Legal AI Year in Review 2025 with 2026 RAS Presentation

All right, so today we're going to talk about the legal year in review, 2025, and what leaders must do heading into 2026, which is what we're calling the accountability era. So we just left an era of great. 2025 was a watershed year in, in the space of AI, and now we're heading into 2026. So in 2025, the big shift was it rather ended the era of experimentation with AI. And because as the year went along, AI began becoming deployed to a large extent, it also began the focus on accountability as some of the abilities of AI got to a much more progressed level. So we went from coming soon coming attractions to real consequences with immediate compliance requirements. As some uses of AI blew up on folks in the legal world and evidence standards changed, courts started treating AI outputs introduced as discoverable evidence in litigation proceedings. And finally, autonomous action tools evolved from initially chat assistants using copilot, maybe to chat with online, to independent agents now executing tasks in the background. And the question is, you know, are they being monitored? And in many cases they're not. So the landscape shifted from theoretical risk to operational accountability. So taking or adopting a wait and see attitude is no longer a viable option. So let's take a look at 2025, kind of in a timeline perspective. So in Q1, we had the regulatory clock kind of start. The EU AI act enforcement begins with immediate audit requirements. And then in Q2, vendor risk becomes a legal risk. Better vetting of vendors becomes paramount as discovery obligations transform vendor relationships into potential liability exposures. In Q3, we saw the rise of agents, N8N make and a few others started to seep their way into normal business. And AI transitions from simply answering questions to now executing independent actions. And of course the residual question is anybody monitoring this? And in Q4, the settlement era, copyright and training data disputes establish industry precedents. And there are a couple of landmark copyright cases involving OpenAI and anthropic that really come to the forefront as far as that goes. So each quarter brought its own fundamental change on how we need to approach AI deployment, governance and risk management. So just quickly Q1 regulatory the clock starts ticking. As we mentioned, the EU AI act enforces the black box. Time is over, transparency is in. What changed was that leaders could no longer just adopt a kind of a wait and see strategy and see how things are going to pan out for everybody else. International firms faced immediate audit requirements as a consequence of this act, and transparency mandates fundamentally changed the vendor selection criteria. What leaders ended up doing was early preventative audits. Now prevented operational paralysis. Compliance work cost significantly less than reputational damage. That was an eye opener, but a logical one. Governance moved from policy later to policy now. We have to stem the bleeding now. We have to address these issues right now. A ban on prohibitive AI act practices took effect, forcing immediate organizational responses across the industry. A set of standards. What you can and can't do. Sorry. So in Q2, vendor risk becomes legal risk. And this, this. And here's the typical exchange prompt output log discovery. The critical governance compliance is. There must be a human in the loop. And that isn't guaranteeing a shield by itself. In the legal space. There was a term, lazy associates, that kind of cropped up. These are associates that used a large language model, a consumer product like ChatGPT or others, to do legal research. Even though they were doing that, they weren't verifying the citations and the sources. And the partner is usually the stopgap. That's the last line of defense. Wasn't catching it because they trusted the attorney or they had some prior relationship with the attorney to know they were trustworthy. So they let it through. And boy, there were consequences. If vendor logs are preserved confidentially, confidentiality exposure grows exponentially. So logging is a good thing. Treat prompts as privileged work product from day one. No exceptions. Don't enter things into a prompt that you wouldn't enter into a conversation. Right. If you're on the phone talking with someone or a client, there's attorney client privilege. There's a whole host of legal standards that go into place what you can and cannot say. It's the same thing with a prompt. And build audit trails that you can defend in court. So if you are using AI to introduce content, make sure that you have the citations, a complete audit trail of where you got this information. It has to be more than just for internal purposes. And then we talked about again, some of the vendor activities going on. A couple of landmark cases. The New York Times versus OpenAI lawsuit. The Times sued OpenAI for basically taking its entire catalog of back articles and issues without paying any royalties. But the output, the consequence of that whole scenario was that the courts, the federal courts, mandated that OpenAI was required to retain all output logs, which established a new discovery standard and also probably caused a little bit of anxiety for folks who realized that OpenAI, while they're not learning from this data, they're still collecting it. And is there a trust factor involved? There sure was in my case. Q3 saw the rise of agents N8N make and few others that started to infiltrate into the workforce in general. August 2nd we had the EU transparency obligations. GPAI forced new disclosure requirements. In September, Microsoft facilitator terms and Copilot moved towards workflow management working in tandem. In Q3, agents began assigning tasks or number three. Independent workflow management became