GrowthBridgeTRANSFORMATION
← All insights

Progress report

We're at Level 2 of 5. Here's what Level 3 actually requires.

A public read on where the system we run our own business on is. What moved last quarter, what's still hands-on, and the work it takes to climb the next rung.

Update, June 2026: we crossed. Read the Level 3 report →

A short film from inside the work, rendered by our own animation workflow, not a contractor.

Where we are

We're a family practice building the same system we'd build for a client. We publish where we are, honestly and on a fixed cadence, because in our category that's rare enough to be worth doing.

Here's the honest read: we're at Level 2 of 5.

The ladder isn't ours. It's adapted from Chris Cowden's framework for assessing how durable an operation's AI actually is. Five rungs:

  • Level 1 — AI-aware. The team knows the tools exist. Nothing durable runs yet.
  • Level 2 — AI-active. Pilot workflows running, supervised, hands-on. Some of it compounds.
  • Level 3 — AI-operational. Standardised. Semi-automated. Fewer surprises.
  • Level 4 — AI-integrated. Most jobs run on their own. Humans supervise the exceptions.
  • Level 5 — the destination. The system runs the operation. The humans run the strategy.

We map our own business across nine functions: acquisition, sales, delivery, customer success, internal ops, finance, founder operations, brand, research. Five of them operate at Level 2 today. Four of them (sales, delivery, customer success, and finance) are still climbing out of Level 1.

So when we say “Level 2,” we mean it the strict way. An operating system is only as strong as its weakest function. We don't get to average our way up. The honest number is the one held down by the work that isn't done yet.

What changed last quarter

A quarter ago, “Level 2” was more claim than system. Now there's machinery under it.

The work you're reading about was tracked end to end in a system we built called Helm. Every card, every handoff, every status change runs through it. This post was dispatched to a working session as a Helm card and written against its acceptance criteria. The thing reporting on the operating system is part of the operating system.

Sales and delivery got their first real scaffolding: the discovery motion, the offer mechanics, the per-tier onboarding steps, all captured as durable workflows instead of living in one person's head. The leads pipeline moved into the portal. The film above was rendered by our own animation workflow.

None of it is autonomous. Every run has a human in the loop. That's exactly what Level 2 is supposed to feel like: pilot workflows that work, that we still have to watch.

And not everything graduated. Some pilots we ran last quarter showed us where they broke, and then got shelved. That's the cost of being at Level 2 honestly: you try more than you keep.

What Level 3 requires

Level 3 isn't “more AI.” It's less supervision without less reliability, and that's a harder thing to earn.

The gap is mostly process. Our four lagging functions have to reach Level 2 first; you don't standardise a function you haven't yet made durable. From there, a workflow earns Level 3 when it runs the same way twice without us holding it. Fewer surprises. Less re-checking. The same output whether we're watching or not.

The rest is people. Level 3 asks the team to trust the system enough to stop watching it, and to know exactly when to step back in. That trust is earned by the system being right, repeatedly, in public. We're not there yet.

We're also not going to claim it before our weakest function gets there. From a number we could inflate. To a number we can stand behind.

What we're building right now

This quarter's named work:

  • Closing sales, delivery, customer success, and finance from Level 1 to Level 2: the four functions holding our aggregate down.
  • The content engine (the workflow publishing this post) so transparency runs on a cadence instead of on a good intention.
  • The cadence loop that re-checks what's changed across the whole system, flags what's newly unblocked, and surfaces stale judgment before it rots.

Some of this will land. Some of it will slip. When we publish the next one of these, we'll tell you which was which, by name.

Why we publish this

Most of our category sells Level 4 and operates at Level 2. We refuse to do that. The work of climbing from one rung to the next is the marketing. Not the polish, not the press release, not the inflated outcome claim.

Four refusals hold the line:

  • We won't sell you a level we haven't lived in.
  • We won't promise outcomes we can't ground in lived work.
  • We won't take more clients than we can fully deliver to.
  • We won't position done-with-you as something we run while you watch.

Immediate value at the entrance. AI-first at the destination. The bridge between is the work, and this is us building our own, in public, one rung at a time.

Read this in 90 days. We'll have moved. We'll show you exactly how.

— Vance and the GBT team

One email when the next one lands

We publish where we are every quarter.

No drip sequences, no upsells. Just one email when there's a real change to report.