Most of the agentic commerce conversation starts with something you can put in a shopping cart. A buyer describes what they need, an agent compares products, and eventually someone authorizes a purchase. Product catalogs, specifications, inventory, prices. There is a reasonably well-defined thing to buy.
Now try that with a firm to handle a cross-border restructuring. Or an accounting team to support a carve-out. Or a contractor responsible for keeping a manufacturing facility running while a new production line is installed.
These are purchases, often substantial ones. They involve discovery, comparison, negotiation, contracts, and payment. They also involve conversations in which the buyer figures out what the problem actually is, whether the proposed team understands it, and whether these are people they want answering the phone when something goes wrong.
Sometimes the most valuable thing a provider does in that conversation is explain why the buyer needs different work from the work it intended to commission. That complicates the purchase before anyone has compared a price. A system can become very good at comparing proposals against a brief that should have been questioned.
Agentic commerce is going to affect these purchases. The fact that a buyer still wants to meet the people doing the work leaves plenty of room for the process around that meeting to change.
Quite a Lot Happens Before the Conversation
Consider a manufacturer that needs help with unreliable production. Someone asks around. A former colleague recommends a firm. An operations executive remembers a contractor from a previous employer. Procurement has an approved supplier list. The person who understood why those firms were on it has since left.
The company ends up speaking to a handful of providers, selected through some combination of experience, relationships, availability, and whoever happened to come to mind.
That process contains valuable knowledge. It also has a limited field of view. There may be a specialist two states away who has solved almost exactly this problem at six similar facilities. Nobody involved in the purchase knows they exist.
Agents will broaden that search. Relevant experience, geographic coverage, certifications, equipment familiarity, and the kinds of engagements a provider takes on are all useful starting points. An agent can identify gaps in the available information, request clarification, and help the buyer develop a set of firms worth speaking with.
There are already practical pieces of this. PERSUIT’s legal-services platform uses AI to turn matter briefs into requests for proposals, including questions, phases of work, and pricing requirements for litigation and M&A. Globality describes AI-assisted scoping and comparison across complex sourcing projects, including services. These capabilities stop well short of autonomously choosing the right adviser for a consequential engagement, but they reach into work that used to require substantial effort before a firm was hired.
My expectation is that agents will increasingly influence which service firms get a first conversation. A buyer may already have assessed the firm’s capabilities before anyone on its team knows there is an opportunity.
A Platform for Things That Don’t Fit in a Catalog
You cannot put a six-month operational turnaround behind a purchase button and consider the commercial arrangements settled. But a service organization can make parts of the buying process accessible through a common interface.
An agent could establish whether a firm handles a particular kind of work, request evidence of relevant experience, check capacity, understand its commercial model, or arrange a conversation with the appropriate person. Apify’s discoverable software Actors offer a loose analogy, although in services the next step might be a call with a partner rather than running a piece of software.
Imagine an accounting practice represented through several distinct capabilities: transaction support, carve-out financials, outsourced controllership, international tax. Each has its own requirements, evidence, availability, and route to an actual human. The initial interaction results in a diagnostic call. That is a useful outcome even if the engagement takes another six weeks to scope and sign.
Scientific research services are already bought through platforms. Scientist.com connects buyers with research providers and describes AI-assisted sourcing alongside purchasing workflows and human scientific oversight. It offers suppliers access to buyers looking for specialized expertise.
If enough qualified demand moves through platforms like these, established firms will have a commercial reason to participate. They may dislike the comparison criteria. They may worry about fee pressure or losing control of the client relationship. They may be right about all of it. They still have to decide how much business they’re prepared to miss.
A firm doesn’t have to list itself as one large organization that apparently does everything. A specialist practice or regional team gives the buyer something more useful to investigate. A contractor can offer a defined assessment that leads to a larger engagement. Several small firms can assemble a joint offering where the buyer needs a combination of capabilities, provided someone is clearly responsible for making the combination work.
The firm still has partners, phone numbers, meetings, and people responsible for delivery. It also has a presence in the systems buyers use to decide whom to call.
What a Rating Can Tell You
Ratings are already part of buying services. Clutch collects reviews of agencies, development firms, and other B2B providers. Avvo offers client reviews alongside its separate lawyer rating. High-stakes professional services have established rankings too, including Chambers’ research into lawyers and law firms. Nobody needs to invent the idea of evaluating an adviser in public.
The question is how much weight those evaluations carry when an agent assembles a shortlist.
There is useful information to collect. Did the provider deliver on time? How much did the scope change? Was the proposed team the team that showed up? Did the final cost bear a reasonable relationship to the original proposal?