a reality. And how the ROI really changed was stop measuring time saved and start measuring task completion. And where this has particular impact in the legal profession is most law firms even today are still practicing time and materials as it's basically its operating methodology toward its clients as opposed to fixed pricing, which became something also known as alternative pricing about 10 or 15 years ago when something called legal project management started out. So AI is going to drive a conversion from the traditional time and materials model, which doesn't really serve clients, is very inefficient. But law firms have traditionally had a disincentive to move away from that because their life is built around the billable hour and the more hours you bill, the more profit you make, et cetera, which is antithetical to an efficiency model when you think about it. So AI is going to drive firms in that direction and largely because clients will insist on it. But legal project management for example, came out of internal internal counsel from large corporations who were getting who were tired of paying time and materials for predictable matters and then began to demand from their external council to create fixed price scenarios that they could in turn bid out. So this is going to only going to continue more and more into the future. And law firms that don't look very candidly at migrating from the traditional time and materials to fixed pricing model may be left in the dust. So just some numbers. 18% admin reduction with structured data environments through the use of agents. 85% task completion. So more stuff is getting done. But Here's a concern. 40% agent failure rate. So these agents are not infallible by no stretch. They can run a process 19 times in a row and the 20th time it fails again. Another reason that we need eyeballs on this human in the loop and audit trails. Success required infrastructural investment, not just tool investment. So garbage in, chaos out. So Q4 of 2025 is now known as the settlement era where training data risk really got a price tag. And The Barth Barts vs. Anthropic Again another settlement were anthropic or Claude basically appropriated a million books without paying any royalty fees on it. And there was a copyright settlement that they ended up paying of a billion and a half dollars. So there are two Paths forward use of license data which provides for transports, transparent sourcing, indemnification protection and insurable risk profile. That's your Nexus, Lexis, Westlaw, Bloomberg Law type of sources. And then there's scraped content, which is generally done by AI agents or AI affected applications where you have an unknown aspect, unknown providence, litigation exposure. Should the information not prove to be reliable, of course it's uninsurable. So the key message here is cheap tools built on simply scraping content may be uninsurable or collapse ultimately under mounting liability that ends up in equal consequences or financial consequences. So what 2025 taught us just real quick verification is a real service and a necessity and it's our obligation. Output generation is cheap. Anyone can use any of the gen AI products and get a response. Verification is expensive and labor intensive because there's a human component to it. We can't verify AI with AI budget accordingly, the cost of human oversight is your real true operational expense. Data hygiene is also critical. OpenAI has maxed out on all of its unique information, so right now it's training on itself. And it's estimated that 30% of OpenAI's large language model content is merely other people having conversations about existing data which as you can imagine, pollutes the overall data assembly or data lake or data ocean or whatever you want to call it. Agents can't fix broken processes or messy document management if the data is subpar. Garbage in, garbage out. Clean data infrastructure is absolutely the recipe for successful deployment of agents. The risky middle disappears. Choose enterprise grade tools with proper governance or nothing. The middle tier of good enough solutions that are untested. And this applies even to existing solutions out there in the technology space. Legal tech is an incredibly broad category of solutions, but be wary of those that haven't really put forth enough proper governance that they may carry indeed an unacceptable risk profile with them. So we're at a crossroads. So, questions, where's your firm to answer me? The legal industry itself is at a technical crossroads where progress is striking hard in the favor of AI. AI is becoming integral to the legal profession. The reality is AI isn't just futuristic anymore, it's actually part of it should be part of your operating system for survival. Because those who don't use AI have a far less chance of surviving in the legal space. The competitive gap is widening between the traditional and the tech forward firms and it's only going to get more, only going to get worse as AI takes off in 2026. Traditional workflows equal time famine that block growth. If you're not automating those traditional, largely manual workflows, or if you have disparate systems that don't talk to one another, you're really putting yourself in a competitive disadvantage because your competitors are already scaling while others get stuck in these kind of weeds. And again, the firms that fail to adopt AI with rigorous guardrails aren't just falling behind, they're risking their very existence in what's becoming an efficiency driven market. So the real cost of waiting and doing nothing is a competitive reality. Right? When a competitor can sift through thousands of documents while your team takes weeks thumbing through manual document review or using software for that's outdated and legacy, this differential becomes impossible to ignore or manage beyond the