How to build robot-ready contracting playbooks

Admittedly it takes a considerable effort to build a playbook and that’s why attorneys tend to resist doing it. Nevertheless, AI can turbocharge contracting efficiency. In minutes, AI tools can review content, create redlines, insert commentary, and assess the degree of deviation of a proposed contract from industry standards or even a company’s standards. AI can “transform” everything. However, preparations are required before AI can make even a minor impact.

Think of AI as a diligent yet very green attorney. To be effective, that attorney needs opportunities, practice, feedback, and coaching. Ultimately, that attorney must be able to think through the issues and make decisions accounting for precedent, policies, priorities, and context. And just like a new attorney, AI needs practice, feedback, and guidance in assessing these variables so that it can be reprogrammed to improve.

But unlike a green attorney, mentoring cannot just be verbally coaching the AI; instead, what’s needed is a written process map capturing how an experienced attorney thinks through his or her review and processes information to make the best decisions. That calls for detailed instructions that can be turned into detailed coding. That also means crafting very robust playbooks – “robot-ready contracting playbooks.”

The problem?  In a typical legal department, playbooks don’t have that type of content and that is why it’s daunting to create them.

Therefore, what content does a robot-ready playbook need?

AI tools are loaded with significant functionality, which means they:  

• Are equipped to understand nuances in language to provide first-rate risk assessments, comparisons and redlines;
• Have the sophistication to determine if the language in a contract aligns with a preferred position, even when that language is phrased differently;
• Can further deduce if third-party content meets an acceptable threshold;  
• Can choose from different alternatives or modify third-party language to arrive at an acceptable position; and
• Can also make content conform to the defined terms, format, and flow of an underlying agreement.

However, AI can do none of that if it doesn’t have the underlying information to work from. While most legal departments and contract organizations have playbooks, these are usually not well-suited for AI use. The challenge is that most playbooks:

• Predominantly outline the company’s standard position and justifications for insisting on that position.
• Contain few if any alternatives.
• Are based on templates that are unreasonably tilted in favor of the company and do not reflect balanced positions.
• Are largely written by and for experienced practitioners.
• Have not been reviewed or updated on a regular basis.
• Fail to address topics commonly found in the other party’s template if those topics are not found in the company’s template.
• Fail to address situations where the other side refuses to negotiate.
• Are not organized by importance or weighting of the topic.
• Fail to impose a governance model for ultimate resolution of issues.
• Are not organized by ownership of resolution of issues related to that topic (e.g., business, risk management, finance, tax, information security, privacy, HR, export control, legal, etc.).
• Do not lend themselves to the logic of AI programming (e.g., if X, then Y, but if X1 and X2, then Y2).
• Do not reflect what happens in a negotiation and the results of that negotiation – e.g., agreed compromises are considered outliers rather than acceptable alternatives.

For an AI tool to provide useful results, it needs to start from a robust playbook that includes:

• Treatment of every topic that may arise – not just topics covered in a company’s template, but topics that the other side may want.
• Positions or clauses that must be and must not be in the contract (“hard fails”).
• A large, prioritized list of alternative positions or clauses.
• Alternatives that need special approval and by which department.
• Commentary that can be pasted into a redline to provide justification for modifications.
• Contextual parameters that determine which alternative to use (e.g., which party has more leverage, whether this is a one-time or master agreement, whether this is new or repeat business, the dollar value of the transaction or its strategic value, what has been previously agreed to with the other party or what has been agreed to with other parties under similar conditions, etc.).

It should be clear from these factors that robust, AI-appropriate playbooks need to be created by people intimately familiar with the agreement types for which they are written. But in-house counsel may not have the time or incentive to build that type of playbook. And forcing them to take on this responsibility might lead to a weaker result.

Solution Ideas

Take a small step and go generic.

A more generic AI tool can compare a third-party agreement or response to a company’s agreement to what it considers industry standard.

• Pros: Fast, easy, not resource intensive; results in decent information to aid in negotiations; fewer change management headaches.
• Cons: Unlikely to produce meaningful redlining.  For example, the industry standard may not align with a company’s particular needs and priorities or does not consider contextual parameters and is unlikely to convince the other party to change positions.

