Validate one real problem before building
A practical synthesis of YC guidance on interviews and MVPs: replace assumptions with observable demand.
0101 / Narrow the audience and situation
Describe who struggles, when it happens and their current workaround. Choose a concrete task rather than a general-purpose AI platform. Speak to people who recently experienced the problem and distinguish encouragement from observable behavior.
0202 / Ask about a past experience
Explore the last occurrence: how it started, where it failed, the workaround and the time spent. Delay pitching your solution. Separate recorded facts from interpretation, and obtain consent before recording.
0303 / Define the smallest useful delivery
Compare repeated problems, existing alternatives and actions users are willing to take. Pick one pressing task and retain only the steps required to complete it. A small tool or assisted service can test the need before a full platform exists.
0404 / Observe use before investing more
Give interviewees something they can use. Observe task completion, return visits and obstacles. Set criteria for continuing, changing or stopping, then keep the results. Separate positive comments from actual usage and address the main obstacle next.
This is a validation method, not a success guarantee. A few friendly interviews do not establish market size, and AI cannot replace real customer feedback.
Sources & further reading
An original Qiyue synthesis of these sources. Suggested actions are not platform guarantees.