---
title: "Small team, own loop"
url: "https://barbairians.com/blog/small-team-own-loop"
description: "Big companies lose to small ones not on brains but on loop speed: every decision waits on a committee and a vendor roadmap. A small team that owns its stack turns it in days."
---

# Small team, own loop

October 1, 2026·2 min read·BarbAIrianS

Every company has a loop: you notice a problem, decide what to do about it, ship a change, check the effect and correct. How fast that loop turns decides how much the company learns in a year. Not the budget. Not the headcount. The number of turns.

## Where the loop jams

In a large organisation every stage of the loop has someone in the way. Noticing the problem needs a report someone has to prepare. Deciding needs a meeting someone has to schedule. Shipping needs a change to a system that belongs to a vendor — so a ticket, a quote and a slot on someone else’s roadmap. Checking the effect needs another report.

Each of those stops makes sense on its own. Together they make the loop turn once a quarter. A company that turns it once a week learns a dozen times more in the same period.

## Why the small ones did not win before

A small company always decided faster. The owner sees the problem in the morning and decides before lunch. But then it hit the second half of the loop: shipping needed people it did not have and tools somebody else configured. A fast decision waited on slow execution.

That is changing from two sides at once. First, AI-native systems take over a large part of the execution work — retyping, sorting, reconciling, preparing. Second, when the process lives in your stack rather than inside a vendor’s product, changing a rule is a change you make, not a request you file.

## Anatomy of a fast loop

Teams that turn the loop in days share a few traits:

-   **The data is to hand.** “How many complaints of this kind did we get this month?” is answered in minutes, not a week, because the data sits in one place that is theirs.
-   **The decision is made by whoever sees the problem.** No escalation through three layers. A small team has a short path by definition — it just must not get built over with procedure.
-   **The change is small and reversible.** Not a process redesign — one rule, one prompt, one step. If it does not work, it is rolled back within the hour.
-   **The effect is measured automatically.** The same test set or the same metric as before the change. Without it, nobody knows whether the loop actually closed.

## Coordination cost is the real opponent

For decades scale was the answer to coordination cost: a large company could hire specialists and build processes a small one could not afford. But that same scale created coordination cost inside — meetings, sign-offs, hand-offs between departments.

AI does not remove that cost in a large company, because the problem there is not execution but the number of people who have to agree. In a small one it removes it almost entirely, because almost nobody has to agree. That asymmetry is what the whole thesis of this podcast stands on.

## A check question

How long was it from the last time someone in your company noticed a problem to the moment the fix was working? If the answer is counted in months, write down which stop the loop waited at longest. It will almost always be the one where the decision belonged to someone outside the company.

-   small team
-   speed
-   AI-native
-   organisation

## Keep reading

-   [AI factories in Europe: who will own the tokens](https://barbairians.com/blog/ai-factories-in-europe-who-owns-the-tokens)
-   [The model contract: what not to sign with an AI vendor](https://barbairians.com/blog/the-model-contract-what-not-to-sign-with-an-ai-vendor)