Very few people in the legal industry have worked inside as many of its seats as Mary O’Carroll. She fell into the industry more than two decades ago, joining Orrick from a background in banking and consulting, and remembers thinking within her first week that the profession was stuck in time. That observation became the starting point for the career that followed.
She moved client-side to Google on the theory that real change would probably originate from the buying end of the market. She co-founded the Corporate Legal Operations Consortium, better known as CLOC, and served as its president. She then joined Ironclad, one of the leading legal technology companies, and later went to Goodwin Procter as chief operating officer, having always said she would never return to a law firm.
She is now chief executive of LegalEng Consulting Group, executive-in-residence at SignalFire, an advisor to Sandstone, and the host of Pearls On, Gloves Off, a podcast that recently released its hundredth episode. The vantage point is unusually wide.
This week, The Legal Wire sat down with Mary. What follows is not a career profile. It is a piece about how someone who has spent two decades genuinely inside the profession reads where it now sits, and where she thinks it is heading.
The industry she walked into, and the one it kept trying to be
The starting observation, and one Mary returns to across the conversation, is that the legal industry was never modernised in the way other professional-services industries were. She had come from banking and consulting, where clients had already reshaped how work was priced, staffed and delivered. What struck her most was how misaligned law firms were with what their clients actually wanted. The incentives, the business model, and the way firms operated were not really aligned with client outcomes.
That misalignment became the animating question of her career. Everything she did after Orrick was, in some form, an attempt to figure out how the industry could modernise and how legal services could be delivered differently.
Two decades on, her assessment of how far the industry has come doesn’t come across as sentimental. It has come further than a cynic would allow, but it has come nowhere near as far as it should have.
The client side, and why change originates there
Mary moved to Google on a specific theory: that clients ultimately have the buying power, and that if the profession changed, it would be because clients demanded it. What she found inside a large in-house department was more complicated than the theory implied.
In-house teams, she quickly realised, operate very differently from law firms. Different priorities, different pressures, different success metrics. But most in-house lawyers had come up through law firms and had naturally carried over the habits and assumptions they had developed in private practice. In-house was structurally different, but the people inside it had, in many cases, been trained by the very system they were now supposed to reshape.
What was genuinely new was legal operations. That role was not about practising law. It was about making legal departments more efficient and more effective, finding the right balance between speed, quality and cost. It was a completely different way of thinking, and it created the space for the whole discipline to emerge.
CLOC grew out of that movement: legal operations professionals from large companies coming together to share best practices and collectively define what modern legal departments should look like. As it grew, venture capital started flowing into legal technology and the wider ecosystem began evolving more quickly.
TLW: You’ve said in-house lawyers still carry over habits from private practice, which makes them a less transformative buyer than the theory would suggest. Have you seen that dynamic starting to shift as a genuinely operations-native generation reaches senior positions in-house, and if so, what does it change about the pressure firms are feeling from their clients?
Mary: “I’ve seen this shift, and it’s real, though it’s happening slower than I’d like. When I joined Google, most in-house lawyers had one professional identity: they were lawyers who happened to work inside a company. Today, a growing number of legal leaders have spent significant time working alongside a legal ops function. And that changes how they think. They know value isn’t delivered in six-minute increments. Simply put, they’re buyers first. These buyers want to know how a matter will be staffed, what technology will be used, whether the pricing actually reflects the value delivered instead of just the hours logged. And they’re comfortable pushing back, because they’ve spent years focused on measuring outcomes, not activity.
If we’re being honest though, this group is still a minority. A lot of legal departments are still led by former Big Law partners who’ve never worked anywhere else, who haven’t had their eyes opened to how the rest of the world operates.”
Why Big Law is not going anywhere quickly, and why that is not for the reason people think
The question of Big Law’s future is one Mary is asked constantly, and her answer is measured in comparison to what the industry usually gives itself credit for hearing. That is, Big Law will absolutely still exist ten years from now.
Clients already want firms to operate differently: different pricing, different service delivery, effective use of technology. But they do not, in practice, have many alternatives today. That is why change has been slower than many people expected.
The pressure is genuinely increasing. More work is being brought in-house, AI-native law firms are emerging, and some are attracting significant funding and very talented lawyers. But shifting the centre of gravity away from the largest firms takes time. There is an enormous amount of institutional inertia. Big Law has more time than people sometimes assume.
