Why a Roadmap should sometimes say: don't buy AI
If a paid diagnostic can only ever conclude 'hire us,' it isn't a diagnostic. Here's the rule we've written into ours.
Imagine a 9-person specialty retailer convinced they need an AI assistant for customer service. Budget approved, enthusiasm high. Now imagine their ticket volume is eleven emails a day. Eleven. A part-timer handles them in under an hour, well, with a personal touch their customers love.
Automating that would save almost nothing and risk the exact thing that makes the shop good. Meanwhile the thing actually hurting them is inventory reconciliation — three afternoons a month of two people counting and arguing with a spreadsheet — and odds are a $40/month off-the-shelf tool fixes most of it. The right deliverable in that engagement is: point at the tool, show the math, and end there. No build.
Why design the business this way? Because the Roadmap's value rests entirely on the verdict being honest. If every Roadmap magically concluded 'you should pay us $20,000,' it wouldn't be a diagnostic. It would be a sales pitch with a two-week delay.
So the rule is written into the deliverable itself: every opportunity gets costed, including the ones where the answer is 'don't automate this' or 'buy something cheap off the shelf.' If AI isn't the answer for your business, the Roadmap says so — in writing, with numbers.
A meaningful share of Roadmaps should end exactly that way. The ones that end in 'don't build' are the reason anyone should trust the ones that end in 'build.'
About these notes
Scenarios in these notes are worked examples — the patterns we build for, with the math shown. Wondering what your version looks like: sam@outeragent.ai.