The Swivel Chair Problem Holding Back Enterprise AI With Clio

[00:00:00] Do you need AI agents that you can trust? Well, with an AI data layer providing real-time connection within your data platforms, you can trust your agents to provide accurate solutions. So, scale your business by trusting your agentic AI accurately getting the work done for you. Trust its capabilities with Denodo. And you can do that by simply visiting denodo.com to learn more.

[00:00:26] How much of your company's technology strategy depends on somebody swiveling between two screens and holding everything together? Well, my guest today is John Foreman. He calls this the Swivel Chair Problem. And once you hear it, you may start seeing it everywhere.

[00:00:49] And John is the Chief Product Officer at a company called Clio, which is a legal technology platform built to help run a law firm. So, in the episode today, we'll examine why AI pilots stumble when data, documents, billing, communications and workflows all live in separate fragmented systems. But John will also explain why connected context matters more than collecting just another AI tool.

[00:01:18] And together, we'll discuss how legal teams are moving from cautious experimentation into daily use. And learn more about where AI can remove admin work that currently consumes 62% of a lawyer's time. So, if your AI looks clever in a demo but confused at work, this conversation should help you make things a little clearer. But enough from me. Let's get John onto the podcast now.

[00:01:48] So, thank you for joining me on the show today. Can you tell everyone listening a little about who you are and what you do? Hi, I'm John Foreman. I'm the Chief Product Officer at Clio. Chief Product Officers, for those who don't know what they are, we work in research and development, helping the company figure out what to build. I've been doing this for a long time.

[00:02:12] I was previously Chief Product Officer at a company called MailChimp, as well as a company called Podium. So, I've been working, doing software development in kind of the business-to-business software space for a very long time. Awesome. It's a pleasure to have you join me today.

[00:02:32] And before we inevitably go take a deep dive into all things tech and AI, I'd love to learn more about what Clio does and why it has become such a significant platform for legal professionals around the world. Great story. But tell everyone listening about that and what you guys are doing there. Yeah, so Clio is a legal technology company.

[00:02:53] And what sets us apart from other software companies operating in the legal space is we really just sort of ask the question, what are all of the jobs that a law office needs to get done? And then we've said, we'll help you with that. A lot of software companies, a lot of B2B software companies, they kind of pick one thing that they do, right? Hey, we maybe just help you write these documents.

[00:03:23] Or we help you store these files. Or, you know, whatever it is, right? They pick a job. Clio's purpose has been, for a long time actually, to basically be the operating system that a law firm runs on. So, everything you need to run this particular business, we will help you do it. And we've also been dogged about, we're here to help law firms. We're here to help lawyers. We're not here, for instance, to get into other verticals.

[00:03:50] We're not going to work for accountants or get into the professional services space or whatever it is, right? It's all lawyers all the time. And I think it was about 10 years ago that I started doing this. I was talking with a lot of legal firms. Many were accused at the time of being slow to adapt to technological change. That was kind of the reputation at the time. But, of course, that has completely changed now. I've seen so much progress, I would say, about eight years ago.

[00:04:18] But fast forward to present day, legal firms have been experimenting with AI and doing so for several years. But many still struggle to move beyond those isolated pilots. And, again, that's not just isolated to the legal industry either. I hear that in every industry right now. So, from your perspective, what's preventing organizations from operationalizing AI at scale? Any trends that you see there?

[00:04:45] Well, I mean, I think what's interesting when talking about law firms adopting AI is it does sort of model previous eras of technology and what it would look like for a law firm to adopt those things. Right? You said eight, ten years ago maybe law firms were a little bit behind. I mean, it makes sense, right? Lawyers are subject to a lot of regulation as well as a lot of, of course, ethical restrictions that come with the trade.

[00:05:10] It's very similar in that sense to, like, healthcare, right, where they're also governed by a lot of rules. And those rules can affect how you approach technology and how you adopt technology, right? So, you roll the clock way back. Lawyers were concerned about what does it look like to put my files in the cloud? Well, why? Well, you think about things like attorney-client privilege and who has access to this stuff, right? Well, those same questions are being asked today of AI, right?

[00:05:35] If I use an AI model, the company that produces that model, whether it be OpenAI or Anthropic, what are they doing with my data? Are they hanging on to my data? And hence, would it be discoverable? So, is there any sort of attorney-client privilege that's being violated here? Or are they training a model based on these queries and the documents I'm providing them, right?

[00:05:58] In which case, not only are there sort of like maybe attorney-client privilege questions there, but also am I losing IP, right? A lot of law firms have built up IP that's basically, it is represented in the form of documents upon documents upon documents. That is the intellectual property of a law firm, right? That's how they distinguish, you know, what they do is like, oh, it's often embodied in documents. They don't want someone to just eat that up and put that in their model, right?

