Problem Interviews Without Methodology (AI SaaS)

AI SaaS methodology is heavy. Eval frameworks, prompt registries, persona docs, customer development canvases. If that overwhelms you, here is the version stripped to the parts that actually do the work. Three rules.

Rule 1: Talk to People Doing the Workflow

Not your friends. Not AI Twitter. Not VC partners. People who, recently, did the specific workflow you want to AI-augment.

Find them on LinkedIn by job title and team size. Cold message ten of them. Some will say yes. The mechanism is just asking.

Rule 2: Ask About Past Incidents, Not Future Hopes

Do not ask "would you use an AI tool that does X." Hopes are unreliable. Ask "walk me through the last time you did X." Past incidents are reliable. The story includes the workflow, the pain points, the existing tools, the moment things broke.

If AI shows up in their answer, follow it. If not, that is also data. Either way, do not import the AI frame yourself.

Rule 3: Do Not Pitch and Do Not Demo

Two failure modes wrapped into one rule. Do not describe what you are building. Do not show your prototype. The moment you do either, the conversation tilts toward politeness and away from honesty.

If a model or architecture idea hits you mid-call, write it in private notes. The customer should never see your hypothesis during the call.

What That Adds Up To

Three rules. Talk to people doing the workflow. Ask about specific past incidents. Do not pitch and do not demo.

Run ten interviews following those three rules and you will have more useful information about your AI SaaS market than most first-time founders ever collect.

What You Can Skip

You can skip the eval framework. You can skip the prompt versioning system. You can skip the synthetic test set. You can skip the latency dashboard. You can skip the cost-per-token tracker. You can skip the persona doc.

None are bad in principle. They are infrastructure for AI products in front of thousands of users. You have ten interviewees.

What You Cannot Skip

You cannot skip the calls. The reason eval frameworks exist is to verify AI behavior at scale, after you know what behavior is even worth verifying. The interview round tells you what to verify.

You also cannot skip writing things down. Direct quotes, the trigger event, the workaround, the trust boundary - all non-negotiable.

If You Want to Be Even Lazier

Do five calls. Just five. Use the three rules. Take messy notes. Write a one-paragraph summary at the end of each.

You will know more after five than you knew before. The biggest mistake is doing zero because the methodology felt overwhelming, not doing too few because you got tired.

Plain Summary

Find people doing the workflow. Ask about specific past incidents. Do not sell. Do not demo. Take notes. Look for repeats. Stop when you can summarize the customer in three sentences. Then ship the smallest AI capability that respects their trust boundary.

That is problem interviews for AI SaaS without complexity. The rest is decoration.