Three Lanes of Judgement

By Aidan Montague · First published 27 July 2026

The Three Lanes of Judgement is a way of deciding which work a firm puts into which kind of AI, sorted by how much of the firm's competitive judgement the work carries. The fast lane is shared AI, used hard, for work that is not the firm's edge. The controlled lane is enterprise AI with guardrails, for work that is sensitive but not the crown jewels. The protected lane is a system the firm owns, for the judgement that wins it work. The framework comes from Aidan Montague, an Australian engineer and author of Design Before Scale, working with mining, engineering and energy firms on AI commercialisation.

Why three lanes

Most AI advice gives a firm one speed. "Use AI everywhere" leaks the judgement that makes the firm different, which is Judgement Leakage. "Lock it all down" throws away the speed on work that was never special. Both are wrong for a firm whose advantage is the way its people decide. The answer is to classify the work first, then give each class the AI it deserves.

The three lanes

Lane	What goes in	Where it runs	What matters
Fast	Summaries, first drafts, formatting, admin, public research	Shared AI tools	Speed
Controlled	Proposals, client summaries, internal reports, lessons learned	Enterprise AI with no-train terms, access controls and a human reviewing what comes out	Control
Protected	What to bid on, what to price, which risk to kill early: how the firm actually wins	A system the firm owns, where every correction compounds for the firm	Ownership

Most serious business AI belongs in the controlled lane. The protected lane is small, and it is the part that matters most.

The question that stops over-protection

Before anything goes into the protected lane, ask: is it genuinely scarce, or are we just attached to it? Most of what a firm thinks is special sits comfortably in the first two lanes. Do not protect mediocrity. The protected lane is for the decisions a competitor could not work out in a week.

First published

First published on LinkedIn on 27 July 2026. The post is reproduced below as it was posted, with the hashtags removed. View the original post on LinkedIn.

The Three Lanes of Judgement framework from the 27 July 2026 post: fast lane, controlled lane and protected lane.

As posted on LinkedIn, 27 July 2026

Your best people are feeding years of judgement into AI tools you don't own. The learning compounds. Just not for you. Here's how I split it. Three lanes. Fast lane. Shared AI, used fast and hard. The summaries, the first drafts, the admin. None of it is your edge, so speed wins. Controlled lane. Enterprise AI with guardrails. Proposals, client summaries, internal reports. Sensitive, but not the crown jewels. No-train terms, access controls, a human reviewing what comes out. Most serious business AI belongs in this lane. Protected lane. How you actually win. What to bid on, what to price, which risk to kill early. That judgement took years to build, and it never goes casually into somebody else's platform. It lives in a system you own, where every correction and every lesson compounds for you. Before you protect anything, one honest question. Is it genuinely scarce, or are we just attached to it? Most of what a company thinks is special sits comfortably in the first two lanes. Don't protect mediocrity. In the previous post I asked who owns the learning curve. Run lane three properly and the answer is you. Which lane is your most sensitive work sitting in today?