bottom line. Your professional reputation is at risk. Your clients are going to be expecting that you're deploying technology, the latest advents of technology, and they're not going to pay for outdated methods. Certainly. Again, the conversation about time and materials. And this isn't just about business, it's about the integrity of service. You as a law firm with your privileged clients have almost a rite of passage to ensure that you're providing the best tools to do to serve them the best, most efficient and economical and effective way. And if you're not, deploying AI and technology that shines a light on that integrity requires showing up with the best tools available to serve those who trust us with their very lives and their businesses and their well being. So the looming threat modern clients today expect technology sophistication that matches their own. We live in an era where most people are familiar with AI and are using AI in their business. This is not too dissimilar. In 2008, when Steve Jobs thrust the iPhone on us, consumers became so consumed with that superior model that eventually the iPhone permeated its way or found its way into the business side and effectively replaced BlackBerry, which had been the standard for years, but also at the same time caused a lot of security issues as IT departments were forced to try to figure that out. This is interesting. The human oversight gap. Why Guardrails matter more than speed. So it's not the speed to information, which is what AI tantalizes us with. It's take a step back and realize no, the accuracy of the response is critical. 67% of legal professionals report having to correct AI outputs. That's two out of three. That's interesting and scary. 21% of law firms today trust without review. They're using AI sources for drafted work, legal briefs, et cetera, without human review. And that's almost a recipe for disaster because yeah, AI without proper governance, that's exactly what you're looking at. So guardrail imperative. Using AI without the proper oversight is that recipe for disaster. Governance must be the core of your decision. Governance can be anything from what are your habits? Using AI for prompting what are your habits in reviewing output that comes from AI? Validating citations. Every action serves the truth and protects the vulnerable, your clients. Here's a key principle. You can't serve two masters. You must choose between the comfort of old ways and the responsibility of the new ways. And a lot of us find us in that space today. AI is the biggest disruptor in my life. I'm 67. Imagine what it's doing to businesses. Governance isn't optional. It is the foundation of trust with your clients and with yourself. So we're going to talk about the justify the justice multiplier that AI is also can be a phenomenal tool and I'm looking into this as well in the pro bono and personal injury space. So the challenge in those spaces is personal injury law and pro bono services. The scarcity of time isn't just a business issue. It's a barrier to justice, often resulting in scenarios where their clients will plea bargain simply to get an outcome so that in the criminal law case, they're not sitting in a holding cell in perpetuity. So innocent people may plea bargain and plead guilty because of the lack of resources and the lack of ability to really come through and against better financed, better equipped, better resourced opposing counsel. So research shows legal aid organizations like the Legal Aid Society are actually adopting generative AI at twice the rate of the wider legal profession. Why? Out of necessity, they have to. They know they're outgunned and outmanned and out resourced. AI can be the great equalizer in this space, which is why it's particularly exciting for me because the notion of one innocent person going to trial and being absolved of the charges versus an abundance being recorded chronology which is huge in the PI space, personal injury space. And it can reclaim 15 to 20 hours a week for high value work through this automation and return to focus where human souls attorneys can serve instead of paperwork and being overwhelmed protects fairness in a system that often favors the powerful. As I said, the higher the resources and whatnot, the more likely that they'll use that as a cudgel to pledge bargain plea bargain. So AI is a justice multiplier because it allows attorneys to reclaim authenticity by automating the grunt work serving the marginalized and the Voiceless who are often silenced out of the system or again into plea bargaining arrangements and reclaiming what matters. So the power of presence in a tech driven world. The fear is that AI will replace the human heart of the law. The truth is AI is a tool meant to empower, not replace the creative brilliance of dedicated attorneys, especially trial lawyers. They are absolutely brilliant. They're like actors. They can take a cacophony of data information and in this swirl, most of which they do in their heads, they're able to translate that and play an acting rule to try a case successfully. That's an incredible skill and talent that with AI providing predictive analysis and provided additional insights into co counsel who the who the judge is on that particular trial can really help them level up in a big way. So what AI enables again it automates the mundane, creates space for those sacred moments, whether it's in your personal life or otherwise, freeing you to spend time with family and in your faith. And eliminates not enough thinking. With AI, I use it all the time. Tell me what my gaps are, tell me what my blind spots are. It's really good at shining a light on that. Your purpose is to be faithful. And being faithful today means mastering the tools of your trade to lift others higher. That's just a personal statement, but I think we all have an obligation