Hogan Lovells Cadwalader faced a simple choice this month, and refused to make it. Two rival companies, Harvey and Legora, each sell AI built for lawyers. Most firms pick one. On Oct. 16 this firm signed with both1.
Timing matters more than the signature here. Two months earlier, a national survey counted entry-level hires at firms with more than 500 lawyers. The count fell 7.5 percent from the year before4.
That was roughly 540 fewer graduates, Big Law's first drop since 20144. The work that used to justify a large incoming class is the same work AI now does first.
Entry-level hiring at big firms falls 7.5 percent, the first drop since 2014
Change in entry-level associate hires, U.S. firms with more than 500 lawyers, Class of 2025 versus Class of 2024
The magenta share of the ring is how much the Class of 2025's hiring fell short of the Class of 2024's, the first such gap NALP found since it began counting in 2014.
Source: Above the Law [4]. Chart by The AILately.com Desk.
The numbers behind this chart
| Item | Value | Note |
|---|---|---|
| The decline | 7.5% | about 540 fewer graduates hired |
| Whole | 100% |
The Firm That Picked Both Sides
Miguel Zaldivar runs Hogan Lovells Cadwalader as chief executive. He explained the double booking in plain terms. "Through Harvey, we're perfecting the ways to capture and apply the expertise that exists across our firm on a variety of legal work," Zaldivar said1.
Harvey's side read the deal the same way, with less modesty. "Hogan Lovells Cadwalader has a clear vision for how Harvey can elevate its lawyers' work and client service," said John Haddock, Harvey's chief business officer1. Legora, the rival vendor the firm signed the same week, sells a similar promise to a similar list of firms5.
Lawyers inside the firm went further than buying software off a shelf. They built their own agents for knowledge management, capital markets work and international arbitration1. Staff tested each agent against real matters before any client saw its output, and partners kept the final sign-off on every filing6.
Hogan Lovells Cadwalader also runs ELTEMATE. It is a roughly 100-person technology unit that builds tools for clients, and for the firm's own lawyers5. Two outside vendors, one in-house shop, and a governance board to referee all three now sit inside one firm's AI strategy5.
What Junior Lawyers Used to Learn by Doing
Nikolas Guggenberger teaches law at the University of Houston. He studies what happens when software absorbs a profession's entry-level tasks. Junior legal work always did two jobs at once, he argues: it billed the client, and it trained the lawyer.
"If more and more of that work that trains junior associates is being automated, then there's no real material anymore for them to train on," Guggenberger told Axios2. Document review, first-pass contract drafting and basic research memos are the tasks a new associate used to cut their teeth on2. An AI agent now finishes the same tasks in minutes2.
Zaldivar's sentence and Guggenberger's sentence describe the identical pile of legal work. One calls it expertise the firm can finally capture. The other calls it the material a junior lawyer needs, and is losing.
A hiring partner reads the same stack of documents as a training budget. To a chief executive, that pile reads as a cost center finally under control.
The Billable Hour Loses Its Math
Every law firm still prices its work by the hour a lawyer spends on it. AI breaks that math at the root, because a task that took a junior associate a day now takes an agent minutes.
Thomson Reuters surveyed law firm professionals this fall. Roughly two-thirds said their firm kept its hourly prices exactly where they stood3. That held even as AI cut the hours behind each bill.
Lawyers who bill by the clock have little time left to rebuild the clock itself, one legal technology reporter observed3.
Ropes & Gray tried a fix of its own. Its new TrAIlblazers program lets first-year associates spend up to 400 hours on AI training3. Those hours replace billing clients for case work. That is about a fifth of a first-year associate's annual target.
Those hours still count toward the associate's target. They stay off the client's invoice entirely3.
Ropes & Gray carves 400 of 1,900 billable hours out for AI training
How a first-year associate's annual billable-hour target splits under the TrAIlblazers program
The bar is one associate's full-year target. The magenta segment is hours spent learning AI tools; the gray segment is hours billed to clients on case work.
