Key Takeaways

  • Michael Grinich, founder and CEO of WorkOS, argues that calling yourself a non-technical founder signals low personal agency when modern AI tooling eliminates syntax as a barrier.
  • WorkOS built and deployed Atlas, an internal AI teammate, illustrating how small teams can assemble functioning software without waiting on engineering departments.
  • Grinich projects the global developer pool will surge from roughly 30 million professional software engineers to 300 million and eventually 3 billion people as natural language replaces traditional programming languages.
  • Building code is no longer the primary bottleneck in early company creation; distribution, sales pipelines, and financial discipline are much harder to master.

The Technical Excuse Has Expired

For two decades, aspiring startup founders hid behind a familiar excuse: they had great business intuition, but they could not code. They hunted for technical co-founders on message boards, gave away half their equity to freelance agencies, or waited on the sidelines. Grinich says that entire mindset has collapsed.

“So, I think the era of being nontechnical is over,” (play from 1:04:37) Grinich told the hosts. “Saying I'm a nontechnical co-founder is cope. It's like you can just go learn it. It's an indication of low agency.” (play from 1:04:41)

His point is direct. When natural language lets you instruct machines to generate working applications, claiming you cannot build software means you simply chose not to spend an afternoon learning how to direct an LLM. At WorkOS, Grinich watched this shift happen firsthand while building Atlas, an AI teammate deployed inside the company. The tools to build, test, and iterate are available to anyone with an internet connection. If a founder throws up their hands at software creation now, they reveal how they handle every future roadblock.

As Grinich put it: “If learning the code is going to be the thing that stops you from being successful, I hate to break it to you, but you're not going to succeed long term, cuz that's like not the hardest thing about building the business.” (play from 1:05:22)

Three Billion People Will Write Software

Software development historically resembled monastic scribing. A tiny sliver of humanity learned esoteric syntax, compiler quirks, and memory management. That scarcity created massive salaries and gatekept who could build digital businesses.

That dynamic is breaking. Grinich points out that the global constraint on software has never been customer demand; it was always the labor pool required to produce it.

“Like we just can make software forever, and the constraint has always been people making it,” (play from 1:07:02) Grinich said. “And now that constraint is lifted, so I think that the software industry is about to go through this explosion of usage and creativity and we'll go from you know, they're being roughly 30 million professional software engineers in the world to at least 10x that in terms of people doing stuff, probably 100x you know, in our lifetimes.” (play from 1:07:10)

When 300 million to 3 billion people can prompt computers to build software, the value of knowing where to place semicolons drops toward zero. The Changelog hosts captured the core shift in perspective: “MG's point that for the first time we can bring computers closer to people instead of dragging people closer to computers.” (play from 1:11:20)

When writing software becomes as simple as speaking, the real work begins. Product-market fit is harder to secure because supply will drown demand. The founders who win will not be the ones who marvel at their ability to ship an app; they will be the operators who master enterprise sales, design working distribution channels, and secure durable unit economics.

What to Do With This

Kill the phrase non-technical from your pitch deck and your vocabulary. Open an AI coding environment this evening and build a functioning prototype of your product's core loop before Friday. If you encounter an error, paste the stack trace directly back into the model and ask it to correct the mistake instead of calling an engineer.