[00:06:27] So, these things have caused people to perhaps be, you know, what would be characterized as a little bit slow. But when you look at law versus other industries, I would say lawyers now, like you point out, are kind of ahead, right? They're sort of actually blazing the trails, I would say, right alongside software engineers. These are probably the two trades we see where folks are really blazing the trails in terms of what does it look like to deeply embed AI into what I do day after day?

[00:06:55] So, yeah, it went from perhaps caution to people actually seeing the value and now using it in all sorts of manner of like different things they're doing at their job, which is cool. Yeah, completely agree with you. And before you join me today, I was doing a little research on you and I was reading how you've argued that AI is also, in some cases, exposing weaknesses in existing technology stacks rather than solving them.

[00:07:22] So, why does a fragmented mix of case management, documents, billing, research, and so many other systems, how is that limiting the value AI can actually deliver sometimes? Yeah, I talk about this as the swivel chair problem. When you're working a job at a business, humans can solve a lot of problems with their swivel chair and a dual monitor setup, right? Oh, I'm using this tool over here. I'm using this tool over here. They don't talk to each other, but I'm here.

[00:07:52] I can read things on both screens. I know what's going on. I'll just move between the two, right? Humans, in that way, we are the spit gum duct tape of the software world, right? A lot of companies you don't think about, but a lot of companies just rely on people to make all this shit work together at a business. AI exposes the cracks in that, right?

[00:08:15] Because AI shows up and we think of it as, you know, the super intelligence, but in many ways, it's, you know, it's the baby deer on baby deer legs being like, what's going on? And the AI doesn't necessarily have the ability to swivel around in a chair between monitors and be like, oh, you have this record over here and it's got to talk to this thing over here, whatever, right? And so AI, AI is only as intelligent as the context that it has, which, you know, we use the word context in AI.

[00:08:45] It just means data stuff that it can look at. That's what allows it to answer questions appropriately, right? If it can't read a bunch of records, then it can't answer questions about them, right? It needs that context. And it's about as powerful as the skills it has access to or actions it has access to, right? So as an example, if we want AI to, let's say, draft some filing for a divorce, right?

[00:09:12] It actually has to have access to court form data in a particular jurisdiction to understand how you would put together that exact filing. It needs that data. It needs that context, right? Which is in some system. Similarly, if it's going to file, it needs to have access to the systems where it would file that with the courts, right?

[00:09:32] So AI is really only as powerful as the context or data that it has and the actions or skills is the word often used that it has access to as well. And so we've papered over a lot of this sort of like isolated or siloed context issues and siloed action issues with humans who have swivel chairs. That's what we've done. And AI shows up and is like, I don't know what's going on.

[00:10:01] I need all of the data and all of the actions you have access to do anything, right? And so that's the way I think that AI today in 2026 is exposing problems in the way a lot of businesses are built. They're built on a myriad of systems that don't talk to each other. Perfect. Such a great analogy there. And I think for many people listening, they would have experienced that almost inherent temptation to solve just about every new problem by, hey, adding yet another AI application.

[00:10:29] But I was also reading that instead you believe structured, connected data, that is where the real competitive advantage lies. And I've been to that many tech conferences where they repeat the mantra that no data, no AI. But just to dig a little bit deeper on that and for people listening, why is that quality and connectivity of data becoming more important than maybe the number of AI tools that an organization owns? Yeah, yeah. I think that a lot of businesses are popping up.

[00:10:58] A lot of AI native companies are popping up that are like, oh, we're going to solve this one particular problem. And going back to this sort of swivel chair analogy, what people aren't realizing is just how deeply connected all of the processes and all of the data at a business and in particular at a law firm are. And I'll give you an example. So, let's say that I have a client that wants to, we can go back to this divorce example, wants to get a divorce, right? And I've got to put together a filing.

[00:11:28] I might have two AI systems, right? I've got one that's over here answering the phones doing intake. Maybe I've got another that's over here drafting and filing, right? What's important to realize is when it comes time to draft the document for the courts and get it filed, the information that's going into that document, I probably actually grabbed it at intake.

[00:11:50] When I was answering the phones, when I was doing the initial consult documents were submitted, stuff was actually probably just described in a phone call or an email, right? What if all of these systems were connected, right? What if the AI doing my drafting and filing knew about that phone call, knew about that email, was actually the AI that answered the phones, right? That's sort of what Clio is building that other companies aren't, which is this AI that's like, you don't need another point solution.

