AMA: Google Group Product Manager, Android, Shahid Hussain on Practical AI for Product Managers

I don't agree with the implication in the question. AI as it stands today is a super helpful tool but it doesn't replace the core requirements for a good roadmap. We still need to define where we're going, how we're getting there, and prioritise. Define Product Vision (where we're trying to get to) which serves as the north star to ensure leadership alignment and focus. Determine Strategy (how we get there), specify the goals and how success will be measured. Prioritise Impact, by identifying th ...Read More

There's no one size fits all for evals -- we have to define these with respect to the problem we are trying to solve. Some of these will be critical, others may be not needed, but it's up to the PM (you!) to prioritise them. Quality / accuracy of the response: could be benchmarked vs a human response, vs a previous iteration of the tech, but beyond the numbers we have to think through how this would fit with our users. Is it enough? Latency & perceived responsiveness: Users have very low tol ...Read More

It's still early days for AI and the core tech is still advancing rapidly. Consider what we didn't have just a year ago (e.g. nano banana). Hence -- AI bets are often about buying the right to learn, vs traditional bets which are about predicting outcomes. AI investments are often bets on learning and capability, not guaranteed returns. Ask: what do we gain even if this specific feature flops? Data? Internal skills? User habits? Traditional features mostly risk "users don't want this." AI featur ...Read More

Across the industry we have seen many examples of AI integrations for the sake of using AI. But we as PMs need to focus on our users, our organisations and what will work best for them. AI remains an exciting leap forward with lots of implementation work to do -- but users, especially younger users, are pushing back against "ai slop" and we need to be very careful to ship AI when and where it makes sense. Does AI improve vs non-AI solution? The result should be meaningfully better, not just nove ...Read More

AI is a sea change for all of us. First: we all spent many years figuring out what works and what doesn't work for users, what's feasible and what isn't, and when the right time is to ship. AI breaks some but not all of the assumptions coming from your background as a PM. So you have to know when to trust your intuition and when to be open to being wrong. Many of the basics still apply: users still hate latency, trust still matters, adoption curves still work the same way. Others, you have to st ...Read More