Hamid Farmani's case for a new era of software fits in one sentence about subscriptions. He thinks AI now lets people build the two features they need instead of renting the twenty they do not. If he is right, that also changes what is worth tracking at all.
The short answer: Big organisations earn trustworthy data through interdependence and documentation; solo, he builds that discipline for his agent, and trusts a real person's need first.
What Hamid Farmani trusts, and for what
| Evidence | Trusted for | Not trusted for |
|---|---|---|
| Interdependence and documentation in large teams | Keeping data coherent at scale | Solo work, where nobody imposes it |
| Documentation written for his agent | Letting the agent find what each feature does | Deciding what to build |
| Conversations and scenario tests with people | What to build, if anything | Measuring use after launch |
| Approval before irreversible actions | Deletions and other changes an agent cannot undo | Routine steps |
At scale, interdependence keeps data honest
Almost fifteen years in tech took Hamid from QA into development, from Iran to Armenia to Sweden, and through automotive consulting and streaming before his current engineering role. In large systems, many services depend on each other, which forces teams to agree on how they build. More documentation, he argues, is a feature of scale rather than bureaucracy.
Solo, he documents for the agent
On his own projects nobody imposes that discipline, so he manufactures it. He asks his AI to keep good documentation for both of them, and structures folders so the agent can find what it needs. His collaborator is the agent, and the project is laid out for it.
my agent is going to help me, not another human being.
The answer might not be an app
His research-first habit comes partly from his wife's product design instinct: interview people, put them through scenarios, then decide. Sometimes the honest answer is a video, a course or a calculator rather than an app with tracking in it. The hard part of starting something, he says, is no longer the code but deployment, billing and the number of infrastructure choices.
You pay fifty euros for a subscription with twenty features, but you only need two.
What to do with this
Talk to people before you build, and track what they told you matters. Tracking twenty features nobody asked for gives you twenty charts and no answer; tracking the two that people need gives you a number you can act on.
For seven checks drawn from across the series, see the questions to ask before trusting a number.
Frameworks this interview tests
- Jobs to be done (JTBD): Clayton Christensen's jobs to be done says customers hire a product to make progress. The job decides which events matter, so interviews come before the tracking plan.
Read next in The Question
Source and disclosure
Read Issue 13 on A Founder's Friend. Quotations on this page are copied from the published issue.
Skene sponsored this research and commissioned it on a single question. A Founder's Friend ran the interviews, wrote them up and published them, and the series states that Skene paid for the research and did not direct the questions, the analysis or the conclusions. Each of the thirteen interviews was reviewed and signed off by the person in it, named or anonymous by their choice. Appearing in the series is not an endorsement of Skene, and nobody here is presented as a Skene customer.