to use AI to raise the standards for those that we serve. Imagine a practice where you're no longer burdened by not enough time or resources or money. But by integrating strategic systems like AI, you can create space for those moments now that matter. You can be less under, less stressed. Citation validation in action. I'm going to talk briefly about my product, which is called Trial Lift, which was designed to help trial lawyers basically evaluate their current body of evidence, court preparation, testimony, all of those things, even medical information, to see if there are any gaps. Is this testimony weaker than it needs to be or is it? Or is something missing? In addition, if lawyers do go outside of, they may not be able to subscribe to the typical subscription services like Nexis, Lexis, Westlaw, et cetera. If they do use genai such as ChatGPT or others, my product will actually validate those citations. And that's becoming a necessity more and more in the space. There are over 700 cases to date where false citations were introduced by attorneys in court trials, resulting in fines, disbarment, dismissal of cases. So there are real catastrophic effects that happen without citation validation. And you're seeing more and more law firms migrating towards what are called rag Systems, which is really. Valley did a wonderful program on this about a month or so ago. It's essentially a system that sits on top of your gen AI and it narrows the focus down on, on what body of documents information that it will query based on the nature of the prompt. So it basically parks the waters, so to speak, and just incredibly intensifies on just what's relevant to the actual prompting. So again, hallucination detection is so critical in determining before those documents leave your firm or God forbid, end up in court. Audit trails are so important if you are going to be using AI to support a legal trial. Where did that information come from? What are the, where are the citations? Where is all that? Because again, in the EU those audit trails are mandatory if you're going to use AI to introduce information in court. So this is what governance really looks like in practice. It's not just a bunch of policy documents, but there are tools in place that protect your clients and your reputation. And you're going to see more and more of this in the legal tech space as these, especially with the traditional vendors who are playing catch up because their, their products basically predate AI and it's very difficult to immerse AI in a legacy product. It's much easier to start from the ground up with AI and infuse governance and moving forward. So the newer startups and the newer products I believe will have an advantage in that sense. And that's what we're trying to do with trial left little plug there. So the complete solution ecosystem is where automation meets ironclad security. So you have these custom AI frameworks which you can build through agents and whatever the is customized to your particular values and practice areas, ensuring that these tools serve your specific needs. And that could look like simply looking at all of your business processes, where the bottlenecks are, where the manual processes, where the time sucks and then building reliable, transparent ages and processes, agents and processes and governance around that citation validation. Again, tools like what we have and others that validates the total accuracy and prevents hallucinated material from entering into trial and resulting in malpractice. Because that's coming. AI governance is an ethics first implementation, proper oversight, accountability measures in place, audit trails in place and end to end infosec, this is another key piece, is that it starts with your infosec security, then there's cybersecurity and then there's AI governance and guardrails. These are not separate silos. They're all part of an overarching strategy in your organization to offer Protection for data as well as your business. Technology without security is a house built on sand. AI can be ethical, it can be secure, but only when properly governed. It can be the launchpad for your greatest impact when you have that in place. So again, budget for fewer, safer tools, not many cheap ones. I know in the consumer space that's what we're all doing, right? We're all using a myriad of large language models. We're all using a myriad of editing, image generating, video creation, voice generating, all these different things which are fine in our perhaps public or private life. But we need to be far more screen involve far more scrutiny when we bring those into the workforce and the workplace. So what does 2026 look like? What is our forecast? We're going to see work relocation as elements that were previously done by humans that are redundant, time consuming, manual. You're going to shift to agents. Agents are going to be going out there looking for case material that's relevant to whatever you're working on right now. Humans move now more to editing and strategy and judgmental calls requiring true expertise. So it's imagine us more as we're drivers behind the vehicle, which is AI. If we drive that car into a guardrail, no pun intended, that's on us, that's not on the AI. So the era of blaming the AI first of all, it was never appropriate. But even now, more standard delivery discovery requests for AI interaction logs becomes a standard practice in litigation. You have to provide total visibility if you're in law using AI premium tools. We mentioned clean data tools cost more than the cost more, but they become the insurable standard. You must have clean data. You must have clean data that comes back from additional sources. So budget for quality, not quantity. And in the document management space, agents will be integrating inside these systems, looking for documents through elevated