[00:12:20] What you really need is something that hooks into all your systems and understands everything. It understands all of your communications, front office and the law office. It understands all of your documents. It understands all your financial data, billing data. It understands everything coming from the courts, the court forms, how to file. It understands what's going on on the docket with a particular case. It understands the law itself in the form of a legal library. Clio has the Clio library.

[00:12:50] And so it can bring all of this together in one place to get jobs done. Basically, we're just acknowledging that the jobs that someone like a lawyer does, they're continuous across a client journey. Right? The first phone call you have with a potential client, that hooks all the way through to showing up in court. And so why would you isolate those things? Why would you have AI working on pieces of those?

[00:13:19] That's really not how jobs work when humans perform them. And in AI, it's supposed to work alongside folks who are doing real jobs, not just tiny little pieces. One of the things that I love about what you're doing here at Clio and your approach there is how you embed AI directly into the platform rather than just bolting it on as a separate product. So what practical difference does this make, though, for legal professionals, let's say, during a typical working day?

[00:13:48] And why is it leading to better outcomes? I'm sure you've got so many different stories just to bring this to life. But for those legal professionals listening, anything you can share there? Yeah. I think one of the things that people, just to put a little bit more color around the difference between embedding and bolting on,

[00:14:06] one of the things that I feel like has dominated the AI conversation over the last couple of years is this idea that if it's AI, the way that I experience that or see that is more or less a blank screen with a chat bubble in the middle that's like, what do you want to do today? And then you have to bring a bunch of stuff to it, right?

[00:14:33] Originally, you'd brought stuff to it by just putting extra details in the prompt. And then maybe there was a file uploader. And then eventually we had integrations or connectors. You've got to bring stuff into this chat bubble. And that's how you experience it, right? But when you think about doing legal work, oftentimes that's like a wholly inappropriate way to do things. Sometimes that's unwieldy, right? So you want, like, I don't even want a chat bubble. Maybe I want no interface at all. An example of that might be AI answering the phones. Right.

[00:15:03] I don't, if I want AI to answer my phones and do client intake, I don't need a chat bubble. My phone system is sitting right here. Just put AI on that. There's no interface for it. Maybe I got to configure it a little bit, but then it just, it answers the phones. It's like, hey, what are you working on? Like, what legal matter do you have? Where do you live? Et cetera. And you would maybe do a conflict check, et cetera. And there's really no interface for this. It just eventually tees up a console with the lawyer.

[00:15:28] On the flip side, maybe that chat box, rather than being too much, is too little, right? So oftentimes lawyers need things to be repeatable. They need things to be auditable, right? I have to inspect the data that is going into what the AI is about to do. For instance, if the AI is about to draft a filing, I actually want to look at all the fields that are going to be tucked into that form, right?

[00:15:54] And then all of a sudden, the UI that you have, maybe that looks more like traditional software, right? Like, oh yeah, I actually got to go through the entire record of this legal matter. Yep, that's right. These are the children's names. These are the birthdays, et cetera. This is what's going to be, you know, I'm maybe filing something around custody that I need to check all this, make sure it looks good. Chat box might be wholly inappropriate for that, right? And there are other UIs that exist that would be more appropriate to bring alongside that chat box.

[00:16:18] So, I think what we're realizing a couple of years into this sort of LLM version of the AI world is that AI can do a myriad of jobs. They can do some of those jobs in a fully automated fashion, in which case the UI might look like no UI at all. This checks and later is like, I did it, boss. That's the UI.

[00:16:42] And then in other situations, you might actually need to work alongside the AI. The AI works alongside you, in which case you need a lot more review and perhaps a lot more structure than you would need in the form of a chat box. And so, what it then looks like to not just sort of be this thing you slap onto the side is you come into the actual job that's being done and you ask the question,

[00:17:06] if the AI is going to act like an employee, if the AI is actually going to help get stuff done for this law firm, what should the appropriate experience of that be? It's really a design question. It's really a user experience question. These are things software has been answering for a long time. It's very customer-centric. You just kind of got to break out of this headspace that the answer for everything is make it look like chat GPT, which is where I think a lot of people have been living for the last couple of years.

[00:17:37] They really have. And maybe that's one of the reasons that every tech project or at least every AI project is under close scrutiny for the return on investment that it can offer. And on that side of things, another example that caught my attention was how Clio work is actually reducing the time required to build a case-ready litigation timeline from anywhere between 9 and 25 hours to a single prompt, which is just phenomenal.

[00:18:05] But tell me how that works and what role AI plays and why keeping lawyers firmly in the loop obviously remains essential too. Yeah, I think that when it comes to reducing the time worked on a particular legal matter, so many companies have been focused on interesting things. Truly the substantive legal work, right? Oh, let's see if AI can pass the bar exam and go in front of the Supreme Court and all this. And that's very interesting. Can AI produce legal arguments?

