Legalweek 2026: Inside the Shift in Legal AI

LegalWeek 2026 marked a shift in how the legal industry is talking about AI. It’s no longer about whether firms should adopt it—it’s about how to get it right at scale, in a way that stands up to client scrutiny and delivers measurable value.

Jessica Gersten, Legal Solutions Architect at Orbital and former Am Law 100 real estate attorney, was at LegalWeek and shares her insights.

Rethinking Client Expectations

For the first year or two after generative AI arrived, many clients were actively cautious, asking law firms to keep their data out of AI systems altogether. That point of view has reversed sharply. Matthew Beekhuizen, Chief Pricing and Innovation Officer at Greenberg Traurig, observed that clients now not only permit the use of AI—they actively expect it. As Mark Brennan, Hogan Lovells’ Global Managing Partner for Digitalization put it: "AI is no longer viewed as optional or experimental: most clients actually assume firms are using AI in some capacity. Expectations have shifted."

And when clients do notice the efficiency gains AI delivers, they increasingly expect those gains to be reflected in their fees. That creates a new kind of pressure on firms: not just to adopt AI, but to be able to account for it: to show where it is being used, what it is producing, and how it is changing the value they deliver.

The Billable Hour Is Under Pressure

If there was one tension running beneath every session at LegalWeek this year, it was the paradox of AI dramatically increasing efficiency and saving time inside a business model entirely built around how long the work takes. Speakers across sessions were candid about how difficult this is to navigate in practice. Efficiency gains from AI often occur in downstream tasks that clients don’t directly see, making them difficult to quantify or communicate.

The legal industry is beginning to experiment with alternatives to the traditional billable hour model. Hybrid structures, fixed fees on front-end AI work, and rate premiums for AI-proficient attorneys are all being tested. One firm described charging a fixed fee for AI prompting and validation performed by a specialized team, with a separate billing structure for secondary review. Others are exploring using historical matter data, aggregated across comparable cases, to generate fee predictions at the outset of a transaction to provide more certainty for clients upfront.

Underlying all of these experiments is a more fundamental question about how to measure the value of AI. Beekhuizen offered perhaps the most grounded framing: "Your ROI is going to be that you get to stay in business." His argument is that AI needs to be treated as overhead, rather than as a discrete investment expected to generate a line-item return. The firms that are tying themselves in knots trying to justify AI spend in isolation are asking the wrong question. The right question is whether it is enabling them to do better work, take on more of it, and retain clients who expect it.

Build vs. Buy: The Discipline of Tool Selection

As the legal AI market matures, law firms are shifting away from thinking about tools and toward the problems they are trying to solve. But as the number of products grows, so does the difficulty of evaluating them. Several panelists pointed to the rise of “AI washing”: tools that claim advanced capabilities but in practice offer only a thin layer of automation. In response, some teams are adopting more disciplined approaches, including simple “stopwatch tests” to measure how long tasks take with and without a tool before expanding pilots. That kind of grounded evaluation is becoming more common across the industry.

Sergey Polak, Director of Technology Innovation at Ropes & Gray described how this discipline starts before any tool is considered through continuous conversations with practitioners about what they are actually trying to solve and what they wish technology could do. Emily Florio, Director of Knowledge Research at DLA Piper applies a similar filter: is the problem repeatable, is AI the right tool, and does the tool actually contain the right content?

Once a tool does meet the need, the build-versus-buy question becomes more concrete. Some described evaluating third-party tools across functionality, privacy, and regional constraints, while also asking where it might make sense to build on top to better fit firm workflows. A counterpoint raised was that anything built internally comes with ongoing maintenance and support costs that are often underestimated. In practice, the decision on build vs. buy unfolds gradually: through evaluation, testing, and the gap between what a tool promises and what it delivers.

The Reality of AI Adoption

Most of the panelists agreed, the biggest barriers to AI adoption in legal practice are not technical—they are organizational. The challenge is getting people to use tools consistently, in the right way, for the right tasks.

Successful adoption lies more in a cultural shift than in a technology rollout, and one that happens gradually. It depends on identifying champions early and creating the conditions for attorneys to see AI working in contexts relevant to their own practice. As Brennan put it: "momentum builds when people see tangible scenarios connect to how they work and their practices." Matthew Kohel, Partner at Saul Ewing, added that meeting attorneys where they are technologically (through town halls, ongoing training, and an adoption strategy that accounts for the full range of digital comfort levels) is just as important as the tools themselves.

Organizational culture, though, is only part of the equation. A point that resonated across sessions: before any firm reaches for a new tool, it needs to ask whether it actually has a process—or just a collection of habits. AI cannot fix a broken workflow, it can only accelerate whatever is already there. The most important operational move most firms can make is to document their workflows, standardize them, and turn them into policy before layering automation on top.

For more on how real estate legal practices are turning AI into real results, read our latest CIO forum write-up.

Final Thoughts 

For me, the consistent thread across the sessions I attended at Legalweek was that progress depends less on the volume of tools firms adopt and more on the foundations beneath them: the processes, the people, and whether the tool they choose was actually built for the work they do.

The same principle shapes how we think about our work here at Orbital. In real estate, that distinction matters more than most. This isn't generic legal work: it's spatial, location-based, jurisdiction-dependent, and deeply physical. Decades-old easements, metes and bounds descriptions, surveys cross-referenced against title commitments. None of that fits neatly into a generalist "upload your document" workflow (exactly where most legal AI loses momentum). 

When the tool matches the reality of the practice, adoption follows naturally. Orbital Copilot is grounded in that specific complexity, with every workflow designed by our team of 15 former practicing real estate attorneys around the document types, processes, and pain points that define the work. LegalWeek 2026 reflected an industry that has moved past the question of whether AI matters and is now grappling with the harder ones: how to implement it responsibly, how to measure its impact, and how to build organizations that can sustain it over time. Those are the right questions—and the ones that here at Orbital we've long been focused on answering.

Interested in bringing your real estate law expertise to legaltech? We’re hiring Legal Solutions Architects (just like Jessica) and Legal Engineers to bring their law firm experience to Orbital. You'll partner with AmLaw 200 firms and in-house teams as a trusted advisor: guiding pilots, coaching through change, and driving adoption of AI built specifically for real estate practice. Find out more.

‍