The AI Answer for Boutique Law Firms Is Already in Plain Sight, But Are Firms Looking in the Right Place?

Key takeaways - Boutique law firms have many of the same confidentiality, supervision and governance responsibilities as much larger firms, but without the same budgets, teams or resources. - The most useful AI opportunities are not always the most obvious ones. Client intake, matter management and billing may deliver more value than focusing only on drafting and legal research. - Firms may already have much of what they need through Microsoft 365, Claude, ChatGPT or their existing practice management systems. - Buying a tool is not the same as adopting AI successfully. Training, clear policies and properly defined workflows are what make the difference. - AI may help boutique firms compete more effectively with larger practices, but their real advantage remains their relationships, judgement and specialist knowledge. We all know that boutique law firms are being approached constantly by AI vendors promising faster work, lower costs and greater efficiency. What is much less clear is what genuinely works for a smaller firm, what is worth investing in and what is simply unnecessary complexity. That is why I wanted to speak to Rok, founder of MPL. His background spans regulated enterprise environments, startups and small businesses, which gives him a particularly useful perspective on the realities boutique law firms now face. Rok works with legal teams across the UK, Europe, the United States and Australia, helping them understand where AI can make a measurable and practical difference. His approach is vendor-agnostic and starts with the systems and processes a firm already has, rather than assuming the answer is yet another piece of software. In this conversation with Rok, we discuss where law firms are wasting time, why operational improvements are often overlooked and what firms should be doing now if they want to adopt AI properly and avoid any costly software purchases that are not the best fit for their firm. **1.** **You came into legal AI from enterprise-scale regulated environments. What specifically did that background show you about boutique law firms' AI readiness that a pure legaltech background would have missed?** I think it’s not just the enterprise, it’s the privilege of having worked on both ends. I spent years in enterprise, regulated environments, and then a good stretch working with small businesses and startups. And that combination shaped my approach to helping legal teams with AI. Enterprise taught me the obligations, and what real governance machinery looks like when it's done properly - confidentiality, supervision, access controls, audit trails, the whole apparatus. The small business and startup side taught me the other half: the economics and the fast pace. Every move has to pay for itself fast, there's no transformation budget waiting in the wings, and you get good at squeezing what you already have before buying anything new. A lot of law firms I work with sit right where those two worlds meet. Their reality carries enterprise-grade duties - confidentiality, privilege, supervision over the work - but runs on small-business resources and speed. I believe it’s important to be able to tell which piece of enterprise discipline is genuinely load bearing for a small firm and which is overhead they can leave out. I don’t think boutique firms should run like enterprises, and they can’t run like startups either. The answer sits in between. **2. MPL positions itself as vendor-independent. In practice, how often does that independence create friction with firms that have already committed to a vendor before they reach you?** This actually removes a lot of friction. Being vendor-independent means my recommendations aren't motivated by a commercial agreement with any one vendor. It leaves MPL free to find the best solution for the client, whether that's a platform they already have or any product on the market they'd genuinely benefit from. A firm that's already committed to a vendor is a good example. We're not there to displace their tool since we have no commercial incentive to. So we can simply focus on helping them get more out of what they already bought, or give them a straight read on whether it fits, with no conflict of interest pulling the answer one way. **3. You work across four continents. Are the barriers boutique firms face genuinely consistent globally, or is that a simplification — and where do you see the sharpest regional divergence?** I think it's a bit of both. Most of my clients are from the EU, UK, US and Australia, so I can speak much less to the Asian market. The core pressures are consistent, the ones that come from being small, regulated, and in professional services. No transformation budget, the same workflows eating non-billable time, an old core system setting the ceiling. There are a lot of regulatory differences. The US is running on a lighter, fragmented set of rules, so a firm that decides to move has little in its way. Europe carries the EU AI Act, GDPR, and legal-use AI possibly treated as high-risk, which means far more has to be cleared before acting. I think part of the appetite to invest follows from that: where there's less to clear, firms just move faster. The other split is pricing - as an example, a lot of UK based firms are already used to working with alternative fees while I see the US leaning more on the billable hour. **4. You publish to close to 5,000 people in legal AI. What's the question your audience asks most that the legal AI vendor market is still not answering well?** I'm not sure about the vendor market not answering it well, but the question I hear most is some version of: what's actually working inside a firm like mine right now? I think it’s pretty hard to get a straight answer to that because it feels like every answer is directed by the selling incentive. Which makes it hard to get to the honest version of the answer, what actually stuck and what quietly didn't. That’s why I think peer to peer conversations are so important right now, lawyers across the jurisdictions comparing notes with one another. **5. The boutique firm market is being targeted by every legal AI vendor right now. What does the sales narrative most commonly get wrong about how these firms actually operate?