Where the standard critique goes wrong is in assuming firms have not invested. They have. However, buying technology, hiring innovation leaders and making headlines about AI investments is the easy part. The difficult part sits at a much deeper level: firms changing their core business model.
Her answer to that issue gets straight to the point. Their incentive system is still built around the billable hour, and that creates a fundamental conflict with AI. AI reduces the amount of time required to complete legal work. If lawyers are still rewarded for billing more hours, why would they enthusiastically embrace technology that reduces the very thing they are measured on. “People often argue that lawyers will simply be motivated to use AI because it’s better than without,” she said. “Of course they do. But lawyers aren’t irrational. If you’re still expected to bill 1,800 hours every year and your bonus depends on reaching that target, using technology that dramatically reduces the time required creates a very real tension. You can’t simply invent more work if there isn’t enough work to go around.”
The implication is that most of what the profession calls AI resistance is not really about the technology. It is about the incentives underneath it. Until firms address that, they will continue struggling to become truly AI-native. Restructuring the business is the work most firms have not yet begun.
TLW: You’ve made the point that firms have invested in technology but haven’t restructured the incentives that would let it work properly. What have you seen at firms that have started the harder work (i.e., the underlying incentive redesign) and how does that decision reflect in practice for the partnership?
Mary: “Very few firms have done this work, which is exactly why it stands out when they do. Let’s be clear about what ‘doing the work’ actually means. Some firms have added a category of AI hours that count against the billable target. That’s a step, but honestly, it isn’t much of one… you still have your billable hour targets, and that’s still how performance gets measured. Nothing about the underlying structure has changed.
What I want to see is true performance metrics tied to impact: client value, client retention and expansion, efficiency. I like to call this “staffability”. In other words, do clients actually want this person on their matters, over and over again? That’s a very different question than “how many hours did they bill.”
I also like seeing that some firms are shifting some of their focus on profitability and margins rather than just revenue targets and growth. That’s the incentive shift that actually matters, because it lets partners think about growing margins by being more efficient, even if that means less revenue some years, or a different fee structure altogether.
And honestly, it starts even earlier than partnership. Firms are still hiring associates as soon as they get to law school. That’s nearly three years before you actually need them! How can you possibly know what your practice will need three years from now?”
Legal technology, in-house, and where the products work
Another important observation Mary makes, one carrying a real editorial charge for the legal tech industry, is about the market itself. One of the biggest challenges, she says, is selling technology to organisations whose business model is not aligned with the value the technology creates. If your product makes lawyers significantly more efficient but those lawyers still generate revenue through billable hour models, you have immediately created a conflict.
Many legal tech companies have sold successfully into law firms, but widespread adoption and meaningful client impact have not always followed. That is one of the reasons so many legal tech companies have gradually shifted their focus toward in-house legal departments. It is where the technology is much more likely to be embraced and used in the way it was intended.
The in-house environment is often more receptive because the incentives are different. The objective is not maximising billable hours; it is delivering better legal services faster, more efficiently, and at lower cost. Technology aligns naturally with those goals.
Her advisor role at Sandstone sits inside that thesis. She is selective about the companies she advises. The questions she asks herself before saying yes: is the company solving a problem she genuinely believes needs solving; does she believe in the founders; can she actually add meaningful value at the stage they are at. Sandstone, she says, was one of those rare companies where everything clicked within the first five or ten minutes. Having spent years in legal operations, she recognised the founders were solving challenges in-house teams had been talking about for a long time.
TLW: Your observation that legal AI companies are repositioning from firms to in-house teams is one of the more consequential shifts that few people are naming directly. What do you think happens to the founders and products that stay pointed at law firms. Do they eventually catch up on adoption, or does the incentive misalignment simply keep the ceiling where it is?
Mary: “Companies selling into firms are going to find real early success, because firms are under enormous pressure to demonstrate they’re doing something, and buying a tool is the easiest way to show that. But that adoption, as we’ve talked about, is fundamentally misaligned with the incentives underneath the business. Where I do see real opportunity is in anything that doesn’t touch the core practice of law: operational efficiencies, or work that’s traditionally non-billable and gets written off anyway. Those use cases will succeed inside firms because the incentives are actually there to support them.
