Rethinking the Billable Hour: Generative AI and the BigLaw Business Model

Tristan Ramon

Edited by Jordan Perlman, Luis Carvajal Picott, Judge Baskin, and Sahith Mocharla

In February 2026, the company Harvey AI announced that Gabriel Macht, known to many as Harvey Specter in Suits, entered into a brand partnership with the eponymously named company [1]. Macht framed his partnership as support for Harvey AI’s “responsible approach that keeps public interest in view,” emphasizing the platform’s role in improving the way legal teams work [2]. Harvey AI may be the first legal generative AI (genAI) company to sign Netflix’s best closer as a brand ambassador (without his psuedo-lawyer sidekick), but it is only one of several emerging legal genAI firms gaining traction in the legal technology sector. As investors pour hundreds of millions of dollars into large language models (LLMs) designed for legal applications, the traditional BigLaw business model is beginning to face future economic pressures [3]. By compressing the time required to complete routine legal tasks, advanced genAI challenges the economic foundation of the billable hour in an unprecedented way [4]. As legal genAI services continue to grow, firms that continue to fully rely on the billable hour will face an intensifying misallocation of resources between their conventional pyramid staffing structure and the reduced labor demands created by automation. Simultaneously, firms that are prepared to adapt their business model to more sustainable revenue structures will earn a significant competitive advantage. Consequently, large law firms aiming to remain competitive and profitable should plan for a structural transition in two main areas of the traditional ‘leverage’ model: training and pricing.

Currently, the traditional leverage model converts time worked by associates into revenue, with the billable hour functioning as the central measurement tool for pricing and performance. Popularized in the early 20th century, this leverage model creates a pyramidal staffing structure with its base made up of a large pool of associates and its peak made up of a small group of equity partners [5]. One major principle of the pyramid model is that task complexity increases as the pyramid narrows: relatively inexperienced juniors handle repeatable ‘grunt’ legal work, while more experienced seniors earn higher billable rates by taking on nuanced, discretionary projects. Historically, law firms have used the billable hour system to reliably convert attorney work into revenue at all levels of the staff pyramid [6]. As such, regardless of a project’s complexity, a given attorney’s revenue generation is directly proportional to their hours billed. One major benefit of the billable hour system is transparency, as firms and their clients can set expectations for fee services and retainers [7]. This transparency has enabled billable hours to dominate law firm management despite the fact that they enable firms to overrepresent hours worked and, consequently, overcharge clients. In fact, billable hours comprise over 90% of all U.S. legal cash flow and have already survived multiple predictions of their disappearance in the last five decades [8]. Such a strong reliance on the billable hour ultimately makes the traditional leverage model highly inflexible and incredibly entrenched.

Amidst BigLaw’s rigid business model, genAI is currently being integrated into corporate legal workflow. Many modern BigLaw firms have already invested in partnerships with leading legal genAI development companies, like Harvey AI and Clio, with the goal of developing firm-specific algorithms assisting with tasks like research, drafting, and document analysis [9]. A small number of firms have begun experimenting with billable hours dedicated to general genAI training, but the impact of advanced genAI on attorney work is yet to be seen [10]. As heavily funded algorithms continue to develop behind the scenes of the world’s largest firms, it can be expected that genAI will streamline BigLaw workflows on an unprecedented scale.

As genAI becomes widely adopted, the billable hour system faces stress because efficiency gains are difficult to translate into monetizable billables. Unfortunately for traditional BigLaw firms, if genAI-driven efficiency is not accompanied by a proportionate increase in demand for corporate legal services in the future, hourly billing will decrease firms’ profit margins and force a reckoning of the leverage pyramid [11]. When genAI models help accomplish in minutes what once took hours, billed time loses considerable value as a driver of revenue. Firms’ potential outflow of legal services will rapidly increase, but billable-based revenue can only remain healthy if an inflow of demand matches or exceeds new potential output. Ultimately, this represents a unique opportunity for firms to increase revenue by taking advantage of future efficiency and moving away from billable hours for specific legal tasks. In parallel, on the client side, costs of most corporate legal services can decrease and general accessibility can increase only if BigLaw firms strategically adapt their business models to advanced genAI. More specifically, competitive firms should prepare to redesign their approaches to training and pricing.

