When I started out as a dispute resolution lawyer, I worked at leading law firms on high-stakes cases involving everything from safeguarding the human rights of independence fighters to fine art fraud, lifesaving treatments, and superyachts. I believed the best lawyers were the ones who could win the hardest cases. Years later, after handling complex disputes across industries and jurisdictions, I’ve come to a different conclusion: with some exceptions, the best lawyers are the ones who kill disputes early.
Winning vs preventing
That is why I find today’s conversation around legal AI both energizing and superficial.
Energizing, because AI has the potential to transform legal services more profoundly than any technology shift, I have seen in my career. Superficial, because too much of the industry conversation still feels focused on optics rather than transformative outcomes.
Every week brings another announcement, with firms rolling out AI assistants across entire practices. But amid the buzz, a concern arises: is the legal sector drifting into AI window-dressing, focusing more on demonstrating AI adoption than proving business value or ROI?
AI hype vs value
Recently, a UK law firm made headlines after linking a £1 million bonus pool to employees collectively reaching one million Microsoft Copilot prompts. The target was reportedly achieved months ahead of schedule.
Don’t get me wrong. Law has historically been slow to adopt new technology, and firms want to encourage experimentation. Embracing AI is the first step, for sure. But it also raises an important question: when did prompt volume become the metric that matters, over, and above impact delivered by law firms to clients?
Using AI more often does not automatically mean clients receive better outcomes. Measuring adoption is easy. Delivering and measuring a return on investment is much harder. The simple point is that the industry’s ambition should go beyond making disputes more efficient to manage.
Right now, much of the legal AI conversation focuses on accelerating existing processes: reviewing documents faster, summarizing evidence more quickly or drafting submissions in less time. Those improvements matter. Anyone who has worked on large-scale litigation understands how impressive that efficiency can be.
Efficiency is not enough
If AI simply helps lawyers process disputes faster while the volume of disputes continues to rise, we are solving the wrong problem. Particularly as self-serving lawyers may use AI to increase the volume of material to be considered, in order to wipe out any net saving to clients. Notably, 84 percent of in-house lawyers say they have seen no noticeable reduction in fees or billed hours attributable to AI adoption by law firms, and a further 11 percent report that fees have actually increased.
The more transformative opportunity is using AI to prevent disputes from escalating in the first place. The short point: good tools make disputes efficient, but great ones make them disappear. That is why I founded Aavalynx, an AI-driven capital and risk management tool that helps companies reduce dispute length and costs. Our co-development partners include Vodafone.
Prevention drives values
Our early data suggests a ROI for clients of easily over 20x in saved costs and up to 200x when considering the rescued commercial relationships. This is obviously very significant.
Why is this the core of the innovation? Most disputes are not sudden crises. They brew gradually, often fueled by an under appreciation of risks. Our mission is to help identify those patterns early, before they become active litigations, arbitrations, and regulatory investigations. Horizon scanning may, in turn, lead in-house teams to propose operational changes to head off risks early. This underscores the role of the legal team not just as problem fixers but as ones delivering value in a partnership with business and operations.
And that is where I believe the real value lies. In most industries, the highest-value technology is preventative. The best cybersecurity systems prevent attacks before they happen. The best healthcare interventions focus on prevention, and early detection. Modern aviation safety is built around preventing failures, not simply responding to them better. As a large technology company has commented to Aavalynx: “Law firms sell us an excellent tumor removal service. But what we really want is the absence of tumors.”
And of course, not every dispute should be settled. It may be important to establish a precedent or to resist a reputation as a soft target for litigation. In these cases, it is important to have a holistic, data-driven view of the risks associated with each dispute.
To survive, law firms need to be thinking the same way. To predict, resolve and manage disputes proactively. To provide only the apex legal judgment, leading to predictability, risk visibility, and avoidance and commercial outcomes. No more enormous legal teams: the State of in-house 2026 survey shows that, as in-house teams adopt AI to automate routine, and capacity-driven tasks, the matters that remain with external firms are increasingly specialized, high-risk, or judgement-heavy. No old-world bills with a seemingly novel 20 percent AI discount on top, especially as law firms have been offering 20 percent discounts on their hourly rates time immemorial. Not only harnessing AI to leverage a law firms’ internal knowledge-banks – that is useful, but not sufficient. A law firm’s internal innovation must go further, leading to faster, cheaper resolutions.
The firms and legal departments that succeed over the next decade will not be the ones with the loudest AI announcements or the highest prompt counts. They will be the ones that go through a ‘category disruption’ to create a new type of service delivery – think Airbnb, Uber, and Apple. They will ask harder questions. Does this technology reduce disputes? Does it improve decision-making speed? Does it reduce legal spend without compromising quality? Does it help clients identify, and manage risks earlier? Are we sharing the benefits of our internal AI innovation transparently? These are the metrics that matter. We at Aavalynx deliver a significant ROI on these scores.
AI will absolutely become embedded across legal services. That debate is over. The real question is whether the profession uses AI to reinforce existing inefficiencies or to fundamentally rethink how legal value is delivered.
The future of disputes is not just resolving them more efficiently. It is preventing more of them from happening at all.
Hanna Roos
www.aavalynx.ai
Hanna Roos is Founder and CEO of Aavalynx, an AI-powered case intelligence platform for dispute resolution. A former Oxford and Cambridge-educated “top gun” disputes lawyer at Freshfields, Latham & Watkins and Quinn Emanuel, Hanna spent close to 20 years advising clients before building Aavalynx to give enterprises a real-time view of their dispute portfolios, resulting in a transformative reduction in dispute length and cost. Hanna has featured in Forbes, Vogue, ELLE and TEDx and spends her free time boxing, baking and doing car singalongs with her two boys and a sillly puppy.
