Credit: AndSus/Adobe Stock

Corporate lawyering is a grind. Attorneys slog away, drafting and editing dense tomes while clients wonder what takes so long and costs so much. The grind stems in part from the lawyers' role as "designated reader" across organizations and industries. As legal agreements from privacy policies to data licenses become more ubiquitous and more complex, attorneys find themselves responsible for review and translation of opaque language for business principals. Ever the good test takers, we lawyers have opted for careers in reading comprehension.

Since legal documents capture everything from commercial terms to risk allocation, lawyers are often tasked with negotiating these points, even when a business operator, financial professional or risk-management expert would make a better call. From handshake to signature, we approximate that 80% of time on contracts is spent on reading and writing by a lawyer or paralegal; 10% on business advice and 10% on true legal advice.

For lawyers who aspire to practice at the top of their license, the ratio stinks. And because of the density of legal language, the work involved in comprehending it, along with the specificity of the knowledge required to advise clients, legal services are priced out of reach for many common business transactions. The resulting vacuum yields legal forms drafted by, and highly favorable to, repeat players.

Artificial intelligence and the technology behind tools like ChatGPT change the dynamics of corporate lawyering by enabling fast and relevant insights into two common questions clients ask: what's missing and what's market? In any transaction between business parties—and in particular in David and Goliath transactions between a "NewCo" and a "BigCo"—clients want to know: "what are my blind spots? What should be in here that's not already?" And, for those terms already present: "what's fair given my situation?"

While much attention has been given to the ability of AI to draft contracts, the magic happens when the AI replaces the lawyer in reading and comprehending language with sufficient fidelity to propose a range of answers to both questions. An expert—legal or otherwise—interprets the output and advises the client. For simple transactions, a client may no longer need an expert (or the associated cost) to act with confidence

How did we get here? Legal tech burst into the collective consciousness with automated document drafting (recall LegalZoom and pitchman Robert Shapiro). This "editable PDF" era was characterized by standard templates where users filled in blank spaces while coached along by comment blurbs. This analog approach went digital over the past decade with "contract lifecycle management" (CLM) from a variety of technology providers, including Ironclad and Docusign. CLM allowed attorneys and legal operations professionals to generate agreements LegalZoom-style, while enabling tagging and search across the universe of saved agreements.

The tag-and-search models used by CLM tools learn on the basis of boolean, if-then rules (enabling "and"; "or"; "not"-style searching). Today, "natural language processing" learns on a contextual basis (more like the human brain), relying on huge datasets to establish networked connections between words to generate intelligent insights. Where once legal technology users filled in blanks on agreements they hoped were right for the situation, generative AI invites users to provide relevant inputs and have a contract drafted around them.

Relevant inputs are a language lawyers know well. From their first days in a law firm, partners advise associates to become trusted advisors to their clients: effective listeners who solicit their principal's interests and guide them expertly through challenging circumstances. A junior attorney navigates by a checklist that tracks approvals, agreements and signatures required to close a deal.

We expect legal AI models will learn similarly, using multiple choice checklists to capture common permutations of particular contract terms and typical combinations of sets of terms. Publicly available resources—such as the millions of agreements filed with the SEC by thousands of public companies disclosing material contracts—provide plentiful training content waiting to be sifted through the natural language processing tools that are coming to market now. The output results in variables and observations; the variables help answer "what's missing," while the observations help answer "what's market."

The tectonic shifts will ripple in waves across the legal industry. With demand down for reading and up for insight that can supplement AI products, law firms will need fewer associates and those they hire will have greater job satisfaction. Where today, technology fills in gaps left by lawyers, soon lawyers will fill in gaps left by technology. They will combine finance, risk and legal expertise with AI results, into products and services that advance client interests.

In contexts other than law firms, the results may prove even more dramatic. For in-house attorneys—especially the "Davids" selling NewCo products to the "Goliath" BigCo—the bane of commercial contracting practice lies in the so-called battle of the forms over which agreement template to use. Oftentimes, large customers require vendors to use the buyer's standard form of agreement that is highly favorable to the buyer. The associated anchoring effect that pulls outcomes in favor of the drafting party is referred to as the "drafter's advantage." Generative AI eliminates that advantage by scanning terms, assessing consistency with market standards and identifying gaps. Armed with AI that can process a contract of any length against a checklist of "must-haves," and supplemented by AI-generated insights into the relative frequency of such terms in a given market, the dynamic shifts convincingly back towards a neutral playing field.

Playing fields across the economy may be similarly evened. Imagine remodeling your kitchen and reviewing the contractor's agreement through recourse to the "Contracts for the Home" app. Imagine a browser extension that reads website terms of service and identifies what's off-market or what consumer rights typically present are missing. Imagine reviewing your employee offer letter, and your contract bot informs you whether the terms your prospective employer has called mere "boilerplate" are, in fact, industry standard.

A note of caution: market doesn't mean fairness. AI models build from existing datasets which include inherent biases. BigCos typically require NewCos to take on outsized liability for bad outcomes. Consumers regularly sign up to arbitration clauses which require them to submit to a decisionmaker far less favorable to them than a jury might be. Navigating the margin from "what's market" to "what's fair" requires judgment—expert or otherwise.

We are here for it. As career corporate lawyers, we are not afraid of losing our jobs. Instead, we're excited by what our jobs can become. The law offers a demanding and intellectual career, but the modern-day lawyer is deluged in paperwork. Our plan is to delegate as much as possible to the bots so we, as people, can maximize our time counseling clients on complex issues.

Jennifer Berrent is the CEO of Covenant, an AI-driven law firm. She is the former chief legal officer and chief operating officer of WeWork and a former partner at Wilmer Cutler Pickering Hale and Dorr.

Daniel Doktori is the general counsel of Credly, a software company.