First, with regard to training, generative AI changes the nature of entry-level repetition work that traditionally trained associates. Thanks to the leverage model, BigLaw firms typically take advantage of lower levels of the staff pyramid by assigning junior associates mundane tasks that simultaneously train them and maximize their billable hour count [12]. Responsibilities like drafting, research, and document review are historically considered rites of passage for rising attorneys, building competence before moving up the leverage pyramid [13]. However, with the development of legal genAI, the BigLaw junior experience and its entry-level work will inevitably move away from traditional research and writing and instead towards AI-powered prompting and review. As advanced genAI drastically changes work demands, juniors will not spend nearly as many hours actively reviewing texts themselves and, as such, will likely not undergo a traditional form of legal training. The value of legal skills in the face of powerful genAI will quickly be put into question, and digital literacy skills may begin to define the most successful modern attorneys. In that case, the legal field risks facing a major schism between the skills prioritized by both law school and BigLaw firms and those prioritized by a genAI-dominated corporate landscape.

In parallel to the integration of genAI, supervision and verification will become priorities for both BigLaw firms and clients. While policy developments are rare in terms of ethical use of AI in the legal field, it is currently the case that lawyers remain fully responsible for competent representation when using AI [14]. Notably, in the state of Texas, lawyers are able to bill for time spent rigorously checking the output of AI used as a tool [15]. In the same way the Professional Ethics Committee for the State Bar of Texas clearly values verification of AI-assisted work, firms should prioritize transparency in the early stages of genAI adoption. Such transparency would involve external communication of genAI output verification with clients as well as internal accountability tied to junior-level staff training to responsibly use genAI. Structured methods for verifying safe tool use will become critical metrics of professional training as employees move up the staff pyramid. Especially in the early stages of genAI integration, institutionalized supervision during the training process will create a strong foundation of trust essential for successful firms in transition.

Consequently, AI competency can be treated as a worthy investment through solutions like training credit programs. Considering the output potential of future genAI used responsibly, BigLaw firms can smooth integration by treating AI training as a valuable, billable use of time [16]. If rising lawyers can begin developing firm-specific genAI fluency as such genAI develops, they will quickly be able to reap the benefits of its advancements. Such benefits include increased output that, if priced responsibly, will far outweigh short-term margin losses from training credits competing with traditional billable hours [17]. While operating profit margins will initially decrease due to increased wages from billable AI training, adaptable firms will quickly revert that downturn with a revenue upturn as legal genAI is implemented. To enable such a revenue upturn as early as possible, genAI training should become integrated into the leverage model long before the full adoption of in-house genAI. Thanks to this proactive approach, modern BigLaw firms can fully take advantage of genAI at all of its developmental stages while minimizing overhead inefficiencies tied to irresponsible or disorganized AI use.

Moving from training to pricing, hourly billing remains useful for tasks that involve critical evaluation and variance, but AI reduces time cost for repeatable legal work and increases pressure to develop alternate approaches for those components. As evidenced by the traditional leverage pyramid and the related hierarchy of legal task complexity, a corporate lawyer’s responsibilities range widely from entry-level work to nuanced advisory work depending on their experience [18]. When this spectrum is reconciled with the fact that genAI will streamline entry-level work from the large base of the staff pyramid, from the point of view of the firm, it is crucial to separate such work from less replaceable tasks. Hourly billing is most efficient for complex projects that are difficult to scope, so hourly billing will likely persist for responsibilities that make up the upper levels of firms’ leverage pyramids [19]. On the other hand, repeatable work will increasingly be treated as a set of predictable components that can be scoped, priced, and managed around outputs rather than only hours worked [20].

As output verification becomes prioritized, billable genAI use also increases the importance of pricing policies and client communication of such verification. Since genAI use allows a wide range of quality depending on the diligence of supervision and verification, internal approaches to genAI use will become highly relevant to clients being billed by the service [21]. Legitimacy traditionally stems from large amounts of billed hours, but the integration of genAI will require legitimacy to stem from internal policy that institutionalizes and communicates genAI use. One potential approach involves clearly distinguishing different categories of tasks and developing in-house methods to complete them with the help of genAI [22]. Every category requires a unique type and extent of overhead genAI use, and the clarification of different tiers will make concrete pricing expectations possible. Simultaneously, task categorization makes internal standards clear to junior lawyers who otherwise may face pressure to maximize AI-enabled efficiency without well-defined quality controls and disclosures. 

