More software used to mean more engineers.

AI agents are breaking that relationship. Software can now move from requirement to production without a human reading every line. What happens next?

The software dark factory

An operating model where autonomous agents do most of the building, testing, reviewing, and deployment, while humans design the system, define the constraints, and stay accountable for the outcomes. The engineers don't disappear; the system moves them from writing the code to governing it.

1. Can a business trust code no human has read?
2. Who is accountable when no one wrote or reviewed it?
3. Do you still buy hours, or verified outcomes?
4. What replaces the team as the unit of production?
5. When code is nearly free, what stays expensive?
6. Which companies should never build one?

Questions we investigate with builders, operators, and leaders

Dany

Business systems

A venture designer who explores what happens to companies when software output stops tracking engineering headcount.

Vitaliy

Agentic engineering

A senior engineer who builds autonomous delivery systems and tests where they survive contact with production.

Both work at Camplight, where they build and test agentic software-delivery systems on real projects. Dany argues from the business side, Vitaliy from the machine room.

Guests join a running argument: what does a company become when its software output no longer depends on its headcount?

For guests · every two weeks

Episode one is recorded and the first five clips are cut. Watch them:

Five clips from episode one · full episode coming soon