Article
AI has made building products easier than ever, but the basics haven’t changed. Product leadership is still about understanding customers, deciding what matters and creating meaningful outcomes. Nitor’s enterprise coaches Ida Sjöman and Erik Fallenius cover four themes that help product teams stay focused while the pace of delivery increases constantly.
The job of product leadership hasn’t changed because of AI. It is still about fighting fires without losing track of the long-term mission. What has changed is how fast everything around that job moves – and most of that change is hard to see while you are inside it.
Your competitors now move faster than your planning cycle. AI has shortened the gap between a great idea and a credible alternative, and teams can create working prototypes in days. Many organisations respond by speeding up implementation. The bigger value comes from using the same tools in discovery, helping teams learn faster and in greater depth before committing to a solution.
Small teams can now operate at a scale once reserved for much larger groups. AI can handle volume, but it does not ensure quality. Teams do not change their habits on their own, and knowledge silos form fast. Build learning, sharing and feedback loops into the work from the beginning. This keeps quality high as volume grows.
Discovery is faster, but keep it human
AI can read hundreds of support tickets in minutes. Feedback themes surface quickly, and qualitative and quantitative data can be linked with far less manual work. The cost and effort that used to make continuous discovery difficult have fallen. Discovery can become a habit your team practices every week.
Using AI in product discovery allows your team to:
Pitch with a working prototype instead of a slide deck.
Validate assumptions faster with data-rich mock-ups.
Turn transcripts and summaries into a living, central knowledge base for your team.
Lower the barrier for product people and designers to propose small improvements directly to the product.
Just do not let the speed of AI lure you out of the real conversations with real users. The insight that changes everything is often the one that isn’t in the tickets. You hear it only because of the relationship you have built – person to person.
The same applies internally. Be careful about replacing real alignment conversations with AI summaries or prompting. There is value in some of the friction. You align by arguing a problem out together and building on each other’s ideas.
Use AI to transcribe and summarise the routine so you can focus on the nuance.
Your roadmap has a new user profile
Most teams build products for a human who logs in, browses and clicks. The user triggers events and might even give Net Promoter Score (NPS) feedback. But the user base and their actions are changing.
Increasingly, AI agents may use your product, too. They make direct API calls and may never open the screens you designed your digital products for. AI agents skip the friction you have spent months smoothing out for people. Their metrics are different: task success rate, API reliability and human intervention rate.
A human-only roadmap risks missing a growing part of the product experience. What if you are building only half the product?
AI agents might already interact with your product. What matters is whether you have designed for that possibility:
Do you know how your API is actually being used, and by whom or what?
How do you differentiate from competitors when users no longer interact with your UI directly, but through an agent?
How do you keep the human interface worth returning to while making the programmatic experience just as reliable?
Designing for agents does not make the human experience less important. It means accepting that the user interface is no longer the only place where the product experience happens.
Prioritisation needs new criteria
The old question was: “Is this technically possible?”
Now, implementation can look like it is only a few prompts away. When more options appear feasible, teams may start building because they can without stopping to ask if they should. “Can we build this” shifts to “should we build this”.
AI cannot fix a bad idea. Do not mistake fast implementation or AI-generated feedback for validation that you are on the right track. There is always a risk that you are simply solving the wrong problem.
Reassess your criteria to ensure you stay on the right path:
Feasibility: can it work reliably beyond a convincing prototype?
Impact: which customer behaviour or business outcome should this change?
Risk: what could go wrong through hallucination, bias, drift or unwanted AI?
Cost: how will the unit economics change as usage grows? If every API call has a price tag, a successful feature going viral can quickly turn flat pricing into an unplanned expense. Consider pricing models wisely.
Nail these criteria before you talk about the backlog.
Trust is hard to build and easy to erode. When anyone can build products, the real advantage comes from understanding customers better than anyone else. Product teams need to make sure they are still learning from product usage rather than merely passing data through – and turning that learning into better decisions faster than the competition.
The real risk is losing your nerve
For fast-moving product leaders, missing a tool or skipping a process may be less dangerous than losing the nerve to trust their own judgement.
One part of the problem is volume. AI can always produce another signal, prototype or point to one more stakeholder who wants evidence. And suddenly, every decision starts to feel under-evidenced. You start second-guessing calls you would have made without hesitation two years ago. The dashboard shows you every bit of information available – except whether you are asking the right question.
Another part of the issue is AI’s tendency to affirm. If you use an AI that is designed to agree with you, it builds on your framing without really pushing back. This can easily leave you feeling right. Gradually, you stop taking the idea to peers who would have challenged it.
The best results come from people who use AI to move faster when the answers are clear. Then they slow down, ask the right questions and make the hard calls themselves.
Go fast where you can. Protect the rest.
How to get started already this week
Look back at the themes in this article: faster discovery, agents as users, new prioritisation criteria, and the pull to stop trusting your judgement. Which one is your biggest challenge right now and why?
Share it with your team and see which one they pick. If you answered differently, that gap is worth discussing before anything else.
Then pick one thing you could do differently this week. Keep it small enough to do without turning it into a whole project or strategy.
Written by
Erik Fallenius is a Senior Consultant at Nitor with twenty years in the field. He is interested in how AI is changing the way people work together, and he works with teams and their leaders to keep that work connected to customers and business goals. In his experience, the tools are the easy part; the new ways of working are not.
Ida Sjöman is a Senior Product Coach who helps leaders and teams make sense of AI and technology change by working with how people actually think, decide and adapt. Her approach blends strategy, product thinking and behavioural science into practical steps that stick.
