My 4 big lessons after the first weeks of building my second 7-figure AI Startup
I thought I would share this in here Aslam, since it might help someone else who is building a product such as a software: Building my second AI startup has been humbling in a very specific way. For context: Iâve been part of building a 7-figure startup before. Not a âone-tweet, one-launchâ success, but years of real involvement. Product decisions. Customer conversations. Server Outages in the middle of a conference we were presenting our product at. Pressure. Messy execution. The whole thing. And yet, this second startup forced me to relearn fundamentals I thought I had already mastered. What surprised me most wasnât some new AI insight or clever growth hack. It was how valuable the boring stuff still is. The same principles Y Combinator has been hammering into founders for years about building products people actually want. Simple. Uncomfortable. And incredibly easy to dismiss once you think youâre âpast that stageâ. That assumption is exactly what gets you into trouble. Hereâs what building this startup drilled back into me. 1. Talk to your first customers. Early. Relentlessly. This sounds trivial until you notice how easily you avoid it. In my head, I had a clean narrative: problem, solution, ICP. On paper, it made sense. In reality, my first real users immediately poked holes in assumptions I didnât even realize I was making. The biggest mistake isnât building the wrong feature. Itâs building in isolation and calling it strategy. Every meaningful improvement we made came from a real conversation, not from thinking harder or refining slides. Talking to users didnât slow us down. It removed weeks of wasted work. 2. Iterate fast based on what they actually do, not what they say. Feedback is noisy. Behavior isnât. People are polite. Theyâll say âsounds coolâ and never come back. Theyâll praise features they never touch. If you listen only to words, youâll drift. The only feedback that mattered was: Did they use it again? Did they struggle at the same point? Did they ask the same question twice?