Deciding When a Product Needs AI
Judge whether an AI feature solves a real user problem before your team builds it
A taste of a lesson
We want AI to suggest prices to sellers on our marketplace. Is that a good use case?
It could be, but let us test it. Start with the problem: do sellers struggle to price, and what do they do now, maybe copy similar listings? Then the cost of error: a bad suggestion could lose a seller money or sales, and they may not notice. Also compare a simpler option, like showing the price range of recent similar sales, which is transparent and cheap. AI adds value if it handles messy listings better than that. What evidence do you have that sellers find pricing hard today?
Written by the teacher as an example. In your lesson the tutor answers your own questions, and like any AI it can be wrong.
What you will be able to do
- Frame an AI idea around a specific user problem and current workaround
- Judge whether a task suits a probabilistic system using clear criteria
- Map the cost of errors and whether users can spot and fix them
- Compare AI against simpler options such as rules, search or better defaults
- Write a one page decision memo with a recommendation and open questions
Lesson plan
- 1 From feature idea to user problem Restate an AI feature idea as a user problem with evidence behind it. Start
- 2 What AI is good and bad at Use clear criteria to judge whether a task suits an AI approach. Start
- 3 The cost of being wrong Map what happens to users and the business when the AI makes a mistake. Start
- 4 Simpler alternatives on the table Compare the AI idea with rules, search, templates, better UX and doing nothing. Start
- 5 Costs, data and practical limits Account for running costs, data access, latency and maintenance before deciding. Start
- 6 Cheap evidence and the decision memo Gather quick evidence and write a one page recommendation. Start
Try asking
About this tutor
For product managers, founders and designers who face pressure to add AI to their product and want a sound way to decide. You will learn to start from the user problem rather than the technology, test whether the task suits a probabilistic system, estimate the cost of being wrong, compare AI against simpler options like rules, search or better defaults, and consider running costs and data needs. Each lesson uses a real feature idea from your product or a worked example. You finish with a one page decision memo template that sets out the problem, the alternatives, the risks, the evidence you have and what you would need to learn before committing.
Reviews
4.7
3 ratingsSample
- Samuel B.Sample
Great at asking awkward questions. The concierge test idea saved us weeks of building something users did not want.
- Daniel K.Sample
I came in wanting to justify an AI search feature. After the alternatives lesson we shipped better filters first and parked the AI part. My memo got approved without the usual three meetings.
- Aiko N.Sample
The cost of error mapping is now a standard section in our product reviews. Slightly theoretical in places, but the tutor kept bringing it back to my actual feature.
About the teacher
Product management for AI features: deciding, specifying, testing and pricing them well
9 tutors 388 lessons taught Sample
I teach product managers and founders how to build AI features that people trust and keep using. I come from product work on software teams, where I learned that the hard part of an AI feature is rarely the model. It is deciding whether the feature should exist, writing down what good looks like, testing it before customers do and...
See Sanjana's profile and tutorsMore like this
Other tutors on the same or nearby topics.