Most founders get addicted to the idea of being a founder. They read the books, listen to the podcasts, and raise the money. They get hooked on the dopamine of a TechCrunch mention or a feature on Product Hunt. They are playing a game called 'startup'. I am not interested in that game. I am interested in building things that solve brutally difficult problems.
Conventional wisdom tells you to focus. Pour every waking hour into one idea, one product, one company. Go an inch wide and a mile deep. This is sensible advice for most people. But it's also a creativity-killer and, in the age of agentic AI, a strategic limitation.
I am currently building seven separate AI products in parallel. This isn't a boast; it's a methodology. A portfolio approach to venture building is my defense against conformity and my primary engine for innovation. This is the playbook.
Most startup activity is performance art. It’s founders and their first few hires LARPing as a big company. They create elaborate onboarding processes, set up 15 different Slack channels, and spend weeks debating mission statements and company values. They hire a Head of People before they have a product that works.
This is corporate cosplay. It’s a waste of the only two resources that matter in the beginning: time and money. It makes you feel like you're running a real business, but you are merely creating bureaucracy. You are building a PowerPoint deck, not a product.
My contrarian take is this: your first ten hires should be builders. Engineers, product people, designers. People who make the thing. At Creative Marketing Group, my AI agency, I didn't hire a pure-play operations or sales role for the first two years. We were a team of practitioners obsessed with client results, and that obsession directly funded our ability to build products.
We run on a simple mantra: ship or shut up. We have a ruthless intolerance for meetings without a clear agenda and a decision-maker in the room. We don't have an HR department to mediate disputes; we hire adults who can communicate. We stripped out the noise to focus on the signal: building, shipping, and iterating.
Principle 2: The 'One Core Problem' Flywheel
Running seven products sounds like a recipe for schizophrenia. It would be, if they were all solving wildly different problems. They are not. Every single one of my ventures—from Autoemails to AutoSettle—is an attack on the same core problem.
My meta-problem is this: How can agentic AI automate complex, high-value business processes that have historically relied on expensive human expertise?
That's it. That's the obsession that drives me. The products are just different manifestations of that core question.
Autoemails tackles it in the domain of marketing. How do we move beyond basic automation and have AI agents manage entire customer lifecycle campaigns, from strategy to copywriting to deployment? AutoSettle applies it to law. How can an AI agent handle the negotiation of low-value personal injury claims, freeing up lawyers for more complex work? AutoTill looks at retail. How can agents manage inventory, predict demand, and automate supplier orders with minimal human oversight?