With my first book AI Reality out there on Amazon, I’m now writing the sequel. this time the voice is old Marek. As a teaser here’s the Preface and Chapter 1.

Part 1

This story is fiction. But also reality to many of us.

---

Preface

I get a different conversation now.

Three years ago it was — so what do we actually do? Now it’s — we did it. It worked. What next?

That’s a good problem to have. It’s also a harder one to answer.

FoodCo got the foundations right. A strong data platform. The skills gaps closed. Models running quietly in the background — forecasting, scheduling, quality control, preventative maintenance — each one doing its job without anyone needing to present excuses to the board. One focused, successful deployment became many. The flywheel starts turning.

This story is about what happens next.

This story is about growing the business without growing the team.

That’s the question now. Not how do we start — how do we scale. Integrate the models so they work together, not apart. Cross business, business-to-business. Build the intelligence into the data pipelines themselves, so the decisions don’t wait for a person to make them.

I think of it as a lineage, not a leap. EDI moved structured data between agreed parties, decades ago, and it was genuinely transformative — it took the human out of the transfer. Event-driven architecture moved events between systems in real time, but the responses were still pre-programmed; the system reacted, it didn’t reason. Agentic AI is the grandchild. It doesn’t just move data, and it doesn’t just react. It reasons across context, builds a plan suited to the situation in front of it, and acts — adapting when the outcome isn’t what it expected.

The ancestors built the pipes. This generation adds the brain. And the frontier — the part almost nobody is talking about yet — is B2B. Not agents inside one business making that business faster. Agents across businesses, reasoning locally and deciding collectively, a demand signal propagating all the way up a supply chain without a person ever picking up a phone to place the order.

That’s the world this book lives in. Three factories, run as one, through a single control plane. It sounds like science fiction. It isn’t. It’s the next boring thing, waiting to be built properly, by people willing to do the unglamorous work first — because the tool was never the hard part. The foundations are. Trustworthy data. Tools with clear contracts. Governance that says where autonomy is allowed and where a human stays in the loop. Someone who owns the outcome. Skip any one of those and you haven’t got an agent — you’ve got a very fast way to automate a bad decision.

This story picks up three years after AI Reality, at the same company, with most of the same people. But this time it isn’t told through Max’s eyes.

It’s told through mine.

Not literally — Marek is fiction, same as Max was. But he’s the part of me that’s spent thirty years watching organisations get this right or wrong in roughly equal measure. Every semi-retired consultant who gets one more phone call from an old friend and says yes will recognise something in him. I know I did, in writing him.

The company under threat in this book, the board pressure to sell rather than build, the vendor circling with the right buzzwords and the wrong understanding — these are composites too, the same way FoodCo’s twelve failed projects were in the first book. I’ve sat in those boardrooms. I’ve watched good businesses get sold for a quick number instead of built into a bigger one, because nobody in the room could make the case for what came next. This book is that case.

One last thing, same as before. This book has an argument, and it doesn’t hide it.

You don’t grow a business by hiring your way through every new decision. You grow it by building systems that can make good decisions on your behalf, at the boundaries where humans used to be the bottleneck. Agentic or die. Not as a threat — as a fact. The businesses that figure out how to act, not just analyse, are the ones still standing in five years.

The ancestors built the pipes. This book is about building the trust layer.

That’s AI Reality, one more time. That’s Agentic or Die.

Chapter 1 – Purpose

Why can’t I sleep, another night coughing. Am I sick? I stand in front of the bathroom mirror.

“Looking your age today old boy,” I say out loud. But there’s no one to hear. Living alone has its ghosts and they’re quiet these days.

“Pull yourself together man, you’re better than this, Sarah would have words if she saw you like this.”

I have my days, days of regret. Days when I can’t see the point of it all. Days when I remember loss, when the ghosts visit. It’s one of those days.

I’ve heard it said, “Happiness is having something to do, and a reason for doing it.” I need something. I’m not ready for garden centres and Thursday night bridge.

Sunday lunch with my old friend Phil will cheer me up. Nothing like beef, roast potatoes, a good IPA and the company of an old friend.

“Wash that miserable look off your face man and get out there.”

I arrive before Phil. The pub is welcoming with an old charm and subtle décor. I sit at our usual table and wait. Tracy, a pleasant girl, maybe 25, comes over. I like Tracy, she’s got a bit of edge about her. Not one of those clone girls with blonde hair, fake tan and eyebrows. She’s herself and I respect that.

“What you drinkin’, Pops?” She smiles. And Tracy always calls me Pops. I don’t mind. She’s the first person to speak to me all week. I don’t hear from Max much these days. He’s changed ever since Lola died.

“You know what I’ll have, IPA. And best pour one for Phil too, he’ll be here soon.”

“Right you are, sir….” We smile at each other. And somewhere in the back of my head I’m reminded of Barbara Windsor and Carry-On films.

