13 signals you're inside a Supercompany

No one says "AI can't do that," to start.

Hi, it’s Greg and Taylor. 👋 Welcome to Supercompanies, our weekly newsletter about building the world’s most valuable companies.

We talk a lot about what supercompanies are in theory. This week I want to get specific about what they actually feel like from the inside - because if you’re building one, you should be able to look around and see/feel it..

These are the signals we see at Section and at the companies we work with that are furthest along. No company (yet) has every single signal (except for Anthropic and maybe a few startups). But if you’re reading this list and none of them sound familiar, your company likely isn’t on the path to becoming a supercompany.

- Greg

The 13 signals

1. AI optimism is the default. People complain about AI breaking, about outputs being wrong, about agents doing weird things. But nobody says “AI can’t ever do this.” The attitude is “let’s try it and see” (“trust but verify”).

2. Every knowledge task gets filtered through, “Could an agent do this?” Not occasionally - reflexively. When someone brings up a new process or a recurring task, the first question isn’t, “Who should own this?” It’s, “Can we automate this?”

3. Nobody stops at AI’s first output. First drafts - either from AI or humans - are usually sub-par. In a Supercompany, people iterate, wrangle, push through to something good, and there’s zero tolerance for shipping slop. If AI-generated garbage is making it to clients or stakeholders, you have too many AI passengers and not enough AI drivers.

4. Agents have names. “My chief of staff.” “The discovery scorer.” “Adrian.” They pop up organically - someone builds an agent, gives it a name, and other people start using it. When your team starts referring to agents with a shorthand, something is shifting.

5. There’s a new vocabulary. Orchestration, model council, .md files, inference budget, context layer. If you hear these words in the hallway or on Slack, you’re inside a company where AI is part of the operating system of work.

6. People share what they build. Skills, hacks, agents, and automations circulate - partly because AI is more powerful at scale, partly because people want to share the cool things they’ve built. The best builders don’t hoard - they publish to the team and help others adopt what they’ve made.

7. Documents are written for agents, not just humans. When your team starts formatting work so AI can consume it effectively, they’ve internalized that the audience for their work has changed. This means markdown files, structured data, and clear headers. And this isn’t just in written work - in Supercompanies, employees speak in meetings to agents. Everyone has Granola (or similar) running and is optimizing to their agents listening in.

8. Intolerance of friction rises. And so do expectations for quality and speed of output. Deadlines start to shrink - “how about an hour?” used to be a joke, but it’s becoming real.

9. The work ratio has flipped. Work used to be roughly 10% deciding what to make, 80% making it, 10% checking whether it was any good. Now it’s closer to 40/20/40. Human time is spent at the bookends: deciding which tasks are important and instructing AI on how to complete them, then judging the output and refining it with human taste. Start to ask your team member, “How did you allocate your time on this task or project?” and see if these allocations are starting to change.

10. Knowledge questions diminish and some disappear. “How do I do X?” doesn’t get answered by a person anymore. The company has wired up enough context that AI can handle a LOT of questions. There’s very little tolerance for people asking questions without asking AI first. If you ask a coworker over chat a question AI could have answered, start to expect some attitude. Ask AI, but still chat to your colleague about the World Cup or their weekend plans.

11. You’re overworked, not underworked. This surprises people. The most advanced AI users at Section are exhausted - not because they’re doing more manual work, but because they’re thinking more and reviewing more. The cognitive load has shifted from execution to judgment, and judgment is tiring in a way that execution isn’t - especially since agents don’t take a break. As CEO, I love it! But the exhilaration from working with smart agents soon becomes exhausting - and is unsustainable, so I expect our most AI-pilled employees to find their equilibrium.

12. You feel like a bottleneck and also more valuable. Machines work faster than you. You’re simultaneously the most important part of the process and the thing slowing it down. That tension is uncomfortable, but it’s correct - humans are necessary friction, because without them you get slop at scale.

13. You start and end your day talking to your AI. Your agents work overnight, so your morning starts with a briefing on what happened, what needs attention, and what’s ready for review. I start my day with an email from my AI Chief of Staff at 6am and end with the “5 p.m. summary” - which includes high fives for what I got done and reminders of what’s already on tomorrow’s to-do list.

If most of these sound familiar, you’re building a Supercompany. If most of them sound aspirational, you know what to work toward. And if some of them sound insane - well, they would have sounded insane to us two years ago too.

See you next week,

Greg and Taylor