case management software, for example, that peer into document management systems. So watch for NetDocuments and iManager, arguably the two largest document management providers in legal. So watch their roadmap, see what they're doing. How are they using AI? What are their governance standards? Be very mindful of that because the DMS is the backbone of the document collection in a law firm. But it also can be re can be ripe with all kinds of issues if not managed properly. You know, multiple files, duplicate files, et cetera, the Brussels effect. So the EU rules go global as multinational firms and or law firms that operate in both the EU and the us you don't want to have two standards, so you want to amalgamate that or homogenize that into the highest common denominator for compliance. So even though the EU is first and the US doesn't really have from a federal basis anyway standards that are as detailed, it's coming and what happens across the pond really, really should be impacting impacting what's happening on this side. So 2026, where are we headed? How build beats wow so operational integrity now over the hype that was generated in 26. Then here are some strategic imperatives. Build for workflows Stop piloting chatbots Build agents for real workflows like intake, discovery and billing automation Defensible governance if you can't produce auto trails, don't deploy the solution. Documentation is non negotiable. It is in the EU and it should become the standard in the us. Vendor diligence demand training data disclosure if they're using a large language model underlying in their product and indemnification commitments must be in every contract to protect your rights. The focus shifts from experimentation to execution in 2026 proof of concept gives way to proof of governance so here's a quick 30, 60, 90 day plan as I wrap up what to do next without pag, of course. So days zero to 30 is a foundational time. Take a look at your existing auditing tools or audit your existing tools rather. You know which ones meet the standard and the litmus test of necessary governance Eliminate shadow AI usage what that is Many people operate in a Microsoft shop so you're using copilot. Everyone knows Copilot is not a is a substandard large language model compared to ChatGPT, McLaun, Gemini, Grok and others. So the tendency is employees may get frustrated by its limitations and using their own tablet, smartphone or whatever. Use consumerized gen AI products not being mindful of what the ramifications may be certainly from a governance standpoint. So shadow AI has become a fairly significant issue. So you must monitor that. You must have policies in place or use tools that can monitor AI use, although that becomes difficult if they're not they're outside of your network or your infrastructure but clearly government guidelines with consequences. If people use shadow AI and introduce content that ultimately incurs hallucinations and whatnot define clear prompt and data handling policies. That's super important and that's part of what we call commonly guardrails. And again don't introduce personal information client information if HIPAA information ever in a prompt and on the flip side when you get that response back from the large language model if you're using a genai LLM you must certify for accuracy and there are ways you can just do that in prompting what I call the Prime Directive that I use where I tell LLM never hallucinate, never fabricate. Don't exaggerate, don't summarize, just provide the facts, please. Of course it doesn't necessarily adhere to that. So even if I do that, I never take that for granted. So in days 31 to 60 is all about protection. Develop your vendor diligence scripts when you're going out there and assessing bringing in vendors to your environment for various purposes. Establishing logging standards again to meet the EU standards and what is a practical standard. Implement audit trail requirements firm wide so that everyone's on the same page. And then lastly day 6190 it's about execution. Pilot 1 agentic workflow with structured data. Monitor it very carefully, measure the task completion rate and and document any lessons learned from those times where the agent may fail and requires additional tweaking whatnot. This phased approach balances urgency and sustainability. Each phase builds on its previous as a foundation. So lastly, governance is a stewardship. Build it before you scale the agents. If you can't produce a log of what AI did and why, don't use that product. Auditability is not optional, it is mandatory. So your next move is get over the fear of AI and partner with somebody who can hold your hand and walk you through reliable, secure and accurate solutions. Enable you and assist you in deploying that throughout your enterprise. The investment principle Budget for fewer, safer tools, not many cheap ones. Quality and governance do cost money, but liability costs more. Audit your workflow. Identify one manual process draining your energy. Verify your security access if your IT can support advanced it or do you need to bring in other experts? Engage the experts. Find trusted advisors who will understand the government's first implementation. Not changing, not chasing shiny objects, chasing security. The future belongs to the accountable. Some final thoughts. The version of your firm that your clients and family deserve is waiting on the other side of this transition. The path forward from frustration to peace, from scarcity to abundance, and from paralysis to purposeful action. So the future belongs to those who govern responsibly today. Sustainable AI adoption requires accountability first, scale second. So thank you for the time to share this with you. I guess bally I'll throw it back to you if there are any questions, comments and platitudes. Not the last ones of course.

