I was very fortunate to work with the Berlin based entrepreneur Robin Jose this week, who gave an excellent presentation on MVP building to our current Cohort of Founders. The narrative of the presentation was the journey from Zero to One, and it was woven through a kind of bardic flow, linking calculus, psychology, artificial intelligence and, of course, product theory.
It is always a pleasure to hear someone else lay out their thought process and critical path. It wasn’t that there was anything particularly revolutionary in the presentation, but the narrative cohesion and theme linking made it stick, for me. That is what I walked away smiling about. Not the individual ideas, but the way they held together.
“Starting from zero can be a gift. If you don’t have much to begin with, you don’t have much to lose.”
James Clear
I was not thinking about the changes in how we develop products in 2026, because that topic is always in my purview. What Robin got me thinking about was the importance of two things.
Firstly, the human in the loop. We’ve used that phrase in modern AI forever (relatively speaking), and in AI and ML for decades before that, but not in this sense. Here I am talking about something more fluid: the human ability to shift gears in real time, to adapt to the capabilities and personalities in the room.
Empathy, compassion, experience, been-there-done-that, not backed by the sum total of all human learning, but by the well-honed craft skills of an individual. Maybe an expert in an area, maybe familiar with some more, but neither an agent nor an LLM system prompted with a broad and improbable array of skills.
The empathy builds trust. The room opens up. And learning happens.
“We spend a lot time designing the bridge, but not enough time thinking about the people who are crossing it.”
Dr. Prabhjot Singh, Director of Systems Design at the Earth Institute
Secondly, from the product perspective, what really reinforced my own beliefs was a simple analogue. In a world where development effort tends to zero, it is ever more critical that the problem you are solving, the job you are doing, remains sacred.
“People don’t want to buy a quarter-inch drill. They want a quarter-inch hole.”
Clayton M. Christensen
It’s all I’ve talked about for as long as I have worked in product: love the problem, not the solution, ship fast to your early adopters, get feedback, iterate, kill your favourites if they are not solving the problem you set out to solve. That hasn’t changed, but the context around it has.
My approach when starting all client or development work has always been to ask what problem we are solving, and how we hypothesise we will solve it today. From there, and importantly before hitting any sort of technical planning, to do some very expansive (and very quick, if required) discovery work without too many guardrails, in order to find out what we would build, if we could build anything.
Today, your favourite AI coding buddy will always suggest more. Their remit is utilisation, and they are very good at it. More features, more capability, more surface area, an almost gravitational pull toward expansion. Token-maxxing, scope creeping.
So today, as a founder, your job is to know what you want, to describe it clearly, even beautifully, to build it out, and to go no further until you have some feedback. And that is hard.
This approach was popularised in the Design Thinking era, and the once-fabled diverge/converge/diverge/converge double diamond approach to problem solving, or Design the Right Thing → Design Things Right. In my experience, this fell out of popularity, as many design or innovation (or development) methodologies often do.
There were good reasons for that. It front-loaded the development process with too much “fluffy” work, slowing down time to market, and when paired with scrum or your favourite agile flavour, it could make everything feel somewhat nebulous and insubstantial.
“Smile, breathe and go slowly.”
Thich Nhat Hanh
Today, though, the ground has shifted. The upside to front-loading the process, to focusing deeply on the problem, and designing the right solution to it, is a kind of superpower. Putting that effort into the left-hand side of your process, building out a robust specification, and then holding to it, even as your agents try to drift, forges a predictable outcome in a way that we haven’t really been able to achieve before. Test it, feed the results back in, rinse, repeat.
Where building is the cheapest it has ever been, judgment becomes the scarce resource.
These are not the exact words or themes that Robin touched upon, but they are the pictures that he conjured up for me. That the opportunity for the human, amidst the noise and uncertainty of the emergent AI world, is clear and present.
That the economics and resource intensity of shipping product have shifted beneath our feet, and that we now all have a responsibility to be advocates for our customers, more so than ever.
Because when everyone’s build effort tends to zero, we are all builders. The barrier to entry has dropped. And so the moats must be deeper.
Zero is dense now, full of everything that could be built. The work is in choosing what becomes one.