Most founders talk about listening to customers. Very few build systems around it.
When I sat down with Mike Kropp, former Microsoft and AWS leader, and now founder of Iridius - one idea kept resurfacing throughout our conversation: product-market fit is not something you discover once. It’s something you engineer through constant feedback loops.
Mike learned this at Microsoft while building the Patterns & Practices community, where his team worked directly with over 17,000 developers and architects in real time. They didn’t just ask customers for opinions. They shipped code weekly, exposed early SDKs, invited criticism, and let the community shape the product as it evolved.
“We called it customer-connected development,” Mike told me. “The better scale you have in getting requirements, the better product-market fit that you have.”
That line stuck with me because it reframes what great product teams actually do. They don’t build in isolation and hope customers validate later. They build with customers from the beginning.
And importantly, this only works if the company itself is built to move quickly.
Mike explained that one of the key enablers at Microsoft was adopting agile development early. You need the ability to iterate fast enough that customer feedback actually changes the product. Otherwise, “listening to customers” becomes little more than performative research.
What fascinated me was how much of this philosophy still shapes Mike’s thinking today at Iridius, where he’s building AI infrastructure for highly regulated industries.
The timing couldn’t be more relevant.
Right now, enterprises are rushing toward AI adoption, but many are stuck between experimentation and production. Mike believes the issue isn’t excitement - it’s trust.
Highly regulated organizations don’t just need AI that works. They need AI that is compliant, deterministic, explainable, and audit-ready.
“Their projects get all the way to MVP,” he explained, “but they feel it’s too risky to move them into production.”
That’s the gap Iridius is trying to solve.
The company is building what Mike describes as a compliance-by-design architecture for enterprise AI - turning regulations, policies, and standard operating procedures into executable logic that AI agents can understand and follow.
What struck me most wasn’t just the technology. It was how clearly Mike understood the emotional reality inside large organizations.
When auditors arrive, teams scramble for months trying to reconstruct evidence trails manually. Compliance departments are often viewed internally as bottlenecks slowing innovation down. Meanwhile, executives are under immense pressure to adopt AI without creating catastrophic risk.
The result is a market full of AI ambition but very little operational confidence.
And that’s where founder discipline becomes critical.
One of the most important parts of our conversation centered around restraint. Mike referenced Bob Muglia - another Microsoft legend I previously interviewed and highlighted something many founders struggle with: saying no to the wrong customers.
At Snowflake, Bob deliberately avoided certain enterprise customers early because he knew the product wasn’t ready yet. That level of discipline is incredibly difficult in startup environments where momentum feels existential.
But Mike believes long-term success depends on clarity.
“You need absolute clarity on who your customer is,” he said. “Know their pains, their hopes, their desires and obsess about making sure your product fits them.”
That discipline extends internally too.
Mike shared something I’ve increasingly noticed while scaling my own businesses: founders set the pace of the company. Not through motivational speeches, but through behavior.
In startups, there’s nowhere to hide. Everyone works. Everyone executes. And people either rise to that environment or self-select out.
As Mike put it:
“A’s hire A’s. B’s hire C’s and D’s.”
The best founders aren’t trying to be the smartest person in the room. They’re building teams capable of changing industries.
Toward the end of our conversation, we discussed AI hype. Mike was blunt: most AI startups won’t survive.
The reason? Many are building features that can easily be replicated as models improve. The companies that endure will be the ones building deep infrastructure, differentiated platforms, and solving mission-critical problems customers genuinely care about.
That may be the real lesson underneath everything we discussed.
The strongest companies aren’t built by chasing hype cycles.
They’re built by obsessing over customer truth long enough to create something that actually matters.
Until next time,
Firas Sozan
Your Cloud, Data & AI Search & Venture Partner
Find me on Linkedin: https://www.linkedin.com/in/firassozan/
Personal website: https://firassozan.com/
Company website: https://www.harrisonclarke.com/
Venture capital fund: https://harrisonclarkeventures.com/
‘Inside the Silicon Mind’ podcast: https://insidethesiliconmind.com/