The fourth variable
Day 10 of building in public
As I’m having breakfast with a few local VCs today, it clicks somewhere in the conversation. I could build that.
An AI-native GTM agency.
Take everything I’ve used to build high-growth sales teams and turn it into something a company can just buy, with a high-end service component on top. De-risk their growth. Stop making them throw headcount and internal resources at it. Same outputs, less weight.
[If you haven’t heard of AI native service companies, or what the cool kids are calling AINS, check out more here.]
So I started pulling everything I’ve ever built into one Claude project. Emails. Docs. Decks. Call recordings. Models. Eight years of go-to-market in one place, so I could frame out what this could even look like, because I’ve never built anything quite like it.
VCs like a16z and Emergence are backing firms built this way from the ground up (see here, here and here). The part I can’t stop thinking about is where these companies stand. McKinsey at one end. Stripe at the other.
That’s the question I keep noodling on. How do you build a best-in-class services business and a best-in-class product business at the same time?
I don’t have the answer yet. But the pull is real. The market already pays for software wrapped in services. Now you can run services that scale like software, and the customer gets a better product as a result.
Here’s the actual move, if you want to steal it. Put everything you’ve ever built into one project and tell the model exactly who it’s supposed to be. Mine looked like this:
“You are a third-year associate at an elite VC firm. Stanford and McKinsey before that. Build me an investment thesis for my investment committee: problem, hypothesis, product, founder. Then pressure-test it as any investor would. Why wouldn’t someone else do this? How does it break? What is the appropriate initial wedge? What is a reasonable TAM? SAM? Give me the anti-arguments against you. Ask me any questions you have before executing.”
The default setting on these tools is flattery. You have to make them argue against you.
The two questions I run every idea through now:
1/ is the work being outsourced today as high-volume, high-importance, low-judgment?
2/ is it already being outsourced through a SaaS product?
When both are yes, it’s worth exploring, but only if you have domain experience. These businesses depend on the founder’s specific knowledge and skills, making them especially intriguing for those who have been in the game for a while.
All of this makes me realize there’s a fourth variable I’m adding to my list of what I want next. It used to be three. Creative, building, local. Now I’m adding owning. I want the direction to be mine. I want to point at the thing and know I built it.
-Foley