When a client loses trust in their attorney, it's almost never because of one wrong answer. It's because the firm shows up inconsistently. In this episode of Ethical-ish, host Paul Bamert sits down with Jerry Zhou, CEO and co-founder of Supio, to dig into what the ABA's duty of competency actually means in 2026, and why the same principle that earns a client's trust is the one that earns an attorney's trust in their technology: consistent, predictable performance over time.

Learn More and Connect

☑️ Jerry Zhou | LinkedIn

☑️ Supio | LinkedIn | Instagram | Facebook

☑️ Paul Bamert | LinkedIn

☑️ Subscribe: Apple Podcasts | Spotify | Amazon Music

Produced and Powered by LawPods

Sponsored by Case Status

Bad lawyer jokes are an old genre, but if you're a consumer, finding a client-focused practice is serious. If you're an attorney, understanding today's ethical duty of care is critical anchored by current headlines and showcasing major legal minds, Ethical-ish speaks to consumers and attorneys about the modern transparent law practice sponsored by Case Status, powered by LawPods.

Hello everyone and welcome to another episode of Ethical-ish. Excited to be here and explore the gray area of what makes a modern law firm ethical or not. And I'm excited to have the CEO of Supio joining us, Jerry Zhou. Jerry, welcome.

Thank you. Thank you for having me, Paul. Very excited to be here.

So we've got some new props that I look forward to getting to you. I don't know if you're right-handed or left-handed, but the Ethical-ish mug depends how you drink it and how you ... I'm right-handed so everybody thinks I'm ethical. So we will make sure to send you a thank you, but super excited to have you here. I've gotten a chance to work with you and your company. I think it's going on almost a year since you and I crossed paths in person in Oregon at one of those injury board events and loved hearing what you were talking about there. But maybe for those of our listeners who aren't familiar with Supio, maybe give a little bit of background. Who's Jerry? Who's Supio? And I'll put you on the hot seat on a few other things there.

Absolutely. I'm CEO, co-founder of Supio. We've been around for about four years now. And when we started Supio, we really focused on one thing. We had a realization that for plaintiff attorneys, it's just really hard for plaintiff attorneys to understand all of the data inside a case. So we're singularly focused in unlocking that, starting with everything from documents like med records and police reports to voice and email these days. And by understanding all of these documents, we believe that AI can help you just do so much of that grunt work that no one really likes doing.

I think it's interesting, right? I mean, just to come right into it, we talk a lot about AI, but I think AI's been around a long time and we haven't really talked about it. What it's really gotten really good at is the language models, right? And it's a really good reader. And I think lawyers are good readers too. So when I think about given enough time, a human can get through all of those materials and put them together, but in the era of large language models, we can do it a lot quicker and not gloss over any minor detail. And so your reading models really do distill stuff intelligently and very fast. I mean, I think that's a very powerful application to bring that to plaintiffs. I mean, do you think about the time component of how quickness matters in those practices as well, not just getting through it and piecing things together, but doing it quickly?

Think about software and AI, right? It's kind of like building robots. Your time matters in the sense that everyone doesn't really like spending time doing repetitive work and we're really trying to do the high value work that makes our jobs delightful. But on the other side of things, really when you think about quality, quality is being able to scale the same types of operations over and over again. And a lot of times when you think about a lot of the work that's happening in plaintiff attorneys, especially in high stakes trials or even day-to-day work, a lot of this is that consistent delivery. And just like having a robot that reads for you is a really simple example of actually achieving better quality than what people can do because now that robot can pour over every single page of every document that comes in, which is just actually impossible for people to achieve.

Yeah. I love the idea of up-leveling and again, digesting every detail that's out there and certainly a law's not the only place. So I'm kind of curious, why plaintiff? Where was your intersection with bringing this to the plaintiff world specifically on ingesting all these documents that make their cases better, their legal matters better?

I think for plaintiff law, when we started Supio, we went and spoke to over 200 different attorneys across many different disciplines. And Plaintiff to me was an intersection of social good and justice and also a technology that is ripe for disruption in terms of just how we can actually build something people use. Our customers aren't on the billable hour. They're really incentivized to just deliver an outcome for their clients. And we wanted to align to that and we realized we wanted to be in an industry where providing technology that works and we don't have to go against the billable hour or any of these other misaligned incentives for our customers as well.

No, I love it. I've came into legal space from the nonprofit space and to me, it was funny when I was exiting the nonprofit space, I got a few naysayers that like, "Oh, you're going to work with attorneys, make attorneys rich." And of course, we all have that bias to JDs, unfortunately, as a society. But as I've come in and approaching my fourth year, it's like, no, I believe that I agree with you. There is an altruism in why people are attorneys and those that get into what I would call consumer law, that's a lot of different things, but obviously plaintiff as well. The altruism is there, but also the David and Goliath story. I love the idea of how the plaintiff attorney can be stronger, especially when they're going up against all odds sometimes in very deep pockets on the other side of the table.