Take the full plunge (Internally).

Make the effort to create an AI-friendly playbook and work with an AI provider to test results, make modifications and test again. This means finding someone internally interested and motivated to modify the playbook; freeing up that person’s time; ensuring responsive collaboration from stakeholders in legal and other functions with expertise on covered topics; and collaboration from those ultimately benefiting from improved contracting (e.g., the businesses, sales, and procurement). In other words, leadership is needed to push this, remove hurdles, stay committed, champion the benefits across functions and generally create an environment for success.

• Pros: The AI tool can handle the brunt of higher-volume, non-bespoke contracting with fewer escalations and approval exceptions.  Personnel will be able to focus on more strategic relationships and become more expert on relevant laws and legal landscape (meaning less reliance on outside counsel).  You might be able to scale for increased contract volume and reduce headcount.
• Cons: This is not fast or easy.  It is resource intensive.  The general council (GC) and legal leadership need to take a significant role throughout.  You might experience resistance from those impacted or threatened and this will require a significant change management effort.

Take the full plunge (with outside help).

Find an external resource to put in the needed effort. Possibilities include:

• an alternative legal service provider (ALSP) with experience building these types of playbooks and working with AI providers,
• a law firm with AI integrated into its commercial practice and implementation capabilities, or a consultant with both legal and AI application experience to work with internal and external resources.

Legal leadership will still need to champion the cause, stay involved and create an environment for success. The ALSP or selected external resource will need access to templates, playbooks, negotiated agreements, stakeholders, negotiators and internal platforms and systems. And that external resource will have to work closely with the AI provider to build a playbook taking full advantage of the AI tool.

• Pros: Same pros as full internal effort; no arm-twisting of internal resource(s) to take on this project; no need to disrupt current support model or staffing; dedicated effort from a provider with skills and project management capabilities to get the job done; clearer ROI because of cost transparency.
• Cons: still needs to be sourced to a reliable external provider; most likely expensive because of the effort; costs need to be built into the budget; still requires GC and legal leadership commitment; still subject to resistance; still requires significant commitment to change management.

Enabling AI to think like a lawyer

A robot-ready playbook must provide clear guidance on what is preferred or acceptable and how to choose among alternatives. Basically, the playbook trains the AI to “think like a lawyer.” For example, a lawyer reviewing a contract would decide where to accept deviations and where to make edits.

That lawyer will also consider the contextual factors of urgency, dollar value, strategic importance, past history, leverage, etc., influencing whether to make major or minimal modifications.

Context helps an experienced lawyer determine the best approach in the given situation. But how does an AI tool complete the same process? Decisions must be step by step, starting with a default treatment assuming no extraneous factors.

Does a clause conform? If not, what is the next best alternative? Then contextual elements can be added.

The great thing about AI is it can sift through tons of data quickly. So, if the AI is to look for a previously agreed-to alternative with the party in an active agreement, it can scan databases to determine if that alternative exists. If it does, it will use that clause or position. If not, it will go back to the best alternative.

Alternatively, the AI can be instructed to use a position most recently agreed to in a similar contract. It takes only seconds for an AI tool to sift through those records. The playbook can prioritize among contextual factors, such as instructing the AI to look first to previously agreed-to clauses with the other party. In essence, this is the map of the lawyer’s thought process; however, now that lawyer has access to all sorts of data and does not have to hunt for it. The playbook can also include overriding instructions to reject a previously agreed-to clause if it triggers a “hard fail.”

This is not all that different from training a new attorney to understand how to approach a contract review and decide on the right course of action based on all the circumstances. The AI is then given the opportunity to use that training in actual or trial situations. This is followed by a review of the results to see how well the AI imitated what an experienced attorney would have done.  

If the AI does something unacceptable, the playbook has not provided the necessary guidance. That guidance must be adjusted for the AI to make different choices. Then the feedback loop repeats to see if that modified guidance leads to the right result.

