A pod used to be three people. Now it's one.

Hey, it's Eric.

A pod used to be three people: an engineer, a designer, a product manager. Now I watch one person cover all three. Give a strong engineer a few design and product skills plus a stack of AI agents, and boom the pod is now one person. We see the same thing at Single Grain, where one person handling AEO does work that used to take a small team.

We started calling it the pod of one, and we run a six-week apprenticeship to build them.

One thing surprised me. I tried it on a single person on my team and we met around thirteen times. I expected the hard part to be the AI mechanics: the tools, the prompting, the setup. That wasn't it. What moved him most was the tone I set, the expectations, the speed we work at, what good looks like. Once that landed, he took it and ran. He's flying now, and most of that is him. I pointed, he sprinted.

So when you build a pod of one, put your energy into the standard you set. That's the part people rise to.

Two notes on rolling it out without chaos.

Start with a small test group, not the whole company. A big-bang rollout creates anxiety and confusion. A handful of people for six weeks proves the model and keeps the nerves down.

Then use that first group to teach the next one. The people you train in cohort one become the coaches for cohort two, which is what makes this spread instead of staying stuck on you. It compounds.

Do this and you stop scaling output by adding headcount. One trained person with agents does the work of several, and a team of them is a different kind of company.

Want the full walkthrough of how we run it? I broke it down here: watch it here.

To building people who run on their own,

Eric Siu