I watch Phil at the door. I sense something. His skin is pale and the crow’s feet have carved more lines into his face. There’s me dying of boredom. And Phil dying of corporate nonsense, meetings about having meetings, shareholders, board meetings. And round it goes.

This is some horseshit, I think.

“Hey Phil, you look like you need that pint, you ok?”

“Marek, we need to talk.”

This is serious, no small talk, no dad jokes. Straight talking. Phil cups his hands round his pint.

“It’s that Brad Wesley, he’s working the board. Wants us to sell FoodCo. Make a quick buck. And no care for anyone else. No care for the business we’ve built and the people we employ. Families, Marek, we’ve employed generations at FoodCo.”

There’s silence between us for just a few seconds, but it feels longer. I hear the tell in his voice. I see his eyes. FoodCo is Phil.

“I know his type.” I break the silence. “Never liked him, never will. My bullshit radar was loud and clear. Selfish, narcissistic and greedy.”

“And get this, he even brought in the plant manager from Grainstow to see how we operate. Sandra on reception tells me he spent five minutes staring at the dashboard, muttered a few words and left.”

I look at my old friend, today he looks older. I know the board will sell. They have no other option. Then Grainstow will shut down FoodCo with over 70 years of baking heritage in the Cotswolds. Jobs will be lost. And the land sold to developers.

In time a new housing estate will be built with road names like “Bakers Way.”

“The board will sell.” I say. “They have no option.” It’s cold but it’s true — no option.

Tracy brings our Sunday roast dinners. And it dawns on me, we eat roast dinners at lunchtime, not dinnertime. Oh to be in England. My father must have been very confused as a Polish war refugee in the 1950s.

“Here you are gents, two of our very best roast dinners.” And with that smile, she’s off again.

We eat and our conversation moves to small talk. Phil’s grandchildren are a frequent topic. Always some mischief. I think grandchildren are a second chance of getting parenting right — only if you can hand them back after a few hours.

But behind the small talk that other part of my brain is focused on the word — option.

“Phil, fancy a fight?” Phil looks confused, one minute talking about grandkids, now fighting.

“Shouldn’t we step outside first? And at least finish our roasts first.”

“No, fight Brad Wesley and his type. Like we did before with that chump Keith Faulkner.”

“What are you saying?”

“Not saying, thinking. And I think we give the board Option B — expand FoodCo. Become predator, not prey.”

***

That night I couldn’t sleep, seeds of ideas came to me. But not connected, each seed of an idea sat in its own space. No cross-fertilised ideas. Just seeds born out of distant knowledge and experiences.

Option B, come on old man, think.

I lie in the dark and let the seeds sit and come together.

Three factories. Three separate businesses wearing one name. Three managers who’ve never once had to trust each other’s numbers, because they’ve never had to.

Steve’s spreadsheet, I think. Not Steve’s spreadsheet anymore. Steve’s model. Learning, adjusting, predicting — one factory, ten thousand loaves an hour, ninety-one percent accuracy, no Sunday nights. That much is already true. That much is already boring, which means it already works.

But it only knows one factory.

I turn onto my back. The ceiling gives me nothing, so I close my eyes instead.

What if it knew three.

Not three copies of the same model, sitting in three buildings, ignoring each other. One brain. One decision, made in one place, that understood all three ovens at once — this factory short on flour today, that one running under capacity, the third with a line down for maintenance nobody scheduled properly. A human would need a phone, three calls, an hour of politics before anything moved. The model wouldn’t need the phone.

A control plane, I think, and the word arrives before I fully understand why it’s the right one. Not a person in the middle. A place in the middle. Everything reporting to it. Everything adjusting from it.

I think about the flour tankers. Ninety tonnes a day, one factory. Multiply by three and the maths gets interesting fast — but only if the ordering knows what all three factories need before any human has to ask. Not FoodCo deciding what to buy. FoodCo’s systems and the supplier’s systems, talking directly, agreeing terms, adjusting quantities, the way two people who trust each other finish sentences.

Business to business, I think. Not FoodCo’s problem to solve alone. Never was.

And the retailers on the other end doing the same thing back — demand signals travelling all the way up the chain without anyone typing an order into anything. The whole thing breathing in and out, three lungs, one set of lungs.

Same headcount. Same costs. Three times the loaves.

No, I think, and this is the thought that finally sits me up in the dark, heart going harder than it should for a man doing nothing but lying still. Not three times.

Ten.

Three factories running as one is arithmetic. Anyone can do arithmetic. But a control plane that learns — that gets better every week instead of staying exactly as good as the day it was built — that’s not addition. That compounds.

I say it out loud to the empty room, the way I say most important things now that there’s no one here to hear them properly said back.

“Option B.”

Somewhere beyond the window a fox is arguing with something. I don’t sleep. I don’t try to. In my mind’s eye I visualise the data flows, the automated decisions. B2B Agentic Workflows feeding a Central Control Plane – The model is scalable – That’s it.

I scribbled some notes on a pad by my bedside. And that night I finally slept.

Something to do, and a reason for doing it. Purpose.