Having explored the areas of training and pricing within the BigLaw business model, there are many broader implications that follow the integration of legal genAI. First, firms will need to redefine their revenue models around value-based pricing and alternative billing. Moving away from hourly billing requires firms to precisely define and price common legal tasks, which will demand extensive research and development on task categorization, verification processes, and training. In practice, short-term investments in pricing research and training credits will temporarily lower profit per partner before sparking accelerated growth compared to that of non-adapted firms [23]. To remain competitive, new pricing must be strategically designed such that margins are maximized while prices stay optimal for clients faced with rival firms that still charge by the hour. Ultimately, law firms that successfully develop coherent and transparent pricing frameworks around fixed fees or alternative fee arrangements will capture value by turning AI-driven efficiency into profit, whereas firms that cling to the billable hour risk eroding their profit margins.

Inherently attached to AI-driven efficiency, general policy on AI use by lawyers will become extremely relevant. Current guidelines such as the Texas Bar’s Disciplinary Rules of Professional Conduct make clear that attorneys remain fully responsible for any AI-assisted work by verifying AI outputs, protecting client confidentiality, and charging only reasonable fees for their time spent using AI [24]. Notably, the word “reasonable” remains rather ambiguous, preserving the opportunity to further define pricing reasonability in the future. In terms of more specific policy, recent ABA ethics opinions emphasize lawyers’ duty to understand AI tools’ capabilities and risks, to obtain client consent, and to supervise AI like a junior associate [25]. At the same time, some legislatures and courts are moving to impose new requirements: proposals like mandated watermarking of AI-generated text could create compliance hurdles for attorneys using generative AI, and several judges have adopted rules requiring lawyers to disclose AI use in court filings [26]. These developments, while only in early stages, warn that misconduct related to AI often falls outside standard malpractice coverage. Firms must therefore hedge against the risk of malpractice and disciplinary or billing disputes by building strong institutional safeguards for training and usage of genAI itself.

More generally, law firms’ own data will become a prized strategic asset in the AI era. High-quality legal data, including decades of past case files, contracts, memos, and court filings—traditionally a firm’s proprietary knowledge—is the fuel needed to train accurate AI models and as such, demand for such data is surging [27]. However, there are potential obstacles related to database use. Most directly, feeding confidential client data into AI raises complex confidentiality and IP issues as firms must ensure that AI systems do not inadvertently memorize or disclose privileged information. Licensing further involves third-party AI systems that may require granting rights to the output or even the input data, risking loss of control over proprietary content. Robust data governance, encryption, and strict vendor contracts will be essential to manage these risks. Ed Walters, founder of Clio, summarizes this issue with a thought experiment: “Will it be Kirkland Ellis powered by Harvey AI, or Harvey AI powered by Kirkland Ellis?” [28]

Because of these data-driven dynamics, large, established firms are likely to further outpace smaller and mid-size firms. Well-resourced firms not only have more cash to invest in AI technology, tools, and infrastructure and alternative pricing development, but they also hold much more internal data to train and refine proprietary models. Today, large corporate firms are adopting AI tools and applying their extensive legal data at higher rates than solo or boutique firms [29]. In contrast, a typical small firm cannot easily assemble the same volume of training data or fund expensive AI research and development. As such, this gap may lead to a consolidation of market share among firms that can offer faster, more powerful AI-enhanced services. Smaller firms will need to find creative responses to stay competitive, like focusing on specialized niche practices where their human expertise outweighs data limitations, or partnering with technology providers to obtain AI capabilities. Niche differentiation, service innovation, and membership in AI networks or legal collaboratives may become necessary for smaller firms to compete with the giants.

Eventually, the very skill set required of lawyers will transform as technical literacy with AI tools will become a basic competency alongside traditional legal knowledge. Modern lawyers will need to be adept at selecting and overseeing AI systems, analyzing AI outputs for accuracy, and integrating AI results into legal strategies [30]. At the same time, skills like client counseling, strategic judgment, and ethical decision-making will become even more valuable. Essentially, lawyers will act more as problem-solving advisors than mere document drafters. Law schools are beginning to respond by implementing AI-oriented courses and modules to their curricula, but such changes are only nascent [31]. Educators emphasize that lawyers cannot outsource expertise to AI and that professionals must still master the underlying law to supervise AI like a junior attorney. In practice, this means training will need to balance foundational legal training with new AI competencies. However, the traditional legal education system is slow to change, and there may be a substantial lag before new graduates fully possess these combined skill sets.

