David BaszuckiFounder & CEO, Roblox

First entrepreneurial feeling came from a small idea: after Stanford a dorm friend suggested window cleaning. My brother and I knocked on doors and washed windows all summer. That was the first time I saw we could spin something up from nothing and make money.

After college I felt lost. School had a clear path; job hunting did not. I had two or three awful jobs and even called my dad about how bad it was. That disappointment later pushed me to stop forcing it and move intuitively toward work I cared about.

Time off after bad jobs helped. Apple II and Macintosh were rising, educational software was interesting. Instead of analyzing every option, I decided to enter educational software, following fascination with 3D simulation, physics, and consumer software. That step led directly to Knowledge Revolution and later Roblox.

Co-founders came from serendipity and openness. I wrote the Interactive Physics vision; MacUser reviewed it. Eric Cassel read it, called, flew from Cornell, and joined. I did not run big searches I shipped, had a vision, and stayed open to people drawn to it, a pattern that repeated.

Lesson: deep customer empathy is gold. With educational software we knew users well. Moving into mechanical engineering software was more tech-driven and less connected to our natural strengths. It worked, but taught me to think hard before walking away from a group you truly understand.

After selling Knowledge Revolution I needed a reset. We bought a 40-foot RV, towed a car, and drove across Canada and Minnesota with our three kids. I even hosted a short talk radio stint. It was about adventure and clearing my head before the next big bet.

I once tried to optimize life with a spreadsheet of nine careers and metrics. It was a bad way to choose. For big decisions, pure analysis can trap you; intuition about what fascinates you is better than a perfect grid. That lesson returned whenever I chose between safe options and harder, interesting ones.

DynaBlocks was a small 3D puzzle game with about 100 users that vanished in two weeks. We knew the real vision was multiplayer, cloud-based, with creation and play. We shut it down and spent 9–12 months rebuilding the broader platform, trusting the big vision despite the first failure.

When we launched user creation and self-publishing, with maybe 100–200 users, we saw constant building within hours. Four of us watched the viral pattern we knew from past bumps and recognized product–market fit. Later the full virtual economy triggered the same immediate reaction.

In 2007 users grew but dollars per user fell, revenue flat, costs rising. We tried 50 ideas; none fixed it. The answer was building a full virtual economy currency, creators selling, users buying, cash-outs, better discovery. Once launched, developers saw they could make a living within hours.

Minecraft and Fortnite exploded and it felt unfair. We leaned on our pipeline: auto-sharding in the cloud, multi-device support, stable APIs, one engine for mobile and PC. We stayed a unified platform instead of splitting engines or running a big first-party game, playing a long-term infrastructure game.

We choose safety. Parents, regulators, and we all want kids safe. Assume an 11-year-old just got a phone. Invest in age estimation, filtering, and safety tools even before laws demand it. There may be short-term engagement trade-offs, but long-term it's right and an opportunity.

We architected vertically, not just functionally. Roblox Safety is a dedicated group with product, engineering, and live ops inside it, with headcount you can measure. Adjacent groups touch safety, but having a pure safety team at the executive level makes it real, not just a slide.

I had to re-architect myself. Early on you can yell and push with raw energy; at scale that fails. I became more zen about feedback pausing before saying 'that will never scale' and delivering it helpfully. I also trust my gut more without over-clamping it with analysis.

We run hundreds of AI models for safety, search, discovery, and 3D generation. Waves: quiet behind-the-scenes improvements; generative tools letting creators and users build by talking; eventually avatars and assistants. Our advantage is billions of hours of 3D interaction we never sell the data but use it to make the platform smarter.

I am optimistic. Like the industrial revolution, work changes rather than disappearing. For creators, AI will raise quality and shift the landscape, not create mass unemployment. New work appears designing systems, stories, economies, communities that AI accelerates. The best developers will lean in and go further.

Best advice is simple: trust your gut. Many people told me that. My best decisions came from intuition I finally listened to, even when harder or riskier. I still look at data and analysis, but I need to say 'this feels right' and commit.