Thank you so much Richard. Thank you. Thank you so very much. And there was so much you've provided us today with and I'M sure we'll open the floor for questions that people may have. And the thing to be thinking about, even if you're not in the legal space, a lot of what Richard spoke about is applicable to us in our businesses. So looking at how we. We start working and tracking what we're doing. So. Yeah, so I just wanted to put that out there and thank you, Richard. People are going to be asking questions and just this morning, actually, I was speaking with one of my mentors who's in. Based in Brisbane and she delivered one of the trainings and she was saying that she had a lawyer who felt frustrated because the company wasn't. They weren't innovating, they weren't moving forward. So he could feel the frustration because they were just, they were just not thinking about any of the technology. So what, what happened was he then moved elsewhere. But not only that, he took 10 of the people with him. Not only did he move on his own, he took the whole team with him because they were open and the company, that all stands still. And this is a case that we just spoke about this morning. They will, they'll cease to exist. They will no longer survive with this outdated way of working for people who hear, I know that you won't do that, but I know that a lot of firms, if they're holding off doing this, they will. We're seeing a lot more research coming through. That's going to be the reality of it. It's a bit like we've seen this in a lot of innovations where we've seen this with companies like Blockbuster, we've seen this, we've seen many examples of that where they laughed at Netflix and that. And we saw what happened there. So I want to put it out to everyone and thank you so much. There was so much information. Most likely watch this again if you've missed some of it. And, and yeah, that they won't last even they can hold on to that. We know why that happens. They're just. People take a lot longer. They do take longer to move. It's because they don't know. They haven't got the expertise that they don't know how to get going or they'll have one tool and they think that's the way forward. What a question here. Yeah, yeah.

