Built AI-first · Chapter 7 of 9
Tool-chasing is not a discipline problem. It is what happens when an organisation has no written rule for deciding, so every new product arrives as a fresh argument with no precedent and the loudest recent demo wins.
Four tests, three verdicts, and one question asked six months early. The aim is not to make the perfect choice — it is to make a choice you can still defend in the autumn.
The four tests
Answer all four in writing. Most decisions resolve themselves by the third, and the ones that do not are the ones worth a meeting.
| Test | What you are actually asking |
|---|---|
| 1. Is this where our judgement lives? | Does the thing encode decisions that are specifically ours — how we price, how we scope, what we refuse, what we know about this client — or is it a capability any competent organisation in our trade would recognise? |
| 2. Whose governance is better, theirs or ours? | For identity, audit, retention and revocation: honestly, who does it better? A small organisation buying a governed product is frequently buying a security posture it could not build, and that is a legitimate reason to buy. |
| 3. What does being wrong cost, and how do we get out? | Contract length, data export, and how much of the organisation's working practice would have to be unpicked. Reversibility is a feature and it is rarely on the comparison sheet. |
| 4. Is the deciding input even available yet? | Is there a number, a measurement or a fact this choice depends on that nobody currently has? If so, the honest verdict is the third one, and it has a name. |
The verdicts
Verdict one
When the capability is a commodity in your trade, the vendor's governance beats yours, and getting out is cheap.
The trap: buying a thing that quietly becomes where your judgement lives. A product that starts as a tool and ends as the only place your pricing logic is written has changed category without telling you.
Verdict two
When the thing encodes judgement nobody sells — and only then.
The trap: building a governed product that already exists in your tenant, because building it is more interesting. The maintenance cost is real, it lands on the same small team forever, and it is usually the reason the interesting build is abandoned in year two.
Verdict three
When the input that decides this does not exist yet, or the commitment outlives the certainty.
The trap: treating wait as the absence of a decision. A wait without an owner and a review date is not a wait — it is drift wearing a decision's clothes, and it is how organisations arrive at next year having neither bought, built, nor learned anything.
The question
What would have to be true in six months for this to still be the right call — and would we actually find out?
The second half of that sentence is the one that does the work. Most bad technology decisions were defensible when made and became indefensible without anybody noticing, because nothing in the organisation was watching the assumption they rested on.
So the output of a buy, build or wait decision is not just a verdict. It is a verdict, the assumption it rests on, and the thing that will tell you when the assumption breaks. If you cannot name the third, you have not finished deciding.
We apply this to our own product decisions and it has changed several of them — most usefully by turning a "build" into a "measure first, then decide", which is the same ordering rule that runs through chapter nine.
In your world
Offered as a starting position rather than advice — your four test answers beat anything generic on this page.
Records and reporting systems are commodity and heavily governed — buy them. What is genuinely yours is the set of standing obligations and who is owed what by when, and no vendor knows your district's version of that.
Procurement cycles are long and public, which makes reversibility harder and the wait verdict more valuable. A documented wait with a review date is also far easier to defend at a council meeting than a hurried purchase.
Scheduling and dispatch software is a mature market and you will not beat it. How you scope, price and refuse work is the actual business, it is currently in the owner's head, and it is the only thing worth building.
Practice management is commodity. Your method — how you run an engagement, what you check before it leaves the building — is the asset, and writing it down is what makes the practice worth more than its billable hours.
What we will not claim
Where this sits
Having decided what to buy and what to build, the remaining question is how much authority any of it gets — and that one is answered per act, in an order that is not a matter of taste.