Through my career as a startup founder, and from working at several early-stage startups, one of the recurring problems I have come across is product-market fit. Not merely whether a product has product-market fit, but how we identify it in the first place.
This has become more important in the age of AI because building itself is no longer the difficult part that it once was. Products can be built faster, with fewer people, and at a fraction of the cost. In many cases, even the customer you are selling to can build a rough version of the product themselves using existing models and tools.
So the question is no longer simply: can this be built?
The question is whether we understand the customer’s problem precisely enough that what we build becomes important to them.
One method I use to think about this is to look at the delta between what a person is responsible for and what they actually spend their time doing.
Every person inside an organization has some core objective. A salesperson is responsible for generating revenue. A customer success manager is responsible for retaining customers and making sure they receive value. A recruiter is responsible for finding and closing the right candidates.
But much of their day is not spent directly doing that work.
The salesperson is updating the CRM, collecting information, writing follow-ups, preparing reports, and trying to remember what happened across several conversations. The customer success manager is searching through account history, compiling updates, coordinating internally, and documenting problems. The recruiter is scheduling, screening, taking notes, updating systems, and following up with candidates and hiring managers.
Some of this work is necessary. But it is not the core reason the person exists in the organization.
That delta—between what the person is responsible for and what they are actually doing—is one place where the customer pain point can be found.
What I mean by a customer pain point here is the work the customer has to do, often repeatedly, that pulls them away from the objective they are ultimately responsible for.
The next question is whether that delta has an economic consequence.
Is it costing the company time? Is it delaying revenue? Is it increasing risk? Is a highly paid person spending a meaningful portion of their day doing work that does not require their expertise? Has the company already hired people, created spreadsheets, purchased software, or built internal processes to manage it?
The customer may already be paying for the problem. They may be paying through labor, errors, delays, lost opportunities, or operational complexity. The opportunity is to understand whether technology can remove enough of that cost to matter.
Only after understanding that should we ask what AI can automate.
This ordering is important. The process should not begin with what the model can do and then move toward finding someone who may want it. It should begin with what the customer is responsible for, what prevents them from doing it, what that obstruction costs, and then what portion of it technology can reliably remove.
And reliably is the key word.
It is easy to build a demonstration in which AI performs a task once. It is much harder to build something that has the right context, produces a consistent result, handles mistakes, fits into the existing workflow, and can be trusted enough for the customer to depend on it.
That is also why customer understanding matters more as building becomes easier.
If the main value of the product is that it has access to a model or can generate an output, that value will not remain unique for very long. A competitor can copy it. The customer can use a general-purpose tool. Their internal team may be able to assemble a basic version themselves.
The defensibility comes from understanding the workflow at a deeper level: what information matters, when the task happens, what the person is trying to accomplish, where judgment is required, what can go wrong, and what needs to happen next.
The delta does not itself mean that there is product-market fit. It tells us where to look.
Product-market fit begins to emerge when the product reduces that delta and gives the customer something economically meaningful in return: more time, additional revenue, or lower costs. The value created by the product must be significantly greater than the price of the product. When the product costs much less than the problem it removes, it becomes relevant to the customer—and that is where product-market fit can begin.