Gokul Rajaram posted the other day that more people should think about becoming an IC. It really resonated with me, because I've been thinking about this a lot.

I love what I do at True. But if I ever went back into an operating role, I'd go back as an IC. Big tech, an AI lab or a startup, I'd want to be the one building.

In a way I’m already back to being the IC I've always been.

It started in my first comp sci class in high school, building games in Turbo Pascal. I sold websites all through high school and university. I joined Research in Motion while I was still in school. I cofounded BackType in 2008 as CTO. We were funded by YC, True and others.

After Twitter acquired us, I went up the management ladder. I started as the lead of a six-person team, still very close to the code, and ended up running a 3,000-person engineering org. Then I became an investor, and for years founders bugged me to start building again, because they knew that’s who I was. About a year ago I finally took the plunge, and I can't turn back.

This year I've built for work and for myself. I've worked in a small pod and built everything from infra to the app. A lot of it leans on what I know well: data, ML and distributed systems. Some of it is stuff I'd barely touched, like voice. A few weeks ago I built a Mac app for our team on my flight home from our annual offsite. The line, and the excuse, between "my area" and "not my area" is mostly gone.

I'm lucky that building is part of my job at True, where we’re rethinking how we work from first principles instead of bolting AI onto old tools. It's how I see how fast the world is changing and which markets are exploding. It makes me better at sourcing and diligence. And it makes me more useful to founders, because I'm running into the same walls they are.

What surprised me is what actually makes me fast. It isn't the code. It's taste: decades of product and UX, and knowing what good feels like. And it’s fundamentals from years of building and writing hundreds of thousands of lines of code, which let me see around corners and point a lot of AI at the right problem before I end up chasing weird bugs. That's what lets me decide quickly, build with conviction and throw things away when they're wrong.

I've managed at every level, so I don't say this lightly. I don't think of management as a bad skill. I think of it as a skill that's losing value. Hiring, mentoring and leading still matter, and those years taught me a lot about all three. But today you can do all of that as an IC, and a lot of it now goes into directing agents. Meanwhile, big companies are trapping a lot of managers in jobs that are mostly coordination. The people who can make the thing are pulling ahead.

Two takeaways.

Founders: look at managers, directors and VPs as key IC hires. Some of the best builders out there are sitting in management roles. They have years of judgment and are waiting for someone to let them build again. Let them. A few founders I work with are hiring this way, and I think it's clever.

Managers, directors, VPs: think about becoming an IC, at your company or somewhere else. Your experience doesn't go away. It becomes your edge.

Go build.