** Yes, that segment of the market is being targeted hard after years of being underserved, but being aimed at isn't the same as being understood. A lot of the pitches I see are still built for a firm that operates nothing like a boutique. The pitch assumes a transformation budget, a procurement process, a dedicated innovation or IT function, and one high-volume workflow worth a premium tool. A boutique has none of that. The person evaluating is usually a partner doing it on evenings and weekends and every spend has to pay for itself fast. It also casts them as behind, needing to catch up before an AI-native firm eats them. That gets the moat backwards, because a boutique's real asset is the trust, relationships, and judgment built over years, the part the technology can't replicate. So some of that fear pitch misreads where their advantage actually is. **6. Where do you see the clearest gap between what boutique firms think they need from AI and what would actually move the needle on their capacity or profitability?** I see a real gap between the visible legal work and the invisible operational work. There’s a lot of focus on wanting AI for the lawyerly tasks - drafting, legal research, the things that look like doing law. That's fair, it's what the market puts in front of them. But what moves the needle is the unglamorous operational layer around it, such as intake, matter management, billing - the non-billable time that quietly eats the week. There's also a catch on profitability. Under the billable hour, doing the legal work faster can actually cost you. So the cleaner way is usually freeing the partner's time for the things that bring work in, business development, and tightening how billing gets captured. **7. You frame the work around three core processes — client intake, matter management, and billing, with a Buy, Stitch, or Build decision at each. What's the most common mistake firms make when they try to assess this themselves before bringing in outside help?** If I would name one, it would probably be jumping to Buy before squeezing what they already have. Firms assume their current setup can’t do the thing, so they go shopping, when a lot of the capability is already sitting in M365, Claude, or the PMS they pay for. The squeeze step gets skipped because “we need a tool” feels like progress, and testing your own ceiling doesn’t. With a lot of self assessments I am also seeing that firms assess against a blurry version of their own process. The buy, stitch, or build decision is only as good as the map of the workflow underneath it - and a firm's own view of its process is the inside view. It's optimized for the people who run it, never really designed to be looked at. The work is built for the operators, so it's genuinely hard to break it down into the steps where a machine could actually help with some of them and so you end up making a sound decision on a fuzzy map. **8. One of your client testimonials describes getting clarity to move towards fixed pricing as a consequence of the AI process work. Is that a recurring outcome, or an outlier and what does it tell us about where the real value lands for boutique firms?** I see it more often than one would think, especially with clients who are fully bought into AI, who treat it as an investment and are willing to do the work it takes. So I wouldn’t call it an outlier. And I think it makes a lot of sense, because when you apply AI well, you get better at predicting how much resource a given task will actually take. And that predictability is what feeds the pricing, because fixed pricing really depends on knowing how much effort a matter will actually take. I don’t see the value being in the automation itself, but in the self-knowledge. Mapping out the processes to be able to identify good uses for AI forces a firm to see its own operation through a different lens - where the hours go, what’s effective, what isn't. Fixed pricing is just one of the positive side-effects of clarity showing up, because you can't price fixed what you can't predict, and you can't predict what you've never mapped. So the headline outcome people notice is "we moved to fixed fees," but the real asset underneath is that the firm finally understands its own work and starts improving it. **9. You complete engagements in five to eight weeks where others quote four to six months. What specifically gets cut or compressed, and what are the honest trade-offs of moving at that pace?** I would widen this a little, because we're not a full automation shop. Most of what we do is training and advisory and that's the first step for almost any firm. The implementation work is one part of what we offer and it’s not for everyone. On the timeline itself, it’s an average for the squeeze approach and the reason it can move that fast is the method. Going AI-native for an existing boutique law firm doesn’t mean rebuilding everything they have from scratch but rather finding the parts that eat the most time, mostly redesigning or adjusting that process, and wrapping the technology around it using the stack they already have (usually Microsoft + Claude or ChatGPT). Fully bespoke solutions, where I believe the longer timelines are coming from, do have their place, but that’s not where we believe most boutiques should start. Hammers are great, just not for every nail. On what gets compressed, it’s the bespoke, long builds and the enterprise-level overhead you get when building and deploying them in a regulated profession due to all the compliance requirements. The trade-off is that the pace only holds under certain conditions. It assumes someone at the firm can commit a significant portion of their time in those weeks and be properly hands-on with us. If they can’t, it stretches. And if a workflow genuinely needs a deep bespoke build as it hits the ceiling of what’s possible with the tools the firm has, this fast model isn’t the right one and we would scope it separately. **10. Which AI capabilities are boutique firms most likely to implement in the next 12 months that will genuinely change how they deliver work and which ones are they currently over-investing attention in?** I think the pace right now makes it almost impossible to name specific capabilities that far out. The adoption among legal professionals went from 31% to 69% in a single year, and products and features are shipping faster than anyone can track. I also think that to change how work gets delivered, we don’t need new capabilities because the technology is already there. And the way to get there is getting the teams trained on the tools they have, and making sure that AI is embedded into the firm’s actual workflows, with a human in the loop. The basics. And most firms haven’t done them. In that same report, 54% said their firm offers no AI training and has no plans to. So for the next 12 months, the unglamorous basics are what actually move the needle for most boutiques. Where I see attention being over-invested is the fully autonomous agent, the "agentic" moonshot, and chasing the latest and the greatest. It's the loud part of the conversation, but it's not where the value is currently being proven. The boring, embedded, supervised use does more for a firm than another step toward autonomy. **11. You describe yourself as Microsoft-first and Claude where it earns its place. What does Claude need to do better before it earns a larger role in your standard implementation stack?