Look for demand where people are already complaining

The Insight

Most validation advice tells you to go ask people. Run interviews, describe the idea, see if they say yes. The problem is that you are introducing the problem and the solution in the same breath, to someone being polite.

Eugene Cheah did not ask. He read.

Before committing to what became Featherless AI, his team watched communities where people were already stuck. Fine-tuners on Reddit and Discord were publishing models and then hitting a wall they described themselves, unprompted: they had built something nobody else could run.

The decision rule: the strongest demand signal is a complaint someone made before you arrived.

How They Did It

1. Go where the users already gather, not where you can reach them. Eugene names the specific communities: the local LLM crowd, LocalLlama, and Ollama, "who were constantly experimenting and fine tuning with different AI models for different use cases."

2. Look for the problem stated in their words. The recurring question Eugene saw was concrete. In his words: "we saw people asking, hey, on Reddit as finetuners created this model, how do I run this model then? How can I have access to it?"

3. Check the problem is structural, not preference. The constraint was physical. As Eugene puts it, "even if I can afford my $10,000 GPU, my friend cannot run the model because it's on a 10,000 GPU that's maybe inside my house."

4. Ship the cheapest version that answers the complaint. They already had the technology built for their own models. Pointing it at Llama and Mistral cost them a weekend. Eugene's framing of the decision: "we saw there was enough demand in Reddit and it was like there was no harm."

What Trips Up Founders

They ask leading questions. Eugene is honest that the market framing came after the fact: "framing it as we knew the market would be hungry for that actually gave it more credit than what was happening."

They wait for certainty before shipping. The experiment went out with the question still open. Eugene: "we weren't even sure or certain whether there was demand in the market for it. But that's why we did an experiment."

They treat community reading as research instead of a test. Watching the complaints told them where to point. Only revenue told them they were right.

When This Doesn't Work

If your buyers do not congregate anywhere public, there are no complaints to read. Enterprise buyers rarely post about procurement pain, and you are back to conversations.

If the complaint is loud but nobody has tried to solve it themselves, be careful. The signal here was strong because these people were already spending money and effort on workarounds.

The Question

Where are the people with your problem already talking about it, and what exact words do they use?

If you cannot answer that, you do not have a validation problem yet. You have a research problem, and it costs nothing to fix this week.