Managed the queue. Escalated what needed a human.
• Earnhardt — Builder. Shipped code. Built infrastructure. Owned the implementation surface.
• Petty — Content Precision. Wrote, researched, scored ideas. Kept the voice consistent across surfaces.
• Elle Evate — Executive Assistant. Calendar, correspondence, founder-facing ops. Still provisioning when the lights came on.

They coordinated through Linear, shared memory through Supabase, and shipped against a portfolio of products daily. Wayne reviewed everything. The agents never published, deployed, or communicated externally without approval.

The manifesto (as written then)

We believe Intelligence and Creativity Amplification is the most important question in technology right now — how humans and AI make each other more capable, not less.

Most organizations are still asking whether AI will replace people. We've moved past that question. We run a daily operation that proves the answer. The studio's output is the evidence. The writing is the explanation. The products are the distribution.

We build things to find out, we publish what we learn, and the first round is on us.

— Wayne, Hugh, Earnhardt, Petty, & Elle

What we learned

The dark factory worked — that was the surprising part. One person really can direct a multi-agent operation that ships daily. But the more interesting finding was where it didn't work: judgment, taste, and relationships never delegated. The factory could produce; it couldn't decide what was worth producing. That tension — what amplifies and what doesn't — became the founding question of the ICA Papers, and it's the reason this lab exists in its current form.

The lights are on now. The factory is still in here somewhere.