The buyer's AI gets there first

A buyer can investigate a need, compare suppliers and form a preference with AI before anyone on the selling side knows that a commercial opportunity exists.

Working with my AI is changing how I understand that possibility. A loose question can become a structured investigation: what is happening, what might be needed, what would solve it, and who could help?

The supplier enters the conversation after much of that work has already been done.

My working hypothesis is that this will change commercialization far beyond the adoption of another marketing channel or sales tool. It changes who performs the commercial thinking, when they perform it, and what a supplier must make available to survive the assessment.

The buying process can begin inside the work

Consider a company preparing to enter a new market.

A manager asks an AI to examine the plan. With access to the relevant information, it can help identify gaps in contracts, distribution, recruitment, systems or local knowledge. It can ask follow-up questions, distinguish immediate requirements from assumptions, and investigate whether the company should buy help, build a capability or do nothing yet.

The same process can begin with a personal task: planning a trip, reviewing a household expense or deciding whether a service still meets someone's needs.

A prompt can start the investigation. A developed workflow can repeat it as circumstances change. Given defined objectives, connected data and appropriate permissions, a system can flag new gaps in the business or in its own ability to complete assigned work. A missing integration, unreliable source or recurring bottleneck can become a requirement for a tool or service.

There is no independent desire behind this. People set the objectives and authorize the actions. But a person no longer has to formulate every potential buying need from scratch.

That moves the beginning of demand into the ongoing work of individuals and organizations.

Then comes the brutal due diligence

Once a possible need has been identified, the buyer can ask several AI tools to investigate it from different angles.

One develops the requirements. Another challenges the business case. Another compares suppliers, examines sources and looks for contradictions. The buyer can ask what would make each option fail, what evidence is missing, and which alternatives deserve further investigation.

That can produce a demanding assessment before a supplier receives an email.

Does this company understand our situation? Which people have relevant experience? What have they actually delivered? What does the service include? What will implementation require from us? Where are the costs, dependencies and risks? Can we verify the claims?

A polished promise becomes one piece of evidence among many.

I call this brutal because the assessment can be repeated, challenged and compared at a cost that makes much deeper investigation practical. The buyer can ask for another round without protecting the supplier's preferred story.

The quality still depends on the questions, the available evidence and the tools' ability to use it correctly. An AI can confidently misread a situation. Repetition alone does not make a conclusion reliable.

But a carefully directed investigation can give a buyer substantial preparation long before the first sales conversation.

The shortlist may exist before the lead

There is already evidence of this separation between buying and seller visibility.

In Gartner's survey of 646 B2B buyers, conducted in August and September 2025 and published in March 2026, 45 per cent said they had used AI during a recent purchase.

6sense's 2025 buyer research found that buying groups usually ranked their supplier shortlist before engaging sellers. It also found that LLMs were commonly used to compare offerings and synthesize information.

These findings support the direction of the argument. They do not establish that every transaction now begins with autonomous AI research.

They also contain a useful complication: 6sense found that buyers contacted sellers earlier than in the previous year, partly because they needed to evaluate suppliers' own AI capabilities.

More independent preparation can therefore coexist with earlier human contact. The important change is the assessment a buyer brings into that contact.

A supplier may receive a meeting request from someone who already has requirements, alternatives, objections and a provisional favourite.

Another supplier may have been removed from consideration without ever appearing in its own CRM.

What this does to marketing and sales models

A large part of commercial practice is organized around a sequence the supplier can observe: attract attention, capture a lead, qualify it, discover the need, present a solution and close.

The buyer's process can run ahead of that sequence.

AI can help clarify a need before a campaign reaches the person. It can compare services without completing a form. It can assemble answers from several sources without following the route designed on a supplier's website.

Funnels, AIDA and standardized sales stages can still describe parts of a decision. Their use as operating models becomes fragile when the company assumes that its own first recorded interaction marks the beginning of the buyer's thinking.

The consequences reach several familiar practices:

Content must help a buyer assess a real situation, including constraints, alternatives and evidence of delivery.
Important answers need to be accessible before a buyer is willing to exchange contact details.
Sales discovery must establish what the buyer has already concluded, which sources shaped it, and where the reasoning needs examination.
Measurement must account for the gap between what a company can track and what a buying group has actually done.
Brand and relationships must help a supplier earn consideration before a particular buying process becomes visible.

A familiar supplier can still have an enormous advantage. AI will also work with the preferences and assumptions people bring to it.

However, once a supplier is being examined, recognition alone may leave too many questions unanswered.

Competence becomes part of the commercial infrastructure

This creates a company-wide competence problem.

Marketing needs enough understanding of the work to communicate concrete value and defensible claims. Sales needs enough domain knowledge to contribute to a discussion with a well-prepared buyer. Specialists need to make their experience understandable outside their own profession.

Leaders need to connect those contributions with actual delivery.

A case should explain the situation, the work performed, the result and its limits. A service description should make the scope and practical requirements clear. A specialist profile should help someone understand which problems that person is equipped to solve.

Keeping these things accurate requires the organization to learn from its work and maintain a shared understanding of what it can deliver.

That is commercialization becoming a capability across the business.

AI may expose the distance between what a company says, what its people know and what its operations can support. Producing more persuasive text will leave that distance intact.

The buyer's organization has its own competence requirement: defining a good problem, challenging an AI's assumptions, checking sources and deciding when direct specialist input is needed.

An impressive comparison is only useful if it compares the right things.

What changed

I increasingly see the buyer's AI as a participant in commercialization.

It can help formulate the problem, examine whether a purchase is justified, define criteria and assess potential suppliers. Those activities can happen repeatedly inside ordinary work.

The seller may meet the buyer after those activities have shaped the opportunity.

This changes the first useful commercial question from how we get someone into our process to how we become relevant within a process they already control.

What I think matters

Companies should examine what a serious buyer could establish about them today, using the same AI tools that buyer has available.

Can the buyer understand the problems we solve, assess the relevant people, verify important claims and identify the conditions under which we would be a good choice?

That exercise connects market understanding, service development, digital presence, technology, selling, organizational learning and leadership.

It also changes the value of the first meeting. A capable seller can test assumptions, correct errors, reveal an overlooked dependency and help the buyer make a better decision.

As preparation becomes cheaper, that contribution may become more valuable. Repeating information the buyer has already assembled will be harder to justify.

What remains uncertain

We do not yet know how widely continuous, connected buying workflows will be adopted. Their usefulness will vary with data access, purchase complexity, risk and the buyer's ability to evaluate the output.

Public information can be incomplete or misleading. Hidden competence can be overlooked. An AI's recommendation can reflect stale sources, existing brand preferences or faulty reasoning. Direct dialogue, references and accountable human judgment remain essential in many purchases.

My expectation is that progressively more of the work preceding a purchase will be supported by AI, often beyond the seller's visibility.

That is enough to challenge the operating assumptions behind much standardized marketing and sales.

Commercial readiness increasingly means being ready for an assessment you may never know took place.

Sources

The sources describe surveyed B2B buying behaviour. The examples and wider implications in this note are my interpretation and working hypothesis.