Finally, the overall volume and accessibility of legal services is likely to rise sharply. As generative AI makes routine tasks far cheaper to perform, the latent market for legal help—individuals and small businesses who previously could not afford assistance—will gain access to legal services [32]. Many compare AI in the legal field to the 1913 Ford assembly line in the automobile industry. Just as assembly lines made cars available to the masses, AI has the potential to mass-produce basic legal documents and advice such as NDA and lease review, simple employment and contractor agreements, demand letters, and much more [33]. Lower-level legal work will become more affordable and available, meaning more legal issues can be addressed. Lawyers will spend more time working closely with clients on complex issues, which may increase job satisfaction for attorneys focused on those higher-level tasks. Notably, to account for billable time saved, firms will inevitably face the option to narrow their staff pyramids by decreasing the size of their junior pools. This cost-cutting incentive is attractive, but there are two main pressures that can help mitigate a narrowing of the traditional staff pyramid. First, if the inclusion of the latent legal market is substantial enough to drastically increase demand, junior attorneys will generally be faced with far more deliverables; genAI would reduce the time required to execute such deliverables, yet an increase in workload volume would counteract this compression of time worked. Second, given that vast entry-level work will become streamlined, firms can increasingly prioritize time-intensive “higher-level” tasks discussed above and assign them to genAI-enabled junior attorneys. This would preserve the value of billed time for tasks categorized as complex and decrease the need for a drastic change in the shape of the leverage pyramid. It is worth noting that the above implications hinge on careful, ongoing development of AI; administrative staff and regulators will need to continually refine policies to ensure quality, fairness, and ethical standards are maintained. If managed properly, the net effect could be a legal system that is both more efficient and more accessible, with attorneys applying human judgment where it adds the most value.

Ultimately, as generative legal AI becomes widely adopted, large firms aiming to remain competitive and profitable should plan for adjustments to training and pricing because AI changes the cost and supervision profile of core tasks while hourly billing and leverage remain deeply embedded. The evidence reflected in current market commentary suggests that the billable hour will remain resilient for particularly uncertain matters. However, repeatable categories will increasingly invite a shift toward pricing structures that are fixed in tiers. In parallel, firms will need to treat AI literacy, verification discipline, and supervision as core components of professional training. Together, these changes reflect a pragmatic restructuring of the leverage model, since the historic linkage between junior billable volume and both revenue and training becomes weaker as genAI reduces time worked.

Accordingly, the most competitive firms will be those that integrate workflow engineering, training reforms, and pricing design into a single strategy that clients can understand and accept. This approach aligns with the idea that value is increasingly measured by results and reliability rather than by time expended. Leading legal AI firm Legora’s website captures this exactly: “spend less time managing process, and more time delivering value” [34].

As a closing note, nothing is certain about powerful genAI and its place in the legal field. LLMs seem increasingly tailored for the tasks of modern lawyers, and there is no doubt that firms will feel pressure to adapt their business models to such unprecedented technology. In the words of Legora’s founder, “We don’t know exactly where the future is going, but neither do you. So, let’s work together to make sure we’re both winners in whatever happens” [35].

---

[1] Harvey Team, Gabriel Macht Partners with Harvey, HARVEY AI BLOG (Feb 17, 2026), https://www.harvey.ai/blog/gabriel-macht-partners-with-harvey.

[2] Bob Ambrogi, Harvey Partners with, Well, Harvey, Its Namesake as Brand Spokesperson, LAWSITES (Feb, 17, 2026), https://www.lawnext.com/2026/02/harvey-partners-with-well-harvey-its-namesake-as-brand-spokesperson.html.

[3] Sara Merken, AI Boom Fuels Fresh Wave of Legal Tech Investments, REUTERS (Nov 12, 2025), https://www.reuters.com/technology/ai-boom-fuels-fresh-wave-legal-tech-investments-2025-11-12/.