Many vendors are pivoting to in-house because that’s where lawyers are genuinely leaning in to change how they deliver legal work, not just adopting a tool around the edges. It’s still too early to know who wins this. I’d love to be wrong because there’s real innovation happening inside firms right now. But I do think we’re approaching a ceiling on that side, if we haven’t hit it already, and the acceleration from here is going to come from in-house out of necessity.”
The junior lawyer question
If AI increasingly performs work that associates traditionally learned from, is the profession creating a training gap or a different kind of lawyer? Mary’s answer is unambiguous: it is a different kind of lawyer.
Historically, lawyers were trained first by law schools and then by large law firms. Neither, in her assessment, has adapted quickly enough. Law schools have not fundamentally changed how they educate future lawyers. Law firms have not fundamentally changed how they develop junior talent. So the question becomes: where will future lawyers acquire the requisite skills.
The answer is already visible. Young lawyers now have many more career options: technology companies, in-house teams, AI-native firms, legal engineering, careers that barely existed only a few years ago. The explosion of AI-native firms is creating entirely new career paths. Those organisations often have partner-heavy structures supported by technology, rather than large numbers of junior associates performing routine work.
That does not mean junior lawyers disappear. It means their role changes. Instead of spending years doing work technology can now perform, they will begin contributing at a much more strategic level much earlier in their careers. If they learn to leverage AI effectively, they can develop far more quickly than previous generations. This, in her account, is one of the most exciting opportunities AI creates.
Precedent, permission, and the message she wants to leave
At the end of the conversation, Mary was asked what one message she would most want readers to take away. Her answer arrived without hesitation.
People should feel empowered to create change themselves. Do not wait for somebody else. Ask more of your law firms, your legal tech providers, your organisations. And equally important, do not be afraid to experiment yourself. We need cultures where it is safe to try new ideas, test different approaches, and occasionally fail. That is how industries evolve.
Legal has traditionally been a profession built around precedent. People often wait to see what everyone else does before acting themselves. Mary hears the phrase “we’ll just fast-follow” constantly, and she really dislikes the mindset. If everyone waits for someone else to move first, progress becomes painfully slow. The legal industry, she says, needs more people who are prepared to experiment, take calculated risks, and lead rather than follow.
The final observation serves as a corrective to the loudest parts of the current conversation. A lot of people assumed early on that Harvey or Legora would dominate legal AI simply because they moved first. Many other companies have since emerged with completely different products and business models, and they have been incredibly successful too. The market is still in its very early stages, and nobody can honestly say today who the long-term winners will be. What is encouraging is seeing organisations become willing to make different choices instead of automatically following everyone else. That is how real innovation happens.
TLW: You’ve said the profession’s fast-follow instinct is one of the things holding it back. For a reader of this piece who works inside a firm or an in-house team and wants to be one of the people who leads rather than follows, what is the first concrete thing you would ask them to try this quarter?
Mary: “Pick one thing you have always assumed you had to do a certain way because that is how it has always been done, and go test a different approach. It does not need to be dramatic. If you are inside a firm, that might mean proposing a flat fee on a matter you would normally bill hourly, just to see what it teaches you about your own efficiency. If you are in-house, it might mean piloting a new technology on a real matter instead of a sandbox, with a clear way to measure whether it actually saved time or improved quality. The point is not the specific experiment. It is building the muscle of trying something before you know it will work, and being okay if it doesn’t. Legal has spent decades rewarding people for not making mistakes which has quietly trained everyone to wait and watch. That instinct is exactly why this profession loves the fast-follow mode in the first place… it’s safe. Instead, have the courage to go first. You do not need permission from your general counsel, your managing partner, or the market to run one experiment this quarter. Just do it.”
Where the conversation leaves the profession
There is a pattern to the message Mary delivers during a conversation. The technology is not the problem. The business model underneath it is.
Buying more tools will not fix a firm whose incentive system rewards the very behaviour the tools were designed to reduce. Announcing an AI strategy is not the same thing as building one. The parts of the profession that are truly changing are the parts that have done the harder, less visible work on their own incentives, their own operations, and their own willingness to move without waiting to be told they should.
What comes through is not doom but a considered urgency. Mary is clear about what is not working, and thoughtfully elucidates why the profession has not fixed it faster. The industry is difficult to change because the parts do not move at the same speed, not because the people inside it are unwilling. That framing is what makes her voice one you keep wanting to return to.