And in particular in the legal space, they tend to be, having served them for over 25 years, technologically adverse. And again, a lot of that is they have an older business model in, again in the whole time and materials space and legal project management tried to Move them out of that. But it's difficult because when the billable hour is directly tied to profit, you have a disincentive to be efficient. But AI is going to force firms to confront that and those that don't or will be left behind is what we're hearing too, with all the. With AI advances that are going to be impacting those who don't use it. That's critical.

A question from Ivan. So Ivan, over to you. And then I know Tom posed a question too, so we'll.

Sure. Tom was peppering me with questions.

Let's go to Ivan first, then we'll come to full circle. Back to Tom in a moment. Okay, Ivan.

Okay, cool. Thanks, Bali again. Thank you, Richard. Really enlightening topic actually, and great presentation.

Thank you.

I've got a million notes here. But one thing that strikes me is I was on a call with a chap called Alex Osterwalder who's from Strategizer in Switzerland, and they were talking about business strategy and business models. And I was interested in your point about the legal firms because they're still on the old model and if they don't change, they're going to be left high and dry. Because, as you so rightly put it, the client's expectation today is that you use the technology. And because you use the technology, I expect you not to charge me this big number, but this slightly reduced number. And that then poses a problem for the business because now all of a sudden you've got all these extra heads and you're going to have to either find new business or lose a couple of heads, which is a knock on problem later on. And then just to exacerbate that, I was watching a talk on the Diary of a CEO and the chap being interviewed was Roman Yaplospi. Happy to pop it in the chat if you want, but really interesting conversation between the two. And his concern is about the security behind AI and AI running away with itself and us not being able to stem the flow of what it's doing. I'd be interested in your view on. It's moved so quickly. Just your timeline just, just shows it's exponential. Right. And at some point the horse is going to run away from us. Can you see how we can stay on the horse?

Yeah.

Ray Kurzweil talks about singularity, which is basically these agents working independently without oversight. And science fiction has taught us Skynet, you cannot outsource something to AI without human oversight. They're talking about doing this for nuclear power plants like Making smaller sized nuclear power plants where human era was really the root cause of Chernobyl and Three Mile island that could actually be staved off by AI, but again AI with proper governance. I'm an old reporting guru, so I always was an advocate of dashboard reporting. Putting a tool at the fingertips of those that make the decisions and the strategy. Originally you had these digital frameworks like Tableau that were great for sales presentations. You could manipulate them and work with them and expose the markets that were lagging and the salespeople and all that kind of good, which is really good data. So I'm a firm believer that you have to develop systems that enable you to, in a real time basis, monitor what your key performance indicators are. And legal, I tried to do that in my products. So if an attorney, you know, sends a prompt out with invalid information, it snags it, it redacts it if they get a response back from AI and there's hallucinated or invalid material in there that gets, it gets stagged and gets isolated or put into a container until a human being can go in there and work with that, eliminate or redact out the invalid components. So it's really from that mindset that and the challenges with AI is it's this great, wonderful thing and it's super easy, it can be super fast. Using AI to me is like an Excel spreadsheet. You know how 90% of your time and effort goes into the proper setup. Now if you're using an Excel spreadsheet, for example, Ivan, you have to make sure that your formulas are copied over to the right place. And if you're doing paste locks or if you're doing any kind of high advanced work in Excel, there's a lot of overhead to get it to where it needs to be. Because yeah, one failed formula can be the difference between a misstated cost projection or a profit and loss statement. And I've been there, so I've learned the hard way. I think AI is in the same manner. If you're going to use it, you really need to spend the upfront focus and attention on anticipating where it can go wrong and put those safeguards in place. Does that make sense? Did I answer your question? Yeah.

I think it's just interesting because I grew up in the automation world, so things like Scarta Systems and all the rest of it, and part of the reason of bailing out of it, because I remember I was at a biscuit factory up in Scotland and I said, why don't we just get rid of Those people that are really, really flippant. And the guy said to me, and he said, yeah, but what happens to them and their families? We get a government grant to employ them. And that really left a mark on me because I was automation all the way and I never really gave a consideration or a thought to what happens to the people. And I think AI, it's already doing that. To be fair, if you think about the stuff that we do today where we would normally have had to hire somebody in order to have the thing done or at least outsource it to another thing with hair on top, we don't have to do that today. And I think it's only going to get magnified. And that, that for me is a concern actually just whilst we're in the honeymoon phase. It's awesome and I will milk it for everything it's looking.