[4] Robert J. Couture, The Impact of Artificial Intelligence on Law Firms’ Business Models, HARVARD LAW SCHOOL CENTER ON THE LEGAL PROFESSION (Feb 24, 2025), https://clp.law.harvard.edu/knowledge-hub/insights/the-impact-of-artificial-intelligence-on-law-law-firms-business-models/.

[5] Fern S. Sussman, The Large Law Firm Structure–An Historic Opportunity, FORDHAM LAW REVIEW (1989), https://ir.lawnet.fordham.edu/cgi/viewcontent.cgi?article=2838&context=flr.

[6] George B Shepherd and Morgan Cloud, Time and Money: Discovery Leads to Hourly Billing, EMORY LAW JOURNAL (1999), https://law.emory.edu/_includes/documents/sections/faculty-and-scholarship/articles/cloud/1999-Time-and-Money.pdf. 

[7] James W. Jones, History of the Billable Hour, THOMSON REUTERS (Feb 11, 2025), https://www.thomsonreuters.com/en-us/posts/legal/billable-hour-history/. 

[8] The Legal Market at a Crossroads: 5 Forces Reshaping Law Firm Success in 2026, THOMSON REUTERS (Feb 6, 2026), https://legal.thomsonreuters.com/blog/the-legal-market-at-a-crossroads-5-forces-reshaping-law-firm-success-in-2026/.

[9] Gine B. Lynch, Paul, Weiss Partners With Harvey AI on New AI Workflows Innovation, PAUL WEISS (June 11, 2025), https://www.paulweiss.com/insights/firm-news/paul-weiss-partners-with-harvey-ai-on-new-ai-workflows-innovation. 

[10] Sara Merken and Mike Scarcella, Law Firms’ AI Experiment Gives Lawyers Break from Billable Hours, REUTERS (Nov. 6, 2025), https://www.reuters.com/legal/government/law-firms-ai-experiment-gives-lawyers-break-billable-hours-2025-11-06/.

[11] Karen Kwok, AI Dooms Billable Hour Big Law Earnings, REUTERS (July 2, 2025), https://www.reuters.com/commentary/breakingviews/ai-dooms-billable-hour-big-law-earnings-2025-07-02/. 

[12] Sara Merken, David Thomas, and Mike Scarcella, Is Big Law’s Pyramid Due an AI Makeover?, REUTERS (Dec 11, 2025), https://www.reuters.com/legal/government/is-big-laws-pyramid-due-an-ai-makeover-2025-12-11/. 

[13] The Hon. Martiza Dominguez Braswell, Legal Training in the age of AI: A Leadership Imperative, THOMSON REUTERS (Apr 30, 2025), https://www.thomsonreuters.com/en-us/posts/ai-in-courts/legal-training-ai-leadership/. 

[14] Ethics and Professional Responsibility, AMERICAN BAR ASSOCIATION, https://www.americanbar.org/groups/professional_responsibility/committees_commissions/ethicsandprofessionalresponsibility/ (last visited Mar 31, 2026).

[15] Tex. Comm. On Professional Ethics, Op. 705, TEXAS CENTER FOR LEGAL ETHICS (2025), https://www.legalethicstexas.com/resources/opinions/opinion-705/.

[16] Artificial Intelligence Practice, ROPES & GRAY LLP, https://www.ropesgray.com/en/services/practices/artificial-intelligence (last visited Mar 31, 2026). 

[17] Harvey Team, Measuring What Matters: A Practical Framework for Legal AI ROI, HARVEY AI BLOG (Feb 6, 2025), https://www.harvey.ai/blog/practical-framework-for-legal-ai-roi.

[18] Bryce Engelland, More Over, “Death of the Billable Hour,” Legalweek 2026 Has Found a New Existential Crisis, LegalWeek 2026: New Existential Crisis, THOMSON REUTERS (Mar 19, 2026), https://www.thomsonreuters.com/en-us/posts/legal/legalweek-2026-new-existential-crisis/. 

[19] LawPay, The Billable Hour Reimagined: A Toolkit for Future-Focused Law Firms, AMERICAN BAR ASSOCIATION (Feb 12, 2024), https://www.americanbar.org/groups/law_practice/resources/law-technology-today/2024/the-billable-hour-reimagined-a-toolkit-for-future-focused-law-firms/.

