Approach
How we intend to work.
This is our operating approach: how we decide what to build, and how we stay accountable while autonomous systems do the work. It is a method we hold ourselves to, not a record of finished projects.
Purpose first
An Engine has to earn its keep. A business that cannot support itself cannot keep serving anyone. But making money is the constraint, not the point.
We start Engines to make ordinary work lighter, to give small teams and communities tools that were only ever built for someone else's scale, and to show that autonomous systems can be run with honesty and care. That is the soul of the studio, and it is what we want each Engine to inherit.
In practice that means we choose problems by who they weigh on, not only by what a solution could earn, and we would rather run a few Engines that are plainly good for the people around them than many that are merely profitable.
What we believe
- Real problems before product ideas.
- Evidence before expansion.
- Useful products before impressive demos.
- Honest claims, controlled risk and human accountability.
- Fewer businesses done well, rather than an endless stream of launches.
Selecting problems
We start with problems, not product ideas. A good candidate is a task that recurs, costs someone real time or money, and is handled today with a workaround: a spreadsheet, a manual check, a process that breaks too often. We look for problems we can describe plainly, that a small focused product could genuinely remove, and whose removal helps more than the person paying for it.
Testing demand
Before building at scale we try to learn whether people want the problem solved and would actually use a solution. That can mean conversations, a narrow prototype or a limited pilot. If the evidence is weak, the idea stops there. Evidence comes before expansion.
Building narrowly
When something earns a build, we keep it small. One clear job, done reliably, beats a broad feature list. A useful product matters more to us than an impressive demo.
Measuring outcomes
Once something is in use, we measure whether it actually helps: does the task take less time, break less often, or stop needing the workaround? We report what we find honestly, including when the answer is no.
Autonomy and accountability
Autonomous AI systems do much of the work: research, drafting, building, testing and routine operations. That does not make them responsible for it. People set the direction, review the evidence, approve what ships, and answer for the outcome. Every Engine has a named human accountable for it.
We keep firm limits on what autonomous systems may do on their own, especially around money, customer data and public commitments, and we review those limits as we learn.
What we won't build
- Deceptive products, or anything that relies on misleading people.
- Manipulative subscriptions, dark patterns or traps that make leaving hard.
- Invasive surveillance of people who did not ask for it.
- Products that depend on spam or unwanted outreach to work.
If a business can only work by doing one of those, it is not an Engine, however well it might pay.
Know a problem worth solving?
What takes too much time, breaks too often, or still lives in a spreadsheet? Tell us what happens today and what better would look like. If it is the kind of problem an Engine should solve, we will say so.