Our Take
Legal AI adoption is real and accelerating, but success depends on sustained effort and clear strategy, not hype or half-measures.
Why it matters
Law firm leaders no longer debate whether to adopt AI; they debate how fast to move and who will be left behind. The billable hour is cracking as a business model, and Big Law's 50-year dominance is being tested by NewMod competitors who've integrated AI into their core workflows.
Do this week
General Counsel: map your firm's current AI usage against a 12-month roadmap (not a vision statement) before your next CFO meeting, so you can budget for integration and training without overpromising delivery.
Legal AI moved from pilot to production at Legal Innovators California
The Legal Innovators California conference, held in San Francisco's Mission District, surfaced a consistent pattern: major law firms and in-house legal departments now treat AI as operational infrastructure, not an experimental line item. Ryan Walker, co-founder of NewMod firm General Legal, told an audience of general counsels that his firm combines AI tooling with elite Big Law talent to deliver work at a fraction of traditional firm rates and 10x faster. The response was immediate; attendees rushed the stage after his talk.
At the same conference, Clio founder Damien Riehl observed that the billable hour model, long a flashpoint for dissatisfaction, is "ill" rather than dead. The sentiment was echoed repeatedly by speakers and attendees, including those working inside major law firms. Gil Perez, head of innovation at Freshfields, underscored that change is a marathon, not a sprint. His firm's partnerships with Anthropic and Google required months of internal analysis before deployment began.
Nicole Diaz from OpenAI told attendees that she is hiring lawyers who function as builders and architects of AI workflows, not just practitioners of law. This framing signals a shift in what "the lawyer of the future" means: someone who designs and iterates on AI-driven processes for themselves and their clients, rather than delegating all technical decisions downstream.
The gap between progressive and conservative AI adoption is becoming a competitive wedge
Clayton Christensen's "Innovator's Dilemma" applies directly to law. Big Law is itself an innovation that solidified roughly 50 years ago and dominated the landscape accordingly. Annie Datesh, head of innovation at Wilson Sonsini, noted that ignoring disruptors tends not to end well. The legal market is not designed to move at the speed AI now enables, which creates friction for incumbents and opportunity for new entrants.
Alon Shwartz at Trellis framed this as a choice between progressive and conservative AI strategy. Firms pursuing conservative approaches implement only what's necessary to avoid falling behind. Progressive firms embrace the possibilities AI unlocks for their business model and client delivery. The NewMod firms already operating at scale demonstrate that the gap is not merely competitive; it is structural.
The billable hour is the linchpin. As long as firms are compensated for hours spent rather than outcomes delivered, AI-driven efficiency creates a margin squeeze, not a revenue opportunity. Sam Presvelos of Lexiden invoked Bezos: "Your margin is my opportunity." The firms that fix this incentive problem first will capture outsized value.
Build a concrete legal AI strategy before next year
The conference highlighted a common failure mode: floating without a plan. Attendees heard repeatedly that legal AI transformation requires sustained attention across three dimensions: AI capability (what the model can do), lawyer attention (how and when lawyers apply it), and attention to AI output (verification, editing, oversight). David Wang, head of innovation at Cooley, distilled this as an attention economy.
Scott Stevenson, CEO of Spellbook, presented statistical evidence that lawyers working without AI assistance make frequent errors, many serious. This is not an argument for blind trust in AI; it is a baseline against which to measure AI-assisted work. The comparison matters because it reframes the question from "Is AI perfect?" to "Does AI reduce error rates below human-only performance?"
For firms serious about embedding AI, the path is clear: commit to a multi-month discovery phase, then a sustained rollout. Freshfields' partnership work with Anthropic and Google took months of internal analysis to design workflows. Quick pilots generate visibility but not lasting change. The legal AI market is no longer in the phase where shipping a tool counts as strategy. Execution now means integration, training, and willingness to redesign billing and staffing models around the new constraints and opportunities.