Real AI PM work usually starts from a product problem, not an architecture. Start from the problem to see which systems it needs — or start from the system you already know you're building.
Product scenarios: the problem, the system it takes, and the decisions it raises. One scenario draws on several kits.
Recommendations & Personalization. What should we show this user next? Candidate generation, ranking, constraints, feedback, cold start, exploration, bias, experiments, and feedback loops — each framed as the decision you have to make.
Open scenario →OutlineSearch & Discovery. How do we get people to the right thing when they only half know what they want?
Open scenario →OutlineDecisioning / Next-Best-Action. Which action — offer, message, route, or nothing — is best for this customer right now?
Open scenario →OutlineWorkflow Automation. Which steps of this process can AI run — and where must a person stay in the loop?
Open scenario →OutlineCustomer Support. How do we resolve customer problems with AI without deflecting people who need a human?
Open scenario →System kits: a reusable workflow for building and shipping one type of AI system, from first decision to production.
Answers grounded in your own documents. Confident answers that aren't in the source — or are in the source but don't answer the question.
Open kit →Tool-using agents and chained AI workflows. Every step can fail, and failures compound: five steps at 90% complete only 59% of tasks.
Open kit →Drafts text or edits a person reviews before using. Suggestions users accept and then quietly rewrite — usage looks healthy while time saved is zero.
Open kit →Talks to customers directly, often with no human in the loop. Deflecting customers instead of resolving their problem.
Open kit →Sorts, flags, or routes items — fraud, tickets, content. A threshold picked once and never revisited — wrong for what a false alarm and a miss actually cost you.
Open kit →Generates, completes, or reviews code. Code that passes a benchmark but fails in your codebase.
Open kit →A scenario is the product problem and its decision framework. A kit is a reusable way to build one kind of system. Recommendations and search both lean on retrieval; workflow automation and decisioning both lean on classification.