The Empty Boardroom
The Silence of the Deal
The conference room was cold. It smelled of aggressive air conditioning and that specific, static scent of heated electronics. On the table, there were no printed contracts. There were no sweating sales representatives trying to hide a production delay behind a smile. There was just a laptop.
I was sitting with a client, the CEO of a mid-sized precision manufacturing firm. We were watching a negotiation happen in real-time. But we weren’t participating.
His purchasing agent—a piece of software—was locking horns with a supplier’s sales bot. They were haggling over the price of high-grade aluminum alloy. The text on the screen scrolled faster than we could read. Bid. Counter-bid. Logistics adjustment. Volume discount trigger.
“It’s aggressive today,” the CEO muttered, sipping his coffee.
He was right. The AI was fighting for a fraction of a cent. It was relentless. It didn’t care about the supplier’s relationship history. It didn’t care that the supplier’s daughter was getting married next week. It only cared about the parameters we had set: cost, speed, risk.
In less than three minutes, the deal was done. The purchase order was cut. The logistics were booked.
I felt a strange mix of German efficiency-lust and profound human obsolescence. This is the new reality. We are moving from a world of “people buying from people” to a world of “machines trading with machines.” And if you are not ready for the silence of the empty boardroom, you are going to get crushed.
The Death of the Lunch Meeting
For thirty years, my career was built on friction.
Procurement is—or was—a contact sport. You argue. You charm. You threaten. You go to dinner. You visit the factory floor and look for the dust in the corners that tells you they are cutting corners on maintenance. That friction was where the value lived. It was how we controlled risk.
Forrester and Gartner keep putting out reports saying that B2B transactions are moving to “autonomous agents.” They use fancy words. Let me translate what that actually means for us in the trenches.
It means the “relationship” is dead.
An AI agent doesn’t need to be taken to a steak dinner. It cannot be charmed. It cannot be bluffed. It negotiates based on cold, hard data access. If your inventory levels are visible via API, the buyer’s agent knows you are desperate to sell before the quarter ends. It will bleed you dry, not out of malice, but because that is the math.
The friction is gone. But so is the mercy.
The New Class War: Algorithm vs. Algorithm
Here is the dirty secret that the software vendors won’t put in their brochures. This shift is creating a brutal new class system in the supply chain.
We used to worry about “information asymmetry.” That’s when the seller knows the car is a lemon, but the buyer doesn’t.
Now, we are facing “model asymmetry.”
Imagine a poker game. Player A is a human. Player B is a basic computer program that knows the rules. Player C is a supercomputer that has simulated this hand fifty million times before the first card is dealt.
Who wins?
The big players—the massive OEMs, the retail giants—they are building Player C. They are deploying purchasing agents trained on oceans of historical data. These agents know the global price of copper down to the second. They know the weather patterns affecting shipping lanes.
Then there is the small supplier. The mom-and-pop machine shop in Ohio or the specialized component maker in Shenzhen. They don’t have an AI defense system. They have email.
When a sophisticated buying agent hits a manual supplier, it is a slaughter. The agent will exploit every inefficiency in the supplier’s pricing model. It will demand terms that a human would recognize as unfair, but the supplier accepts because they don’t have the computational power to counter-argue.
We are splitting the world into two groups: the AI-Haves and the AI-Have-Nots. If you are a supplier without an algorithmic shield, you are not a partner anymore. You are prey.
The Illusion of Control
“But Victor,” my clients say. “We still set the rules. The AI works for us.”
Does it?
Technically, yes. We move from being players on the field to being the referees. We set the “guardrails.” We tell the agent: Do not buy if the price is above X. Do not accept delivery later than Y.
This sounds powerful. It feels like being the general instead of the soldier.
But here is the trap. The complexity of these interactions is moving beyond human intuition. When two neural networks start negotiating, they might find solutions that are mathematically optimal but strategically suicidal.
I saw a case recently where a procurement bot optimized for “lowest immediate inventory cost.” It did a great job. It slashed warehousing fees to near zero.
How? By forcing suppliers into a Just-In-Time delivery schedule so tight that a single flat tire on a delivery truck shut down the entire assembly line for two days.
The AI followed the rules. It saved the pennies on storage. It cost the company millions in lost production. The “minimal friction” approach created a catastrophic failure point because the AI didn’t understand context. It didn’t understand that sometimes, spending extra money on inventory is an insurance policy, not waste.
We think we are the rule-makers. But if we don’t understand the game the machines are playing, we are just rubber-stamping our own destruction.
The Rise of the “Black Box” Negotiation
Transparency is the first casualty of this war.
In the old days, if a supplier screwed me on price, I could call them. I could demand a breakdown of their Bill of Materials (BOM). I could scream until I got an answer.
How do you scream at an algorithm?
As these agents take over, pricing becomes a “black box.” The supplier’s AI offers a price based on a thousand variables: your credit rating, current demand, commodity futures, and maybe even the time of day.
Why is the price $4.50 today when it was $4.20 yesterday?
“The model decided,” is the answer.
There is no logic you can audit. There is no spreadsheet you can check. It is just a floating number generated by a neural network. This makes “trust” impossible. You can’t trust the process because you can’t see the process.
This pushes us toward a strange paradox. We will have more data than ever before, but we will understand less about why things happen. We are trading understanding for efficiency.
What Human Procurement Becomes
So, should we all just retire? Go to a beach in Thailand and drink from coconuts?
Not yet.
The role of the procurement professional is changing, not disappearing. We are no longer negotiators. We are risk architects.
If the machine handles the transaction, the human must handle the exception. The machine can buy the steel. The human must decide what happens when the steel mill burns down. The machine can optimize the contract. The human must decide if we even want to be in business with this supplier at all.
We need to stop training junior buyers how to haggle. It is a useless skill. We need to teach them systems thinking. We need to teach them how to audit code logic, not just financial ledgers.
The Minimalist View
As a minimalist, part of me loves this.
The best process is no process. The best negotiation is the one that happens instantly, without ego, without wasted breath. There is a purity to algorithmic trade. It strips away the lies. It strips away the “sales fluff.”
But I am also a pragmatist. I know that factories are messy, physical places. Things break. People lie. Roofs leak.
An AI agent sitting on a server in Virginia cannot smell the burning plastic in a factory in Dongguan. It cannot sense the tension in the workforce that precedes a strike.
Use the agents. Let them handle the boring stuff. Let them buy the paperclips and the raw materials. But do not let them replace your eyes and ears.
If you stop looking at the reality on the ground because you are too busy looking at the dashboard on your screen, you deserve the disaster that is coming.
Keep your passport valid. Keep your boots muddy. And never trust a deal you didn’t check with your own gut, even if the math looks perfect.