AI is now your buyer

I've spent twenty years helping companies build go-to-market systems. I've watched SEO evolve, sales methodologies shift, marketing automation mature.

What's happening now is different.

Your buyers aren't just researching differently. They're not even researching anymore—not in the way you understand that word. They're running AI agents to evaluate you before any human at their company knows your name. And in most cases, those agents are deciding you're not worth talking to.

Here's the uncomfortable bit: you won't see this happening.

Your pipeline metrics look normal. Your sales team reports the same conversion rates. But underneath, something fundamental has changed. The point at which you get eliminated from consideration has moved earlier in the buying process than you can detect.

Let me be specific about what I mean.

The evaluation point has moved

In complex B2B, buying used to start when someone picked up the phone or sent an email. Before that moment, prospects might browse your website, download a whitepaper, attend a webinar.

You could track this. You had visibility.

That's over.

Today, a procurement manager at a £200M industrial manufacturer doesn't start by googling "aerospace polymer suppliers". They open a procurement agent, describe their technical requirements, and let the AI do the work.

The agent crawls supplier databases, parses technical specifications, checks certifications, evaluates risk signals, and produces a shortlist.

Your sales team never sees the 47 suppliers that got filtered out. They see the three that made it through. They think they're competing against two other vendors.

They're not seeing the evaluation that already happened.

This isn't coming. It's here. Keelvar, GEP, Aerchain—and a dozen other platforms—are already running autonomous sourcing workflows in enterprise procurement. They're handling supplier discovery, RFx orchestration, compliance checking. No human intervention for routine categories.

The question isn't whether AI is involved in buying decisions.

The question is whether your company can be properly evaluated by an AI that's never heard of you.

This is not search

Here's where most companies get it wrong. They think this is about being "visible to AI" or "optimising for AI search".

Wrong frame entirely.

Search assumes a human is looking for something and needs help finding it. What's happening now is different: buyers are delegating the entire evaluation process to agents that think in constraints, not keywords.

A typical procurement query doesn't look like a search. It looks like this:

"I need a thermoplastic with tensile strength above 90 MPa, operating temperature range -40°C to 150°C, UL94 V-0 flame rating, FDA food contact approved, processable by injection moulding, available in Europe with lead time under 8 weeks."

That's not a search query. That's a specification.

And the AI doesn't search for it—it filters for it.

It pulls structured data from every supplier it can access. It checks technical datasheets, certifications, manufacturing capabilities, geographic coverage, delivery performance. It applies boolean logic: does this supplier satisfy all constraints?

Yes or no.

If your technical specifications live in PDF brochures, you're invisible to this process. If your capabilities are expressed in marketing language rather than machine-readable attributes, you don't get evaluated.

You get skipped.

The AI doesn't rank you lower. It doesn't see you at all.

Agents eliminate, they don't discover

This is the bit that makes sales leaders uncomfortable.

These procurement agents aren't trying to find the best supplier. They're trying to eliminate unqualified suppliers as efficiently as possible. Their job is to reduce a universe of 500 potential vendors to a manageable shortlist of 3-5 that meet all requirements.

Elimination is cheap. Evaluation is expensive.

So the agents are optimised for aggressive filtering.

If anything is ambiguous—unclear certifications, vague capability descriptions, missing technical data—the agent doesn't investigate further. It moves on. There are 200 other suppliers in the database who documented their capabilities properly.

Your £100k marketing website that positions you as "a leading provider of innovative solutions" tells the agent nothing. It cannot extract constraint satisfaction from that language.

So it eliminates you and evaluates the competitor who published their technical specifications in a structured format.

You never know this happened. The buyer never knows this happened. The agent simply doesn't include you in the shortlist it presents to the human procurement manager.

Your sales team inherits a shortlist

This is where the commercial model breaks.

Sales has always assumed their job starts when the buyer makes contact. That assumption is now wrong.

By the time a procurement manager schedules a call with your sales team, the meaningful evaluation is already complete. The agent has already decided your company satisfies the basic requirements. You're on a shortlist of 3-5 vendors who all cleared the same filters.

Your sales team isn't winning the deal at this point. They're competing to not lose it.

The agent has already done the heavy lifting: technical fit, compliance, capability verification, risk screening. What's left is price negotiation, relationship assessment, contract terms. Sales influence matters, but the fundamental question—"should we consider this supplier at all?"—was answered before sales got involved.

This is a category shift, not an incremental change.

Sales used to control entry to the conversation. Now they inherit the outcome of an automated evaluation process they cannot see and did not influence.

Every buying cycle is now a re-qualification event

If you're an incumbent supplier, this is where it gets worse.

You used to have a massive advantage. Once you were in, you were the default. Buyers would stick with you unless something went seriously wrong. Switching costs protected you.

That protection is dissolving.

When a buyer runs an agent-mediated evaluation, it doesn't care about your existing relationship. It doesn't factor in "we've worked with them for five years". It applies the same constraint filters to every supplier in the database.

Including you.

If you can't demonstrate compliance with the new requirements as clearly as a competitor who's never worked with this buyer, you get eliminated alongside suppliers they've never heard of.

Every buying cycle—even renewals, even contract extensions—becomes a full re-qualification event. The agent starts from zero each time.