[00:18:34] Can it do legal research, et cetera? We provide that, of course, in Clio work. Our AI named Vincent can actually go into the Clio library and do research, et cetera. But when it comes to saving lawyers time, what you've got to admit about the practice of law is that for many lawyers, the majority of their time, on average, 62% of a lawyer's time is spent on non-billable work. There's a lot of administrivia that comes into being a lawyer.

[00:19:04] And that drudgery, that administrivia, that toil, oftentimes AI can get that done too, right? And so it could come in the form of, hey, I've got a ton of documents here. I need to sift through them all and extract these case facts from the document that are going to be really valuable in creating this filing. Or I need to make sure I'm using the right forms.

[00:19:30] Or when I submit a filing, maybe it gets rejected and the clerk comes back with notes and I've got to actually change how I do this, right? Or I've got to follow the rules of this particular jurisdiction when I'm actually doing the filing. Or I need to track my time. That meta idea of tracking your time, guess what? You don't bill for the tracking of the time. You track the time so that you can bill.

[00:19:53] The AI can actually start that stopwatch and stop that stopwatch and group things together into an actually well-described element of your bill. Oh, and then you've got to transmit the bill and collect it. And you've got to maintain a trust account and all these things, right? Like AI can do all of these things too. And oftentimes lawyers are giving them away for free, right? Just a small example.

[00:20:16] Local law firms are not necessarily talking about large law here, but just your basic hung a shingle on my door, you know, main street law firm. The way that people find them is on, you know, they're using AI, like they're using chat GPT or they're using Google or they're using Google Maps to find a law firm near them that does a particular thing. The main way in which that law firm gets found, the main source of data that's used to rank these law firms, including with AI, is review data, right?

[00:20:45] It's like Google reviews on Maps. A lawyer actually needs to ask people, would you leave me a review? When a legal matter is done in order to get that review, in order to get ranked, in order to get found. That is free, right? The lawyer is not billing when they're asking you to leave a review and then maybe following up and be like, did you ever leave me a review? And maybe responding to your review and you leave it on Google. That's done for free. Guess what? An AI could do that whole thing, right? You don't have to do that bit of toil.

[00:21:14] The AI can detect, hey, this was a, you know, legal matter that was solved satisfactorily. We're all done here. Last bill paid or whatever. I'm going to text this person and be like, hey, would you mind leaving us a review and, you know, following up with them? So that's just an example of like the 62% of time that goes into a legal matter that people don't get paid for. AI could come in and do that. Oh, it can also do the substantive legal work too, right? It can help you research. It can help you draft. It can look at like things like a motion to dismiss coming in over the docket.

[00:21:43] We can use Clio Docket to check that. Coming over the docket and be like, oh, what's going on here? What are the arguments being made? How should you respond? So that's how we get that time down. We're involved in everything, right? If you're just involved in one little piece, you're not going to go from 20, 25 hours to nine hours or whatever. But if you're involved in every little piece, including all of that toil, you can make things go a lot faster.

[00:22:04] And I was also reading how Clio is working with Harbor to help the law firms and Fortune 500 legal departments all move from experimentation to enterprise-wide adoption. So beyond the technology itself, what role does things like governance, implementation and change management? How do all these things play a role in making AI successful too? Yeah. So the Harbor example you bring up is an interesting one, right?

[00:22:32] This is a service partner that we work with to make sure that our products can be implemented appropriately in larger organizations. So when you think about this thing I described around the swivel chair problem, it gets worse when you get into huge organizations. Like you think about the largest law firms in the world, right? How do these law firms work? They have tons and tons of systems, right?

[00:22:57] We've seen with a large law firm that may be different areas of it will actually have different ways of storing documents. So over here, they're using SharePoint. Over here, they're using NetDocuments. Over here, they're using iManage, right? And then when you walk into that law firm, you're like, hey, let's use AI. The AI is like, I don't know where anything is. And lawyers at the firm are like, I don't know where anything is either. And so you have to start to think about as I, if I want to bring all this together and we need a single pane of glass, right?

[00:23:25] And so for us, that's something called Clio Operate, which you bring into these larger organizations that basically hooks into these myriad of systems, often duplicative systems just used in different parts of the business to say like, okay, you want to know where all your documents are? Clio Operate can now tell you and can actually go into all those places, right? Or you want to understand everything about your finances. We've hooked into all the different financial systems, et cetera.