Source: LawFuel [3]. Chart by The AILately.com Desk.
The numbers behind this chart
| Item | Value |
|---|---|
| AI training | 400 hrs |
| Client case work | 1,500 hrs |
| Whole | 1,900 hrs |
Paying for training time is a slower, costlier path, chosen on purpose. Skipping that program keeps a firm's margins intact today. It also bets that a smaller incoming class still learns enough to replace its partners in a decade.
Both bets show up in the same NALP numbers, filed under the same heading of efficiency34. Clients rarely see which bet their own law firm made. The gap shows only once the associates who made partner start to thin out.
What to do
A buyer choosing legal AI should ask a law firm one question before any pricing question. How does your firm train a first-year associate, now that AI drafts the first-pass memo? Ropes & Gray's answer, a paid training block built into the billable-hour target, shows the handoff can be managed on purpose3.
Ask which tasks a firm's custom agents absorb, beyond counting how many agents it built. Hogan Lovells Cadwalader's agents cover knowledge management and capital markets research1. That is the exact ground where junior associates used to learn the firm's own playbook. A client buying faster turnaround is also buying a quieter training floor.
By the numbers
- 7.5 percent: the drop in entry-level hiring at firms over 500 lawyers, Big Law's first decline since 20144.
- About 540: fewer graduates hired into those entry-level slots, year over year4.
- Two: AI vendors, Harvey and Legora, that Hogan Lovells Cadwalader signed with in one week15.
- Roughly 100: people on ELTEMATE, the firm's own in-house legal technology unit5.
- 400 hours: AI training time Ropes & Gray carves out of a first-year associate's 1,900-hour annual target3.
- About two-thirds: law firm professionals who told Thomson Reuters their hourly prices held steady this year3.
What to watch
Zaldivar's case holds if the firm's custom agents free senior lawyers for harder work. It also needs some other path that still teaches the juniors what those agents replaced. Guggenberger's case holds if entry-level hiring keeps falling the way NALP's data shows. It also needs firms to skip building any training path to match.
Watch next year's NALP hiring count. Also worth watching: whether more firms copy Ropes & Gray's paid training hours, or simply let the incoming class shrink. A third path is already visible at Hogan Lovells Cadwalader: keep hiring at today's pace, then measure in five years. The test is whether the smaller pool of tasks still produced partners who can run a matter alone.
Sources
- Harvey, "Hogan Lovells Cadwalader Adopts Harvey," Harvey, Oct. 16, 2026, https://www.harvey.ai/blog/hogan-lovells-cadwalader-adopts-harvey
- Axios Staff, "AI threatens Big Law's talent pipeline," Axios, May 2, 2026, https://www.axios.com/2026/05/02/ai-lawyers-law-firms-artificial-intelligence
- Rachel Williams, "Ropes & Gray Gives Juniors 400 Hours to Learn AI. AT&T Is Already Cutting Its Firm List.," LawFuel, Oct. 13, 2026, https://www.lawfuel.com/biglaw-ai-training-billable-hours/
- Staci Zaretsky, "Biglaw Is Bigger Than Ever, But Its Associate Classes Are Getting Smaller," Above the Law, Aug. 6, 2026, https://abovethelaw.com/2026/08/biglaw-is-bigger-than-ever-but-its-associate-classes-are-getting-smaller/
- Legal IT Insider Staff, "Hogan Lovells Cadwalader unveils firmwide AI strategy, signs with Harvey and Legora," Legal IT Insider, Oct. 16, 2026, https://legaltechnology.com/hogan-lovells-cadwalader-unveils-firmwide-ai-strategy-signs-with-harvey-and-legora/
- Unite.AI Staff, "Custom AI Agents Built by Hogan Lovells Cadwalader Lawyers, Harvey Says," Unite.AI, Oct. 16, 2026, https://www.unite.ai/custom-ai-agents-built-by-hogan-lovells-cadwalader-lawyers-harvey-says/