KEY TAKEAWAYS
What to take away
- Distinguish investor interests from market demand.
- Check behavior and payment beyond demo reactions.
- Measure AI value across the full operating workflow.
Use overseas investment themes as a source of hypotheses
On November 8, 2026, Y Combinator’s Requests for Startups displays Winter 2027 interests including infrastructure and specialized software connecting AI to the real economy. These are the firm’s questions and investment interests, not global market statistics or proof that a particular business will succeed.
Our interpretation is to turn overseas themes into customer questions rather than assume demand in Japan. Who does the task, where do delays occur, and who is responsible when it fails? Understanding a workflow makes the testable problem clearer than focusing on a technology label.
Ask about recent behavior rather than expectations
YC’s user-research lecture emphasizes specific experiences rather than hypothetical approval. Our recommendation is to ask about a recent difficult task, current alternatives, time spent and decision makers, including an interview stage before showing a demo.
“That sounds useful” does not establish a budget or daily use. If buyers and users differ, investigate both. Separate participants’ statements from the researcher’s interpretation, and do not generalize a finding from a small set of conversations into a conclusion about the whole market.
Compare before and after in the customer’s terms
An attractive AI answer may not produce operational value. Our suggested test records time, checking, rework and operating burden before adoption and compares them afterward. Keep the population and period consistent and disclose exclusions and human assistance.
If a task is faster but verification and support increase, review the whole operation. Quality needs differ by customer, so do not declare PMF from a single satisfaction score or usage count. Consider tracking continued use, payment and expansion as observed behavior.
Let the evidence decide the next scope
Passing a raw request list to development can overrepresent the loudest customer. Our recommendation is to document the verified problem, desired change, test and deliberate exclusions together. Evaluate an iteration by the question answered rather than the number of features added.
When expanding from Japan, recheck differences in workflows, pricing, data handling and procurement. Overseas success stories and investment themes do not replace validation. These steps are general editorial recommendations, not a universal PMF threshold or a business-outcome guarantee.
SOURCES & VERIFICATION
Sources & verification
This editorial analysis references the primary sources below. Check the applicable program, jurisdiction and conditions; consult the relevant organization or professional for individual decisions.
- Y Combinator｜Requests for StartupsChecked: 8 Nov 2026
- Y Combinator｜Startup School Week 1 RecapChecked: 8 Nov 2026