[00:23:50] Then once you have that single pane of glass, you can hook AI into that and it can start to find its way around, right? It can start to have access to all of these things. And so when you think about something like Harbor, what we're doing there is we're bringing in a service provider alongside the software company that is Clio to get you all hooked up, to get everything put into this single pane of glass called Operate so that then your employees as well as AI can find everything.

[00:24:18] It's definitely one of those problems that you have once you become a big grown-up massive firm. It's like you create messes, you create IT messes and we work with folks who can help solve them. And finally, obviously our conversation has focused on legal services today, but the underlying challenges that we're talking about do sound familiar across healthcare, financial services, manufacturing and so many other industries.

[00:24:45] So I'm curious, what lessons do you think leaders listening outside the legal profession could maybe take from Clio's experience if they want AI to become part of their everyday operations rather than just another promising pilot that fails to scale? Yeah. As I said, in legal, what we're doing is bringing together actions and data from across the business. Everything that a business does, the front office, the law office, the back office. We're looking at marketing data.

[00:25:15] We're looking at communications data, document data, financial data, legal data, et cetera. This applies to all sorts of businesses, right? Where they would have to bring together a complete picture. You know, let's say you're a med spa. You're bringing together, oh, I've got intake. People are calling the front desk. They're asking about Botox, et cetera. I'm thinking about my providers. I'm thinking about my inventory. What rooms are available to inject that Botox? Oh, that can only be done by certain injectors. You know, what's their schedule like? And you got to bring that whole picture together.

[00:25:44] You think about plumbers, same deal, right? It's like, oh, I got to think about who I've got on the schedule, what trucks are available, yada, yada, yada. So all sorts of businesses, no matter the vertical, they're going to need to bring together the front office, the back office, everyone working at the firm, what they can do, et cetera. Their notion of what inventory is in order to solve problems.

[00:26:06] And so I think the AI that's going to win in each of these spaces, each of these verticals, and each of these businesses is one that understands the complete business. That's where we're headed, right? I know right now things are very fragmented. People are buying point solutions all over the place. And it's not me. I forget who it was, either Andreessen or Horowitz, one of the two, who said something about how the story of technology is really a story of bundling and unbundling repeatedly, right?

[00:26:35] I think over the past few years, maybe we've gone into sort of an unbundling state where a lot of AI natives have shown up and be like, we're just going to do this one little piece. I think we're moving back into a bundling era where people are starting to think about the best AI is the AI that has access to everything, understands everything about my business operations, all of my data, and brings it all together where it can do something coherent. So that's sort of how I would think about this generally. And I think that's a powerful moment to end on.

[00:27:03] But before I let you go, for anyone listening, interesting, learning more about how what we're talking about here is shaping the legal industry and also what other industries can learn from this evolution and anything we talked about today. Where would you like me to point everyone? If you're in the legal space, I would point you to all things Clio. You can follow us on social media, come to the marketing site. Shit, you can reach out to me on LinkedIn if you want to continue the conversation with me personally.

[00:27:30] But that's where you can learn a lot about the goings-on of legal AI. Love it. Do that sound self-interested enough? I think legal is the expert in the space. Well, we need people to carry on this conversation. And when we're talking about fragmented tech stacks, limiting AI adoption and value, and why structured and connected data matters more than just adding more AI tools and everything we're talking about in the legal industry here,

[00:27:57] I know there's going to be a lot of people listening wanting to carry on this conversation with you. So I will include links to your website, to your LinkedIn, and the social channels. No excuse. I encourage people listening to go check you guys out. But thank you for talking about all this in a language everyone can understand today. Appreciate your time, John. Yeah, thanks for having me. This was really fun.

[00:28:20] I think John's swivel chair analogy there deserves to travel far beyond the legal profession that we focused on today. Because everywhere, people have spent years compensating for disconnected software with memory, copy and paste, and heroic tab management. It doesn't have to be that way.

[00:28:39] Especially because AI cannot inherit that invisible labor unless it can access context, understand the workflow, and act through connected systems, not fragmented ones. And the practical lesson here is simple. Before rushing out and buying just another AI application, map the complete job, identify where your information lives, and ask whether your systems can support a continuous journey.

[00:29:08] And Clio's experience shows where value might be hiding. Yes, AI can reduce research and drafting time, but it can also tackle billing, intake, reviews, filing, and so many other unpaid administration. But over to you. Where is the swivel chair problem hiding inside your organization? For all things Clio, remember you can connect with John on LinkedIn and also visit the Clio website. For me, it's Tech Talks Network. Love to hear from you.

[00:29:39] But that's it for today. Let me know your thoughts. I'll be back again tomorrow with another guest. Thanks for listening. Bye for now. Bye for now. Bye for now.