For product managers, one of the hardest things to do is to say no, especially to everyone at the same time when new, innovative technologies come around. But just because something is new, disruptive and your organisation is willing and capable to introduce it into your product, doesn't mean you should.
Early in my career at Microsoft, I worked in Developer Experience — helping academia, startups, and big enterprises adopt new tech. Toward the end of that era, I spent time in what we then called Cognitive Services and chatbots — basically Microsoft AI today. It was 2016, long before ChatGPT, and every organisation wanted a chatbot.
My job was to help them figure out if they actually needed one. Most didn’t. Leadership wanted to be innovative, but the customer problem to be solved and the solution to solve it was usually simpler. Often, a well-designed CX flow or better customer success training solved the same problem better.
That lesson stuck with me: just because the technology is impressive doesn’t mean it’s the right answer.
Fast-forward to today, and I feel like we’ve come full circle. It's been a few years since ChatGPT made LLMs and Generative AI a household name. Everyone is talking about AI, and it’s genuinely powerful. But the same question remains — why are we using it? I always ask myself with any new technology:
- What is the customer problem we're solving?
- How can we measure whether it’s actually helping?
- What is the best way to solve the customer problem with the best ROI?
AI features often look exciting and inexpensive to build, but the true ROI comes from sustained value over time — not just shipping something new. If a simpler, cheaper, more reliable fix exists, don’t be afraid to pick it and advocate for it. Remember, our role is to solve the customer's problem — they don't care how, they just want it solved.
In large organisations, it’s easy to get caught up in what’s technically possible and forget to ask what’s strategically right.
Before deciding where to apply AI, step back and think about the bigger picture:
- What markets do we actually want to serve or expand into?
- How do we expect our customers’ needs to evolve over the next five years?
- What capabilities — human or technical — do we want to own versus partner for?
AI, especially as it becomes more “agentic” — systems that act and decide with less human input — will tempt us to automate faster than we can think. But strategy is about pacing as much as progress. Just because something can be agentic doesn’t mean it should be.
The most effective AI strategies I’ve seen are grounded in long-term clarity: clear outcomes, clear customer value, and a clear sense of where human judgment still matters.
When your AI roadmap ladders up to your business strategy — markets, customers, brand — it scales naturally. When it doesn’t, you end up with clever demos and confused teams.
AI is our generation’s industrial revolution. It will reshape work and creativity in ways we’re only beginning to understand. But just like then, not everything needs to change. The best products will still be built by people who listen, care, and choose the right tool for the job — even if that tool is a 20-year-old piece of tech that still works beautifully.
So before you add “AI-powered” to your next roadmap item, ask yourself: does this make life better for the customer, or just newer?
Let me pose a non-techie question — impact drivers have changed construction and DIY and made it infinitely more productive, but every builder still has their trusty hammer. Why is that?