Yeah, I think we're beyond the honeymoon phase. I think that was 2025, but I think 2025 had so many advancements in so many areas of AI and 2026 with some of the open claw and how that's going to permeate the market where you can have agents that not only do specific end to end tasks, but actually think autonomously and extrapolate on your original task and go out there without your consent and build on that and just we're going to see a lot, a lot of that more and more. And that's like the watershed moment I think that's going to happen in 2026. The, the explosion of these incredibly diverging of the agentic aspect where I can do one like a virtual assistant, it can do really well most of the time with taking that and then what are the backend deficiencies that can be tacked onto that not necessarily by having to daisy chain agents, but having these autonomous super agents that like open cloth that can just continue in perpetuity. Like the greatest example of that for me is I also dabble a little bit in software design and development. I use Cursor and a few of these other products I could just expose on GitHub, which is basically a repository where you keep your code to this open claw kind of agent and tell it, here's my application, go out to the Internet, go out to my competitors and come back each day with 10 features that are not in my system, throw them into a separate like branch or a separate place and then I'll pick and choose what I want to elevate to my product. This is the reality where we're going a single user can with AI can basically compete with larger entities. I think that's 2026 is really going to kickstart how that impacts on legal I think it's another opportunity where super law firms in deploying these kinds of incredible agents can go out and get competitive intelligence on their competition. Obviously have incredible access instantaneously to information that they pull down from all different sources but do that proactively and preparatory not doing it reactively as which is the more typical case in a court proceeding.

Thank you Richard. I'm going to I know we're on the app so we've got Tom's question. Thank you for in that Ivan. That's and what comes to mind for me is how we'll see a speed will be one and also niching in specific areas of law if people have been spread trying to have more clients so they'll be able to and delivery on and seeing that and the experts will need those experts that sense.

Yeah.

Your question Tom. I know we've got we've gone over time. Thank you everybody. But let's have Tom's question and then after Tom's question we'll call it there is an opportunity to speak with Richard. I just want to I'll put this in the email. We'll follow up. Tom, you had a question.

We will or maybe we could do what Bill Maher does take it after hours on YouTube or something. Tom, you had a bunch of questions so floor is yours in eight or.

Nine questions for you actually I hope you don't. No, no. It was a great very nuanced and lots of facets to it. Of course it's interesting. My son is in his first year of law school right now and he along with us last night he was talking about he's looking into different like summer internship opportunities and he's very good about researching the different law firms and of course he wants to get into a top one and he said there's the at these firms he notices there's the people who've been there for decades who are well respected. I think of like our founding fathers.

With their powdered wigs.

Hey, times are changing they said then there's the rainmakers that bring in the new business. And certainly what you're teaching us today relates to you must know this information or you're not serving your clients and so on. And I just thought one of I'll just touch on one one aspect that spoke to me because I'm in the PR business and I like what Ivan said about Will there Still be need for humanity and that kind of thing. And when you mentioned about strength and content, I was thinking coach people on, do it yourself prior to, and encourage them to write their own press releases and so on. And one of the things that I learned, and I'm sure you learned as a journalist over the years, is like when you're writing your copy for press release, you can't make claims. If you're in a health sector, you can't. This will cure or Parkinson's for everyone. You need to rephrase that, my friend. And when you're talking about liability and the people that write the press releases, usually for their own business or their own project or their own service, are not necessarily the people that are scraping that content and putting it out there to the world. And what kind of liability problem does that raise? And I would hope that means there's still some need for people, experts like you and me that can just say, wait a minute, let's be a little more cautious here because you don't want to step into a liability problem.

Yeah, no, without question. And Tom, you and I have discovered recently on various social media platforms, for example, the plethora of deep fakes that are coming out because of tools like 11 Labs and Agen and so forth, that they're actually. And the ability through large language models to like download a bunch of documents or written material by a particular person. And through technology, yeah, you can post stuff. Saw something on Rachel Maddow that was totally fictitious, George Will and all they're interested in. And they're not making political statements, they're interested in getting the eyeballs and gaming the algorithm to get paid on platforms like YouTube and Instagram. And that's just one example to your point where things are really, it's really troubling. And again makes the point for, for governance and seeking the truth, all that more critical in this AI age. And that's only going to get worse until they figure out ways to track this stuff. But AI was used through Cambridge analytics, through elections and through the Brexit double down.

Thank you, Richard. Thank you everyone. Really appreciate you all being here, asking your questions. And Richard, this has been super, super insightful. Thank you for walking through that. And, and on that note of fake news, we are going to be looking at fake news and AI. This has been absolutely fantastic. I know people perhaps couldn't join today. We will be sharing the recording and we'll also share some notes. So please get in contact, Rich, and I will include your details when I send everyone an email.