Most contracting playbooks contain instructions that make perfect sense to experienced lawyers.
“Generally resist this provision.”
[AI Built for Litigation. Verified by Design.](https://bit.ly/3SwC2Af)
Grounded in authoritative content and verified at every step, Protégé is the only legal AI tool that delivers work you can trust—without exception.
“Accept if commercially reasonable.”
“Escalate material deviations.”
“Use judgment.”
A human lawyer can often work with that language. An AI agent cannot, at least not reliably. “Use judgment” is not an executable instruction.
[From ‘Vendor’ To ‘Partner’: How LexisNexis Is Deepening Law Firm Relationships](https://abovethelaw.com/2026/08/white-glove-service-in-ai-era/)
The company is emphasizing ‘white glove service’ in the AI era. Here’s what the initiative is delivering for clients.
As B2B agents begin participating in contract negotiations, companies will need playbooks designed for machines as well as people. That does not mean converting a Word document into a database. It means making the company’s actual contracting logic explicit enough that an agent can follow it, recognize its limits, and know when to stop.
Consider a familiar negotiation issue: limitation of liability. A conventional playbook might identify the company’s preferred cap, offer one or two fallback positions, and instruct the negotiator to escalate anything materially less favorable.
But what counts as material? Does the answer change based on transaction value, data sensitivity, the type of service, or the counterparty’s importance? Can the negotiator trade a higher cap for stronger insurance obligations? Who may approve the exception? Does an uncapped obligation require escalation in every case, or are there recognized exceptions?
Experienced lawyers often know the answers, even when the playbook does not contain them. They have negotiated similar deals, understand the company’s risk tolerance, and know which exceptions have been approved before. They also recognize when the business context makes an otherwise acceptable term dangerous.
An agent needs those connections stated.
A machine-usable playbook requires structured positions, fallbacks, limits, and approval requirements. It should identify the company’s preferred position, the range an agent may negotiate independently, the factors that change that range, and the point at which human approval becomes necessary.
It must also capture relationships among terms. Contract provisions rarely operate in isolation. A company might accept a different indemnity position if the liability cap changes, or permit broader data use if the data is sufficiently deidentified. An agent following clause-by-clause instructions could produce an agreement in which every individual term appears acceptable while the combined risk is not.
The harder challenge is capturing tacit exceptions. Many legal teams rely on unwritten rules that sound like this: “We normally accept that language, except for strategic vendors,” or “Legal approves these provisions, unless Security has concerns.” Those rules may work because the same experienced people apply them repeatedly.
Agents will expose how fragile that arrangement is.
This is not necessarily bad news. Preparing playbooks for agents may finally force companies to improve playbooks for everyone. Vague standards, conflicting policies, missing approval paths, and undocumented exceptions already create inconsistent negotiations. Human lawyers compensate for those weaknesses through experience, memory, and internal relationships. New team members, outside counsel, and business partners may struggle with them too.
The goal should not be to encode every possible negotiation outcome. Contracts are too contextual, and genuine judgment cannot be reduced to a very large decision tree. The better objective is to define where the organization has made a repeatable decision and where it has not.
That distinction matters. Agents can handle repeatable decisions when the company has established clear boundaries. Novel, consequential, or highly contextual decisions should remain visible as such and move to a person with the appropriate authority.
In-house teams can begin by examining their most frequently used playbooks. Look for words such as “reasonable,” “material,” “standard,” “significant,” and “generally.” Each may conceal a decision that humans understand differently. Ask what facts determine the answer, what range is acceptable, and what triggers escalation.
The exercise is not really about teaching agents how to negotiate. It is about discovering whether the company understands its own negotiating positions well enough to teach anyone.
B2B agents will need contracting playbooks. Creating them may reveal that many legal teams have been operating without complete playbooks all along.
*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.