Journal / Product
Product management of AI products
Working with data science and ML teams over the years, it is easy to forget that AI is just a tool. A very powerful one, but still a tool. The product has to create value for end users. The technology serves that — not the other way around.
The temptation of the capability
AI teams are often structured around capabilities rather than problems. A team gets good at a particular model type, or builds expertise in a particular data pipeline, and then looks for applications of that capability. This is backwards.
Product thinking starts from the user's problem and works backwards to the technology. That applies as much to AI-powered products as to any other kind.
Where AI products differ from conventional products
There are a few genuine differences that product managers need to account for when working with AI:
Probabilistic outputs. Conventional software does what it is told. AI systems produce outputs that are probably right. That shifts how you handle errors, edge cases, and user trust. A 95% accuracy rate sounds good until you are the 5%.
Data dependency. AI products are only as good as the data they are trained on. This makes data quality and data governance product problems, not just engineering problems.
Explainability. In many contexts — healthcare, finance, legal — users and regulators need to understand why a system produced a given output. Black-box models that perform well may still be unusable if they cannot explain themselves.
Feedback loops. AI systems can be improved continuously as more data comes in. Product managers need to build mechanisms to capture that feedback and feed it back into the model — which adds a layer of complexity to post-launch operations.
What stays the same
The fundamentals do not change. You still need to understand the user. You still need to define what success looks like before you build. You still need a clear answer to why this product should exist and who benefits.
AI does not remove those requirements. If anything, it makes them more important — because the technical complexity of what sits underneath the product can easily obscure whether anyone actually needs it.