There is also plenty of room for misleading comparisons. A difficult engagement can go badly despite excellent work. A straightforward assignment can produce a delighted client without testing much of the provider’s ability. One office’s reputation may tell you little about another team. The people responsible for the strongest references may have left.
These problems become especially consequential when the buyer is hiring someone to exercise judgment under uncertain conditions. A five-star restructuring adviser has presumably encountered a few unhappy people along the way.
As agents take on more of the comparison work, platforms have a reason to collect more detail about past engagements. An agent can examine details beneath a score: the type of work, the circumstances, the delivery team, and the basis for the client’s assessment. It can also make a shallow rating influential simply by using it as a filter.
For providers, the response should include better explanations of their work and commercial terms. Publish the assumptions behind the price: what is included, who will do it, what the client needs to contribute, and what would change the estimate. Otherwise, the firm that leaves the most out can look like the best value. Procurement has encountered this before.
The Shortlist Goes Back to the Network
After an agent produces a credible shortlist, a buyer will often do something very familiar.
They call a former colleague. They search LinkedIn for someone who knows the proposed partner. They ask an alumni group whether anyone has worked with the firm. They contact a reference and try to get beyond the polished account of how well everything went.
“Would you hire them for this?”
The last two words matter. A provider can be excellent and still be a poor fit for the situation in front of the buyer. Someone with firsthand experience can help explain the circumstances in which the firm did good work, how it handled disagreement, and what it needed from the client to succeed.
I expect trusted references to become more valuable in high-stakes purchases as agents make apparently qualified providers easier to find. The buyer has more options and more information, including proposals that may all be articulate and persuasive. Everyone will have a well-written proposal. This will be a considerable improvement in proposals.
A trusted person can help the buyer decide how much confidence to place in those claims.
Agents can help here too, if buyers give them access to company contacts and records of past work. A search might identify both a specialist firm and a colleague who worked with its delivery lead three years ago. A referral might start the process, with the agent then investigating alternatives and preparing questions.
Discovery will draw on several channels. Platforms, professional networks, past working relationships, and direct introductions will all contribute. Firms with strong referral businesses will retain a meaningful advantage, even as the route by which a new buyer encounters them changes.
For a partner thinking about what to do, maintaining those networks remains real business development. So does making the firm’s experience specific enough that a reference can speak to the team and situation the buyer is considering. “They’re good people” is pleasant to hear. It leaves quite a lot of diligence outstanding.
Sometimes the Brief Is Wrong
There is a limit to how much progress a buyer can make by comparing providers against a fixed specification.
Return to the manufacturer with unreliable production. It may think it needs a maintenance contractor. Conversations could reveal a controls problem, a scheduling problem, inadequate training, or several interacting causes. The right engagement might begin with a diagnostic rather than the work the company initially intended to commission.
An agent that finds twenty maintenance contractors has broadened the supplier search. It may still have left the buyer with a narrow understanding of the problem.
This is where discovery needs a wide aperture. A useful process should expose the buyer to different explanations and approaches, with people involved early enough to reconsider the brief. A specialist needs an opportunity to say, “We can do what you’re asking, but I don’t think it will solve the problem.”
That is also an opportunity for service firms to rethink their front door. A defined diagnostic gives the buyer something concrete to evaluate without pretending the eventual solution is already known. The firm can specify the required inputs, participants, duration, price, and output. An agent can help compare those initial engagements, while the experts take responsibility for investigating the situation.
Case studies can support the same approach. Explain what the client initially believed, what the team found, and how that changed the work. Where confidentiality prevents a public account, provide a sensible route to private evidence and relevant references.
Platforms will need room for providers to challenge a brief. Otherwise, they risk becoming very efficient at soliciting comparable proposals for work that should have been scoped differently. The buyer will have three comparable bids, a documented selection process, and the original problem.
Earning the Conversation
Selection still leads into agreements, staffing decisions, milestones, and an ongoing relationship. Agents will help with documentation, proposed terms, commitments, and routine coordination. Established terms and delivery history should also make repeat purchases easier to delegate. The degree of autonomy will depend on how clearly the work can be defined and what is at stake if it goes wrong.
For service organizations, preparation starts with making their expertise easier to investigate. Give specialist capabilities a clear entry point. Explain the conditions under which the work succeeds. Make pricing assumptions visible. Offer a diagnostic where the problem needs investigation, and maintain the relationships through which buyers can validate what they find.
A future client may arrive having compared the firm against a dozen alternatives, examined its proposed approach, and spoken to a former client. The first meeting can begin with better questions and a clearer understanding of what remains uncertain.
For a firm that does good work, a better-informed buyer is a welcome prospect. The firm still needs to be found and considered, and the conversation still needs room for someone to question what the buyer has come to buy.