** I wouldn’t really call it Microsoft-first. What I say is squeeze what you already have. And for most boutiques that happens to be Microsoft, so it comes up first by circumstance. If a firm is already on Claude or ChatGPT, we start there just as readily. We don’t really have an implementation stack we’re selling, I think that would put us in a position where we would be commercially motivated to push for a solution as opposed to matching our clients with what’s best based on their reality. We do setups and training across Claude, ChatGPT, Microsoft Copilot and Copilot Cowork, working with the IT side, the lawyers as users, and AI champions with the train-the-trainer programme, and we adapt to whatever PMS or DMS the firm already runs. In practice most boutiques already have Microsoft 365 plus something, and more and more that something is Claude - used well, it is a strong platform for legal work that integrates cleanly with M365, so we usually end up working within both. As for what Claude needs to do better to earn a bigger role, honestly, not much. Right now, before you get to the vertical legal-specific tools, I think Claude Cowork is the strongest general AI platform out there for legal work, and far enough ahead that it will take the others a while to catch up. So it's not really waiting to earn its place. **12. What's the failure mode you see most often when a boutique firm attempts an AI transformation without specialist support — and how recoverable is it typically?** I often see firms treating AI as a purchase rather than a change in how the work actually gets done. They buy the tool, the team pokes at it like a chatbot, and the matter folders, the skills, the real workflows all stay untouched. The outputs come back generic, so people quietly drift off, and the one person who championed it ends up burning out doing it in the evenings. The good news is it's usually very recoverable, because nothing's actually broken. The tools are still there, the spend is already made - what stalled is adoption, not the infrastructure. So the fix tends to be cheap: train the team, wire it into one real workflow, encode a bit of how the firm actually works. The problematic part is belief, because the people who got that flat first experience now think "we tried AI, it didn't work," and that's harder to undo than the setup. The longer it sits there as disappointing shelfware, the harder that is to shift. **13. The boutique firm market is structurally different from BigLaw. Do you see AI widening that gap in the boutique firm's favour, narrowing it, or accelerating consolidation within the boutique segment itself?** My take on this is that AI is closing the gap with BigLaw on a lot of work, and it’s tilting toward the smaller firm. A boutique can now deliver things that used to need a much bigger team behind them, and it can move on them faster, because there are no legacy systems and no committees slowing the decision down. Clients are feeling fee pressure too, so more of the work that doesn’t justify a BigLaw rate is already coming down to boutiques. It doesn’t close at the very top though. The most complex, bet-the-company work still sits with the big firms, and AI doesn’t hand that to a boutique. So I think the gap narrows on the solid middle. The bigger shift, as I see, is happening inside the boutique world itself. The technology is something everyone can get now, so the tech on its own stops being the advantage. What lasts is the moat a boutique already has - the trust, the relationships, the judgment built over years - and whether they encode it so AI runs on it. The firms that do that are already pulling ahead, and they start taking work that used to go up the chain. The ones that just buy a tool without training their team to actually use it fall behind their peers, and over time those are the ones that I think will get squeezed the most. I think the consolidation we see happening is large-firm M&A and AI-platform consolidation. For boutiques, AI mostly widens the gap between the ones who use it effectively and the ones who don’t. **14. If you were advising a managing partner today who is 12 months behind their peer firms on AI adoption, what's the single most important thing they should do in the next 30 days and what should they stop doing immediately?** Being twelve months behind on the tools is not such a big deal. The tools are a click away, and they’re far better now than they were a year ago. So they're not really behind on technology, they’re behind on *using* it. And that closes faster than it sounds, because they’re starting from today’s much stronger starting point, not walking the same path the early adopters did. So the single most important thing in the next 30 days is to get the team trained on what they already have. Almost every firm already has something, as I mentioned earlier, so nobody needs to buy anything for this. It's about getting people to know how to use AI well - its quirks, limitations and caveats - how to handle client data properly through it, and how to check its output before relying on it. Plus a simple internal policy so everyone knows what's okay and what isn't. And I would treat that policy as a living document. Early on it’s rough and it changes a lot, and that’s fine, you tighten it as you learn. It also helps a lot if the partner gets hands-on themselves, even just using AI on their own work for a few weeks. You can’t really lead something you’ve never felt. What I would say to stop right away is the shopping and the waiting. The lining up of demos, the comparing of platforms, the "let's hold off until the market settles". That instinct is usually what puts them behind in the first place. You don't buy your way out of being behind, and you don’t wait your way out. You get your people using what’s in front of them. If you want your own conversation with Rok to learn more about where your AI efficiencies already exist, book a call with him here. For further discovery see Rok's video explainer 'How to Roll Out AI in Your Law Firm: A Practical Roadmap'