This means incumbency is no longer sticky by default. You have to re-prove your fit every time they evaluate the category. And if your knowledge representation has degraded while a competitor's has improved, you lose the renewal to an automated evaluation you didn't know was happening.

You'll find out 12-18 months from now when your pipeline starts looking anaemic and you can't figure out why.

The market is bifurcating

What happens when evaluation becomes machine-mediated and elimination costs drop to near zero?

The market splits.

There will be suppliers who are legible to agents—who publish structured technical data, maintain machine-readable specifications, expose their capabilities in formats that agents can query efficiently. These suppliers get evaluated in every relevant buying process, whether or not the buyer has heard of them.

Then there are suppliers who remain illegible—capabilities locked in PDFs, technical details buried in marketing copy, certifications mentioned but not documented. These suppliers only get evaluated if a human already knows to look for them. They depend on prior relationships, brand recognition, trade show presence.

The legible suppliers operate in an open market where agents can discover and evaluate them automatically.

The illegible suppliers operate in a closed market where access depends on human-mediated awareness.

This is the walled garden effect, and it's not a metaphor. Agents literally cannot see over the wall into your capabilities if you haven't built the gate.

The companies who figure this out early get evaluated more often, in more buying processes, by more buyers. They get more at-bats.

The companies who don't get fewer opportunities each year, and they won't see why until the revenue impact shows up on a delay.

Leadership denial happens because exclusion is invisible

Here's why most executive teams aren't reacting to this yet.

When a sales team loses a deal, they know it. The prospect says no. It hurts, but it's visible. You can analyse what went wrong and adjust.

When an agent eliminates your company during automated evaluation, nothing happens.

No rejection email. No lost deal in the CRM. No sales conversation that went badly.

You just never appear in that buyer's consideration set. And since you didn't know they were evaluating suppliers, you don't know you were excluded.

This creates a brutal time lag between cause and effect.

The agents are evaluating suppliers today. The shortlists are being generated today. But the revenue impact—the deals you should have won but were never invited to compete for—shows up 12 to 18 months from now when your pipeline metrics start deteriorating.

By that point, hundreds of buying processes have happened where you were invisible to the evaluation.

The commercial damage is already done. You're not behind. You're absent.

Most leadership teams will interpret this as "sales execution is slipping" or "our competitive position is weakening". They'll invest in sales training, brand campaigns, new messaging.

All of which are the wrong response to a structural problem they haven't correctly diagnosed.

Knowledge is now a commercial surface

This is the part that requires the biggest mental shift.

Your website is not your commercial surface anymore. Your brochures are not your commercial surface. Your trade show booth is not your commercial surface.

Your commercial surface is the sum of what an AI agent can determine about your capabilities when it interrogates the knowledge you've made available in machine-queryable form.

Think about that for a moment.

If your technical specifications are locked in a PDF, they're not queryable. If your certifications are mentioned in marketing copy but not exposed as structured data, they're not verifiable. If your application examples are written for human persuasion rather than machine interpretation, they're not usable.

The agent cannot work with this material.

So it works with your competitor's instead.

This doesn't mean you need to build an API or publish everything as structured data tomorrow. But it does mean you need to understand that the direction of travel is clear: B2B commerce is moving toward machine-queryable knowledge as the baseline for commercial presence.

Companies are already building MCP endpoints and agent-accessible interfaces because they recognise that evaluation will increasingly happen bot-to-bot before it happens human-to-human. This isn't a speculative future. It's a natural consequence of constraint-based evaluation at scale.

The question is not whether this will happen.

The question is whether your company will be legible to agents when it does.

You cannot opt out of being evaluated

The final illusion to collapse: control.

You do not get to choose whether AI agents evaluate your company. They're already doing it. The only choice you have is whether you're legible to the evaluation process.

If you're not legible, you're not protected. You're just invisible.

Some companies think they can avoid this by focusing on relationships, brand strength, or installed base advantage. That works until it doesn't. The moment a buyer runs an agent-mediated evaluation—and they will, because it's faster and cheaper than human research—your relationship doesn't protect you from being filtered out if you don't satisfy the constraints.

Your brand doesn't make you visible to an agent that's applying boolean filters. Your installed base doesn't prevent re-qualification in the next buying cycle.

This is not about optimising for AI. It's about recognising that the commercial surface has changed. Evaluation has moved. The point at which you get excluded from consideration is now earlier, cheaper, and invisible.

You're being evaluated right now by agents you'll never meet, for opportunities you'll never hear about, in ways you cannot detect.

The only question that matters is: can those agents figure out what you're capable of?

What this means for you

I'm not going to tell you what to do about this. That's not the point of this essay.

The point is to make you see something you cannot unsee: your buyers are using AI to eliminate suppliers before sales ever gets involved, and most companies are invisible to that process.

If you're running a £50M+ B2B business, you probably have 7-8 figures of revenue at risk over the next 12-18 months from buying cycles where agents will filter you out during automated evaluation.

You won't see it happening. It'll show up as pipeline degradation on a delay you cannot trace back to the cause.

This is not a marketing problem you can delegate. It's not an IT problem. It's a leadership problem about how your commercial presence is represented in a world where machines evaluate suppliers before humans do.

You need to decide whether you understand this shift, whether you believe it's real, and whether you're going to do anything about it while there's still time to influence the trajectory.

That's a decision only you can make.

But you need to make it now, because the agents are already running.