So kind of leveling the playing field, do you ever think of it that way of, again, back to the social justice, the social impact side of what you're doing?

100%. We work with so many attorneys where there are so many cases out there if you get into the details where there's something that just went off the rails and all of these rules that society created isn't set to correct this one edge case. And now you have an attorney that's almost like a societal engineer that's now meant to go and correct that. And if it's something huge like product liability, then you're changing something in a way that it will never happen again. And then if it's something for an individual like a single event case, then you're helping that individual get justice. So to me, a lot of the most interesting parts about our job is just understanding that even though we have this society like at face value, all of this falls in the cracks and it's an enormously expensive proposition to deliver justice for every single client.

And having this type of technology is something that absolutely helps us reset that in a way that becomes, I don't want to say effortless, but at least the work itself is a lot more streamlined.

Yeah. I love the idea of scaling the expense, making everything, again, a litle bit of a stutter on my end, but where I'm hung up, I've heard some cases where the law, the case has fought over a decade and the lawyer can go arguably bankrupt trying to service that case that they truly believe in. And to me, it's like pharmaceutical industry. It's like you think about all the capital that has to go up front to ultimately get to being able to sell on a market, getting through trials, no pun intended, so that you can actually bring that to market. Plaintiff cases, when we talk about expensive, there are the run of the mill daily ones, the operational ones that keep the lights on, but there are other ones that need to be fought for. And I love the idea that you can help keep those expenses down so that the fighter for the everyday American has the resources to go the distance on it, right?

I think that's another cool altruistic side of what you guys have hit on over there. Transitioning a litle bit into, this is Ethical-ish and we like to play in that gray area as I introduced. I'm a little bit into 14 months of trying to go through different aspects of the ABA model rules of professional conduct. And we've looked at the number one bar complaint here at Case Status. We talk a lot about communications and where the balls get dropped with a modern consumer and what we can do to solve that. But that's not the only one, right? We've got confidentiality, every tech provider should be thinking about that. We've got competency, financial. So I'm going to go down the competency aspect, which is case status is in the trust business from client to staff. You're kind of in the trust business of do you trust the models?

Do the professionals that you work with trust what's coming back to them? So talk a little bit about that. I'm just kind of curious how, because there is a competence to your modeling and bringing back synthesized analysis and insights from all of the inputs that you get it in those documents. Talk about how you establish trust with the practitioner that the quality of what you're bringing back is good. I think you kind of hit on that earlier, but there has to be a litle bit of a leap of faith in this space to make sure that what you guys are delivering with your models can help the practitioner move forward in all the value proposition that you're saying. Anything to comment on that front?

I would say there's two parts of that, right? The first part is the realization that if your work is not very accurate, then you end up not really providing any value for your practitioners at all because now they have to go pour over every single part of every single piece of work. Now, ultimately practitioners are responsible for their work, but at the same time, you want to have a level of commitment in the service that you deliver. So for Supio, we're committed to delivering human level quality outputs. And what that means is that we also utilize human in the loop actively to review places where we're just not confident. It could be anything from going into a system and suddenly understanding that there's a piece of handwriting that no one can make out. We'll flag that and we'll actually go and try to get a person to go look at that and transcribe it for the attorney to having documents that we've never seen before.

And that is what creates this level of trust between us and our customers that makes it able for them to build more workflows on top of our foundation.

Yeah. I think the human in the loop is definitely discussed, although I don't know that it's discussed as pervasively as I would like it to be. The idea of augmenting humans in a particular job to be done, the lawyer, the paralegal, the caseworker, obviously you're augmenting what they do, but also putting them into the formula and having that oversight. I think that's powerful because I think there's a lot of use cases for AI that are very Wild West that don't have that oversight. And in some cases they're trying to eliminate the human in the loop on the job to be done and they're also not using safeguards of inspection. So I think that's unique. I mean, was that a hard decision or no decision at all when you guys were building out the model? That's been there since day one, I would imagine.

It was not hard for us. I think when we started this company, we wanted to have that type of relationship with our customers to build that trust. So certainly the evolution of human loop for us today is much more efficient. It's much faster than it started, right? Technology has gotten a lot better, but the value proposition of delivering human quality work for our customers is always going to hold true. And as we evolve, we're really excited to see how that can simply diversify across many different types of data and many different tasks beyond what we do today.

So with Supio, when you're trying to get a new firm to adopt what you have, do they try it? Is that how they build that trust with Supio and your models? Or is there another way that you establish that with the practitioners that join your family of firms?

To get started, what we'd encourage is for them to send us a case, whatever they're working on, and then we'll do what we call a showcase and one of our solutions consultants would jump on and we'll show them and then we'll hand over the keys and they can play with that themselves. And then I think very quickly they'll understand that this is just a new way to work and it's really exciting because for most folks that are masters of their craft, finding better ways to do their work is always something that's really rewarding.

