A short list of things we will not build:
Systems designed to reduce headcount. We've turned away three projects in our first year for this. If the brief is "replace these two people with AI," we're not the studio. Our work returns hours to people who already exist — it doesn't replace them.
Customer-facing chatbots that pretend to be human. Disclose that it's AI, or we won't build it. Your customers will figure it out anyway, and the deception costs you more than the conversation does.
Systems we can't measure. If we can't write a clear success metric — hours saved, errors caught, throughput gained — we don't take the work. Vibes-based AI deployments are how people get burned.
Systems that need ongoing prompt engineering by the client. If you have to learn to "talk to" the AI to get good output, we built it wrong. Quietwork systems should be invisible — input goes in, output comes out, nobody needs to think about it.
Anything related to weapons, surveillance of individuals, or content moderation at scale. Not our thing.
This list keeps growing. That's a feature, not a bug. The point of a studio is to have a clear taste — what we make, what we don't, and why.