911±¬ÁÏÍø

Before taking From Zero to Product, what assumptions did you have about startups or building a new product? How did they change?

I will preface everything by saying I believe in Finland's 100x100, and I will be building globally from day 1 while based out of the Nordics/EU to be a part of the 100.

After Aalto Founder Sprint and heading into this course, I assumed staying disciplined with your validation was the most integral part of the process. Of identifying pains and problems to solve.

But depending on the type of company you are aiming to build, I believe you can make significantly faster progress by just trying to sell from day 1 instead.

Asking open-ended questions and identifying themes or patterns through user interviews is great, and I don't mean to suggest it is the wrong approach (nor that I've "cracked the code" or whatever, either), but honestly a lot of the time people don't even know about their own problems, or can't verbalize them well, at least when put on the spot.

By the time you talk to a sample size large enough to start identifying patterns, you could have already tried to sell your (non-existent) product and received immediate, clear validation or invalidation of your hypotheses.

Tell us about the idea you worked on. What was the most important insight you uncovered through customer validation?

The hypothesis I wanted to test was that "shadow AI" in highly regulated industries is an increasingly big problem. Shadow AI is essentially uploading private, confidential, or sensitive data into a public LLM or unapproved tool.

My core assumption was that AI and LLM adoption is inevitable, but these particular industries—healthcare, legal, and others—will remain underserved, creating an increasing risk of shadow AI-related incidents.

Honestly, nothing I discovered through my validation work led me to believe my hypotheses or assumptions were incorrect. However, I already had some conception of a potential solution, and I eventually realized the willingness to pay for that solution was not high enough. Not because the pain was not real, but simply because it was not intense enough for the average end user to care.

The typical employee does not feel the risks, dangers, or effects of shadow AI, but they see increasingly huge productivity benefits. Why champion a solution to this problem and wait through a long sales and regulation cycle when ChatGPT, Gemini, or Claude are one prompt away?

That was probably the most important insight for me. Again, going back to my previous point, if I had just tried to sell this personally identifiable information sanitization layer from the start, I would have come to the same conclusion much faster.

Was there a moment when the evidence challenged something you strongly believed?

When speaking with those whom the risks of shadow AI directly affect—heads of compliance, IT, and security—I assumed a solution to the problem would be very attractive to them.

I misjudged this and found out that they want to minimize exposure risk above all else, even at the expense of potentially huge productivity gains.

So even with a potential solution that would seemingly solve both, if it introduced any potential new risk—or frankly, additional work for them, since they would need to champion and own it internally—most would opt to do nothing.

I think the reality with generative AI right now is that a huge feeling of FOMO is driving adoption. But for many industries, especially those that have historically adopted technology slowly, the return on investment is not immediately obvious.

After all is said and done, I believe this idea and space still have immense potential. But I also believe I am not the right profile or person to execute on it.

What was the most uncomfortable or difficult part of the validation process, and what did you learn from it?

The most uncomfortable part is always gauging the emotional reactions of the people you speak with.

Especially when it seems they have not given much thought or concern to the subject at hand, which often makes you wonder whether the direction you're pursuing is completely wrong or irrelevant—or whether you're simply speaking with the wrong personas.

Narrowing signals and making assumptions based on incomplete information is also uncomfortable. But doing this quickly helps you discover hard truths faster and minimize wasted time.

Another challenge is realizing that you may not be the right person for the problem or space. That is completely normal, but it becomes more difficult to accept the longer you work on something and the more emotionally attached you become.

Looking back, what is one lesson from the course that you think every aspiring founder should learn early?

If the customer is not willing to pay even for an unrefined proof of concept, there is a good chance it is not solving an intense enough pain.

And for those from technical backgrounds such as myself: fall in love with solving a problem instead of the technology or a particular solution.

This may not apply as heavily to deep-tech or certain other domains, but the broader point still stands. It is a lesson that only becomes more difficult the longer you spend in the industry.