Much of the conversation about agentic commerce begins with a consumer asking an AI agent to find a flight, order groceries, or buy a pair of shoes.
The technical questions are fascinating. How will the agent verify identity? How will it discover products, compare options, communicate with sellers, and complete payment? Which protocols will allow agents and commercial systems to interact?
But B2B commerce adds a more difficult question: On what terms may the agent transact?
Consumer transactions often occur within standardized structures. The buyer selects a product, accepts the seller’s terms, provides payment, and receives the goods or services. The transaction may involve significant legal complexity, but the consumer agent is usually choosing among options within rules established by the seller.
B2B transactions are different. The terms may be negotiated. Authority may be divided among legal, procurement, finance, security, privacy, and the business. The agreement may create continuing obligations involving data, service levels, insurance, audits, indemnification, renewal, termination, and regulatory compliance.
An agent cannot solve those issues by finding the right product and confirming that a payment method works.
Imagine a procurement agent selecting a software provider. It compares functionality and price, confirms budget availability, and identifies a preferred vendor. The systems can communicate. The agent can authenticate itself. The payment infrastructure is ready.
Then the vendor proposes its standard agreement.
Can the agent accept a limitation of liability based on 12 months of fees? Can it agree to automatic renewal? May the supplier use company data to improve its products? Does the agent have authority to accept a particular governing law, audit restriction, or indemnity obligation?
Technical interoperability does not answer any of those questions. The systems may be perfectly capable of completing the transaction while the agent lacks authority to make the underlying commitments.
This is where agentic commerce becomes a contracting problem.
In B2B settings, the agent must understand more than the agreement in front of it. It needs the company’s preferred positions, acceptable fallbacks, prohibited terms, and approval requirements. It needs to recognize when a commercially attractive deal creates an unacceptable legal or operational risk. It also needs to know who can authorize an exception.
Even after signature, the contract continues to matter. The agent may need to monitor performance, approve invoices, enforce service credits, manage usage rights, provide notices, or decide whether to renew. A B2B transaction is not a single moment of exchange. It is often the beginning of an ongoing governed relationship.
Lawyers should enter the agentic-commerce conversation before these systems are fully designed. If they arrive later, contractual controls may be treated as friction to be added around an already functioning commercial process. That is how companies end up with agents technically capable of acting but poorly governed when they do.
The legal team’s role is not to slow every transaction or require a human lawyer at every step. It is to help distinguish routine decisions that can be delegated from commitments that require judgment or approval.
That means asking practical questions now. Which transactions may an agent complete under standard terms? What monetary, legal, or operational thresholds change its authority? Which provisions always require review? How should conflicts among legal, procurement, finance, and business rules be resolved? How will the company record what the agent agreed to and why?
It also means making contracting logic usable. A folder containing template agreements and a playbook full of phrases such as “commercially reasonable” will not provide enough direction. Agents will need structured positions, contextual rules, and clear escalation paths.
Agentic commerce is often presented as a new channel for buying and selling. In B2B markets, it will also become a new channel for creating legal obligations.
The companies preparing the technical stack should prepare the contractual stack too. Otherwise, their agents may know exactly how to transact without knowing whether they should.
Olga V. Mack is the CEO of TermScout, where she builds legal systems that make contracts faster to understand, easier to operate, and more trustworthy in real business conditions. Her work focuses on how legal rules allocate power, manage risk, and shape decisions under uncertainty. A serial CEO and former General Counsel, Olga previously led a legal technology company through acquisition by LexisNexis. She teaches at Berkeley Law and is a Fellow at CodeX, the Stanford Center for Legal Informatics. She has authored several books on legal innovation and technology, delivered six TEDx talks, and her insights regularly appear in Forbes, Bloomberg Law, VentureBeat, TechCrunch, and Above the Law. Her work treats law as essential infrastructure, designed for how organizations actually operate.