A playbook should also map out the process for resolving stalemates when the other side will not agree to any acceptable alternatives. AI tools can facilitate the escalation with a workflow tool and ensure proper accountability for deviations from policies or standards. If an issue cannot be resolved, the topic owner is notified and must either agree to an exception or engage directly with the other party. This process establishes a record showing approval for significantly deviating positions.

Specific steps – creating the robot-ready contracting playbook

The good news is that AI tools can gather information from prior agreements and other sources, making this a much less time-intensive endeavor.

1. Gather relevant information, which should include:

• Template(s) for the agreement type, last revision date and process for updating templates.
• Existing playbooks or practice guides for the agreement type along with last revision date and process for updating these knowledge tools.
• A good sample size (say, 20 to 30) of in-scope signed agreements over the last few years.
• World Commerce and Contracting (WorldCC) templates¹ and other industry standards for the agreement type.

• Discussions with escalation approvers and legal personnel who have negotiated the agreement type on:

• • Commonly negotiated topics
  • Must-have and must-avoid clauses
  • Best practices and risk tolerances
  • Governance models for resolving stalemates
  • Contextual variances and exceptions

2. Make a list of relevant topics for that agreement type. These topics can be found in:

• The company’s template. If no template exists, review other templates, or signed agreements to determine preferred language.
• Industry templates. Review the WorldCC’s resources or benchmark template libraries to unearth these standards.
• Signed agreements, aided by an AI-enabled search to produce a list of commonly negotiated topics.

3. Determine the company’s position on each topic.

• Compare to industry standard positions.
• If heavily favorable to the company, push for more balanced position with topic owner.

4. List alternative positions/language.

• Industry standard is always a good start.
• Use the AI tool to help gather previously agreed-to deviations.
• Prioritize a list of acceptable alternatives.
• Add guidance on when to use each.

Example: 

Topic	Topic Owner	Company Position	Industry Standard	Alternatives	Guidance
Assignment	Name approval / escalation owner, practice group, or function	Company may assign to its affiliates or in connection with change of control. Otherwise, consent required.	Neither party may assign without prior written consent of the other party, except in cases of change in control.	Previously agreed-to assignment clause	Use previously agreed-to language if found in an active agreement of the same type (e.g., customer-facing).
				Industry standard	Industry standard is an acceptable alternative. If company is defined to include its affiliates, no further action; otherwise add affiliate.
				Each party may assign to affiliates; otherwise, consent required.	Acceptable if affiliate is not a competitor or an investment company.
				Neither party may assign without consent, which shall not be unreasonably withheld.	Acceptable for large-scale opportunities or strategic accounts, or if previously accepted with the other party.
				Either party may assign upon notice.	Better than having no assignment clause, which causes confusion.
				Only the counter- party may freely assign upon change in control.	Seek approval from topic owner. May occur with RFPs when bidder cannot make any changes.

 

Summary

Robot-ready playbooks are designed to enable purchasers of AI tools to derive actual value from that tool.  On a micro level, these playbooks are designed for use by AI implementers, such as CLMs and stand-alone AI service providers who offer tools to assist with contract review, redlining, and risk assessment.  On a macro level, these playbooks facilitate efforts by legal departments and contract organizations to integrate AI into contracting workflows.  A successful integration enables lawyers and contract specialists to spend much less time on lower value contracting and instead focus on higher value, more bespoke work for their company.

Building playbooks that are robot-ready will take effort. This article has explored various options for sourcing that effort, considering change management considerations. It has also provided a basic blueprint for development. Legal departments that put the effort into these playbooks will reap the benefits of increased capacity for higher volumes, reduced cycle times, consistency in risk management, accountability for deviations and increased bandwidth for staff to take on more strategic work. These are competitive advantages for any company, not just the legal department.

 

About the Author

Andrew Banquer, QuisLex Vice President, Corporate Solutions, is a veteran, hands-on architect and implementer of contracting improvement initiatives and on-going support services. He’s done extensive work with general counsel, commercial practice, and legal ops leaders to turn strategic objectives into actionable projects and activities. 

END NOTE

1.  See also Optimizing Contract Terms & Templates

Return