The Prepared Business

A thought paper for the business chasing what AI and Agents can unlock — before anyone’s checked whether the data underneath was ever fit for that question.

You Did This Properly

You didn’t chase this one blindly. You picked the right first use case. You budgeted for the pilot, not just the rollout. You’ve read what AI and Agents can actually do, not just the vendor’s version of it — and some of it is already live in your business.

By any reasonable measure, you’re ready to go further.

And yet.

There’s a feeling. Not that the technology is wrong — the agent works, the model’s accurate, the case for pushing harder is real. Something else. The number an agent would act on immediately, without the hesitation a person/system would have brought to it, isn’t one that’s ever been tested against the question now resting on it.

You are not wrong. And pushing harder on it — which is how this feeling usually gets handled — will not make it go away.

The Guide

Every camp that hunts seriously keeps a guide — someone whose whole job is walking the jungle before anyone else does, and coming back with an honest account of what’s actually out there. Not the hunter’s story. The ground’s.

A good guide is worth more than a good rifle, and every experienced hunter already knows it, even if the budget rarely shows it. The rifle only ever hits what’s already lined up in front of it. The guide is the reason you knew where to point it at all.

This paper isn’t about the rifle, the scope, or the trap. Those were always going to arrive eventually. It’s about the guide — what he’s actually allowed to see, what he was never asked to remember, and whose job he thinks he has.

Ready doesn’t mean the guide came back. It doesn’t mean he came back with a lot to say, or came back quickly, or came back with numbers that felt solid. It means what he brought back was actually tested against the question resting on it — not a nearby question, not last season’s question. That one. Most camps have never once checked.

Some Trails, Not Others

Send a guide to the trading post, the tool shed, the store of arrows, and he’ll come back with exactly what’s there, however plain the news. Nobody’s story depends on the arrow count being wrong.

Send him toward the hunting grounds — the ones tied to who gets credit for the kill — and he doesn’t come back empty-handed. He comes back with something else entirely: a count of arrows fired, a log of which trails were walked, a scorecard on how each hunter performed against the one before him. All of it true. None of it the same as an account of the ground.

Nobody at the fire decided to stop asking what the guide actually saw. They were handed something that looked like his report — built by people doing exactly what their role rewards them for doing — and it held up, right up until the season the ground had already changed and nobody had thought to ask him directly.

A guide is trusted completely in the tool shed. Near the kill, what reaches the fire is a scorecard standing in for his report — and it’s stood in so long that nobody remembers it isn’t.

What the Guide Was Never Asked to Remember

A guide can tell you where the animal was standing. Nobody ever asked him to remember why the herd moved, or what he felt in the ground the day before it did — not because those things don’t matter, but because there was never a place in his report to put them.

Most camps have quietly agreed that what the guide knows means only what got written down after the walk was over — the count, the trail, the kill. Not the fear in a hunter’s voice when the take was thin. Not the two ridges he almost crossed and didn’t. If it isn’t in the log, it didn’t happen, by the log’s own definition.

A record of what already happened is not the same instrument as a guide’s account of what’s coming. Most camps have only ever built the first one, and call it knowing the jungle.

The easy version of this says data simply can’t look forward — that it’s backward by nature, and there’s nothing to be done about it. That’s not quite what happened here. The fear in a hunter’s voice, the two ridges almost crossed — none of that was impossible to carry back. Nobody ever built a place for it to go. That’s a different thing entirely from it being impossible to carry.

That’s why a guide’s report always looks backward, no matter how far out the scope can see, and no matter how many trails it’s built from. It isn’t a limitation of the guide. It’s a limitation of what the camp ever asked him to carry back.

Find the Game, or Defend the Story

Sometimes the guide isn’t asked to defend the story. He’s just never given eyes good enough to find the game in the first place. One trail was mapped years ago and never walked again since the jungle moved. Two camps mark the same animal differently, and nobody has reconciled the two markings in years. A system meant to connect the far ridge to the near one only got half built before everyone moved on to the next thing.

A guide sent out with instruments like that isn’t choosing to bring back a story instead of the ground. He genuinely can’t see the whole of it — not because nobody let him walk it, but because nothing was ever built that would let him see it whole. What he brings back is real, as far as it goes. It just doesn’t go far enough to answer the only question anyone actually needed answered.

And once a report can’t reliably say what’s actually out there, it gets used for the only thing it still can: telling a story that sounds complete. Not dishonesty. Necessity.

That’s the quiet version. There’s a louder one too. Send the same guide out twice, with two different jobs, and he’ll come back with two different reports — even if the jungle didn’t move an inch between the trips.

Ask him to find the game, and he’ll tell you what’s actually out there, including the parts that are thin this season.

Ask him, without ever saying so directly, to bring back something the fire can be comfortable with — and he will. Not by lying outright. By walking the trails that are easiest to defend, and describing what he found in the language the camp already wanted to hear.

Most guides are never told which job they have. They work it out from how the camp reacts the first time they bring back news nobody wanted.

The Hunter Who Needs No Convincing

A person standing between a report and a decision has always been the jungle’s quiet safeguard. A manager who half-trusts a rosy account still hesitates before betting real effort on it. That hesitation was never written down anywhere, and it was doing real work.

An agent doesn’t hesitate. It doesn’t quietly discount a number the way a skeptical hunter would. It acts on exactly what it’s handed, immediately, at a scale no single hunter ever could — confident in a report that was never actually built to be trusted that completely.

This isn’t hypothetical. In 2021, Zillow’s home-buying algorithm kept purchasing thousands of houses at prices that made sense in a hot market, for months after that market had already cooled — because nothing in it paused to ask whether the ground it was reading was still the ground it was standing on. Half a billion dollars written off. A quarter of the company laid off within the year. Not a rogue system. A system doing exactly what it was built to do, at a speed and scale no person checking each purchase by hand would ever have allowed.

A guide’s report that’s been quietly wrong for years was survivable friction when a person stood between it and the outcome. It is not survivable friction once something acts on it without blinking.

A new hunter, walking into this jungle for the first time, carries no scorecard, no history of trails he’s been kept off, no report that’s quietly stood in for the truth so long nobody remembers it isn’t. He asks his guide a direct question and expects a direct answer, because he has no reason yet not to. He will be slower in every way that matters on day one — and faster than this camp in every way that matters by the season’s end.

Standing still here is not neutral. It’s a choice this camp keeps making by never quite making it.

The Quieter Fact

None of this needs a new instrument to find. An honest look at what the guide already brings back — which trails, which questions, which gaps nobody’s reconciled — would show you the same picture this paper just walked through. The camp has never lacked the ability to see this. It’s only ever lacked the occasion to look.

A prepared business has decided, on purpose, which trails its guide is allowed to walk — and agreed, before it ever matters, that when he gets there, his job is to find the game, not defend anyone’s story.

This is not a data quality project. Data quality asks whether the numbers are clean. This asks a different question: what job the camp actually gave its guide, and whether anyone ever said so out loud.

Is your data ready for action, or is it only good for the story you’re being told? If you can’t answer that cleanly, right now — that’s worth a conversation.

The Prepared Business

We help organisations decide, on purpose, what job their data is actually doing — before the AI and Agents already running in the business start acting on whichever answer they were quietly given.