In product circles, calling a team a feature factory is an insult. It means shipping features nobody asked for and never checking whether they worked. Output with no outcomes. The term is meant to sting, and it usually does.
We think the problem was never the factory. Factories are repeatable, measurable, and relentless - those are virtues. A good factory is the most accountable thing in industrial history: every station instrumented, every defect traced, every batch measured against spec. The problem isn't that software teams are factories. It's that they're factories with the instruments torn out - shipping and shipping and never reading the dial.
Two failure modes, one root cause
The pejorative usually points at one of two patterns:
- Output theater - measuring story points and PR counts as if activity were the goal, so the team optimizes the proxy and the outcome drifts.
- Ship-and-forget - features go out, nobody checks whether the metric moved, and the roadmap fills with more features to compensate.
Both come from the same missing piece: no feedback loop from what shipped back to did it matter.
Build like a factory, measure like a scientist
The fix isn't to ship less or to add ceremony. It's to close the loop:
- Plan from real signal - support tickets, usage, sales calls - not the loudest opinion in the room.
- Build with agents on rails, repeatably, so the how-we-ship is consistent.
- Measure every change, split AI work from human work, and tie each feature back to the metric it was supposed to move.
When the metric didn't move, say so - out loud, in the same place you celebrate the ones that did. That honesty is the entire discipline. It's also what makes the loop tighten over time instead of just spinning faster.
Why AI raises the stakes
When writing code was the expensive step, sloppy measurement was survivable - you couldn't ship that much anyway. AI removes that natural brake. A team that can generate ten features a week without an outcome loop isn't ten times more productive; it's ten times better at manufacturing work nobody validated. The instruments stop being nice-to-have and become the thing that keeps velocity honest.
Reclaiming the term
So we're taking the insult back. A feature factory that measures what it ships isn't the thing to avoid - it's the thing to build. Repeatable and accountable. Fast and honest. More on the thinking behind this in our manifesto, and the mechanics in the product tour.
Measure what you ship - for real.
Put your team’s AI web-development on the outcome loop.
Get early access