AI Product Teardown: Strengths, Trust Gaps and One Prioritized Improvement

Pick an AI-powered app and critique it: what it does well, where it falls short, and what you would change first. ```hint Judge the failure cases With an AI product, much of the critique lives in what happens when the output is wrong. Ask whether a user can tell, what it costs them, and how the product helps them recover. ``` ### Clarifying Questions - May I choose any AI app, or would you like me to critique Perplexity's own product or a direct competitor? - Should I judge it from the view of a typical everyday user, or a specific segment such as students, researchers or people using it for work? - Do you want a balanced review, or should I land on one prioritized change and how I would measure it? - Should the AI output itself (accuracy, speed) get most of the time, or the surrounding experience such as onboarding, pricing and reasons to come back? ### What a Strong Answer Covers - A clear choice of app, its target user, the job it does for them, and how it makes money, set out before any judgment. - Strengths and weaknesses specific to AI products: answer quality, trust and verifiability, latency, and how errors and uncertainty are handled. - Observations tied to concrete moments in the user journey rather than general opinions. - One prioritized improvement with a reason it ranks first, and metrics, including guardrails, to judge it. - The cost, latency and trust risks of the proposed change. ### Follow-up Questions - If you could ship only one change next quarter, which would it be and which number would tell you it worked? - How would you find out whether users trust the answers more after your change, rather than simply using the app more? - A competitor ships your proposed feature next month. Does your priority change?