[20] Tomas Arvizu, From Billable Hours to Agentic Outcomes: Rethinking Legal Value in the Age of AI, THOMSON REUTERS (Jul 15, 2025), https://www.thomsonreuters.com/en-us/posts/legal/rethinking-legal-value/.

[21] From ‘Trust but Verify’ to ‘Do Not Trust Until Verified’: How the Legal Profession is Redeining AI Accountability, THOMSON REUTERS (Feb 26, 2025), https://legal.thomsonreuters.com/blog/from-trust-but-verify-to-do-not-trust-until-verified-how-the-legal-profession-is-redefining-ai-accountability/.

[22] Harvey Team, Introducing Workflow Builder, HARVEY AI BLOG (Jun 24, 2025), https://www.harvey.ai/blog/introducing-workflow-builder.

[23] Toby Brown, Bruce MacEwen, and Janet Staton, Exploring the Impact of AI on the Economics of BigLaw, THOMSON REUTERS (Apr 2024), https://www.thomsonreuters.com/en-us/posts/wp-content/uploads/sites/20/2024/03/Law_com-Pro-Fellows-Report_Exploring-the-Impact-of-AI-on-the-Economics-of-Big-Law_April-2024.pdf.

[24] ABA Issues First Ethics Guidance on AI Tools, AMERICAN BAR ASSOCIATION (July 2024), https://www.americanbar.org/news/abanews/aba-news-archives/2024/07/aba-issues-first-ethics-guidance-ai-tools/

[25] Keith Robert Fisher, ABA Ethics Opinion on Generative AI Offers Useful Framework, AMERICAN BAR ASSOCIATION (Oct 3, 2024), https://www.americanbar.org/groups/business_law/resources/business-law-today/2024-october/aba-ethics-opinion-generative-ai-offers-useful-framework/.

[26] Paul W. Grimm, Maura R. Grossman, and Daniel G. Brown, Is Disclosure and Certification of the Use of Generative AI Really Necessary?, DUKE JUDICATURE (2023), https://judicature.duke.edu/articles/is-disclosure-and-certification-of-the-use-of-generative-ai-really-necessary/.

[27] Gino Guarmere and Sarah Hirebet, How Data Strategy and Sophisticated Training are Intergral to the future GenAI Legal Future, REUTERS (Feb. 26, 2025), https://www.reuters.com/legal/legalindustry/how-data-strategy-sophisticated-training-are-integral-future-genai-legal-2025-02-26/.

[28] Ted Theodoropoulos, Ed Walters: How AI is Breaking the Big Law Training Model. LEGAL INNOVATION SPOTLIGHT (Jan 7, 2026), https://legalinnovationspotlight.com/ed-walters/. 

[29] Clio Team, Legal Trends Report, CLIO (2025), https://www.clio.com/resources/legal-trends/read-online/.

[30] Daniel J. Siegel, Training Lawyers to Bridge the AI Gap, AMERICAN BAR ASSOCIATION (Jan 1, 2026), https://www.americanbar.org/groups/law_practice/resources/law-practice-magazine/2026/january-february-2026/training-lawyers-to-bridge-the-ai-knowledge-gap/. 

[31] Karen Sloan and Sara Merken, AI Training Becomes Mandatory at More US Law Schools, REUTERS (Sept 22, 2025), https://www.reuters.com/legal/legalindustry/ai-training-becomes-mandatory-more-us-law-schools-2025-09-22/. 

[32] Nicole Black, Access to Justice 2.0: how AI-Powered Software Can Bridge the Gap, AMERICAN BAR ASSOCIATION (2025), https://www.americanbar.org/groups/journal/articles/2025/access-to-justice-how-ai-powered-software-can-bridge-the-gap/. 

[33] Roberto Valladares, Legal Tech: A Threat to Lawyers?, MIT PROFESSIONAL EDUCATION, https://professionalprograms.mit.edu/blog/mit-pe/legal-tech-a-threat-to-lawyers/ (last visited Mar 31, 2026). 

[34] Max Junestrand, Legal Work, Without Limits: About Legora, LEGORA, https://legora.com/ (last visited Mar 31, 2026). 

[35] Y Combinator, How This 25-Year-Old Built a $675M Legal AI Startup, Y COMBINATOR (Aug 26, 2025), https://www.ycombinator.com/library/Mq-how-this-25-year-old-built-a-675m-legal-ai-startup-with-no-legal-experience.