So talk about the data. I mean, your models have to eat. I always joke that AI has to eat and it's only as good as the data that it can ingest. So do you ever see a law firm where you got really good potential but they struggle to get the documents, the information that ultimately help your models? And this would go for a human side of it as well, right? If they had paralegals that were having to read and pull together all of the documentation, of course they do. And I know you're enhancing them that human in the loop, but I'm curious on, I feel like with the AI era, there's some traditional legal tech firms that are trying to close off access to systems to the AI companies. And of course the client, as we know, a case status is oftentimes the purveyor of the data you need, right?

Sometimes it's the medical records, which can be a challenge as well. So I'm just kind of curious, when we think about AI models, the data that we're able to ingest and put into that on the sentiment side in our models, on the prediction side on our models, you have to have the data. So how do you guys think about that and making sure that the stream of data is facilitated and made streamlined as possible into these

Models?This is a nuanced topic and I think we probably have some of the most cleanest thinking on this. So what we think about in this world is there's really two types of models that set out there. There's actually generative models, which really is what you talk to with ChatGPT every single day and Converse and Supio has an assistant as well that does work and all of the information that needs to come out of that generative model is only for the firm itself. If you're sharing work product with us, then it forms memories for the agents that can then go and have those conversations and the word memory is intentionalized and that's a memory for your firm. We don't use that in any type of improvement at all. Now we also have models that just ingest data at scale and these are much more mechanical. It's everything from understanding how to split a medical record into the right pages to detecting handwriting slightly better.

And for these, we're anonymizing and just using quality controls to just understand whether we're actually improving our mechanical systems over time. And customers can certainly choose to not do any of that. But in most cases, I think customers are much more concerned about data in their own work products, really the mechanical aspects of how voice text sounds or how documents are classified across diferent types.

I still think that getting access to the document that has the handwriting, again, we'll continue to do our job to get those documents. I know there's other companies on the medical record side, but obviously the documents come as scans. They've got a lot of imperfections to say. So to get through all of that and be able to pull that in, that sounds extremely powerful. But again, the faster, the better in most of the plaintiff firms that I talk to, get the data, get it to your models, have it synthesized and the output ready, amazing what you can do in that overall formula. Sort of changing gears a little bit, I don't know. One of the things, you talk about your chatbot I've actually found in talking to some of your users, that's one of their favorite things because you sort of leave it open-ended as to how they want to engage with your models versus AI that's being deployed to maybe do a specific job.

I mean, we look at that on the case status side when we deploy a piece of intelligence, sometimes it's to do a particular job faster, help the human in the loop. Sometimes it's to help them raise the quality. Sometimes it's to help them do both. Sometimes it's just to deliver an outcome. In your case, it might be a demand letter, right? Why make the human faster in producing demand letter or make the human faster in producing a quality news demand letter? Instead, just get the demand letter, right? Let's just produce the outcome. And then the last category I think about is insights gained. I don't know if you think about jobs to be done by the practitioners in that way, but I do have to say in talking to some of the folks that use you, I think you span the gamut and some of it is the chatbot function of it is it's sort of in the art in the limit of their own creativity of how they can put your models to use.

Is that part of the design? Am I representing that right? Or do you know of all the use cases that these people are talking about at the water cooler and discover on their own? I'm curious how that evolves. Maybe you learn about your product from your happiest customers and you had no idea that it could do what they're describing.

Yeah, it's a good question. There's certainly opinions that we have to form around key workflows and usually those workflows are so complicated that it requires a lot of specialized modeling and product development. So our instant ledger, for example, is quite unique in the fact that it represents the whole case economics of the system. We're ingesting leans data to AMP bills, understanding all the different types of layouts just to get an accurate extraction of all of this and then deduplicating line items and understanding the relationships, right? That's a very complex workflow that we really have to understand. But really what we're trying to do in every single one of those workflows is get to the heart of the data. So once we have that full data, whether it's our medical chronologies or this type of information, then we're very un-opinionated about the output workflows that attorneys want to do.

And that's really what separates us from the pack is when we think about outputs like you're talking about freeform through chat or as we build more agentic tools, we're providing the tools for the AI itself and then for the user, they can navigate this and it could be like any type of conversation they're having. And then the AI can kind of figure out what it wants to do by assembling all this data that we've pre-processed and we've made sure it works. And it's been really exciting to just see how attorneys can take that and take that to the next level.

I would agree. I mean, in the conversations I've had with your family of firms, and it's just a touchpoint, it is valuable and it does deliver value at a lot of different corners. I do like the idea though that you go deep in your particular workflows that you're trying to enhance or improve. I think there is a ... I've seen it in a few places where somebody has some success with an AI model and instead of going deep like you're describing for the plaintiff specific type attorney, they start to go horizontal and make that model do other things. I think about everything that you just described, it's very back office type work, which makes a lot of sense. There's a lot of nuance to it, as you mentioned, a lot of specific intelligence that's needed to do things. That model doesn't necessarily ... You can't put that on a chatbot for an intake to say, "Hey, talk to prospective clients and figure out, right?" But I see some organizations out there that are trying to do that.

So I like the idea of going deep and staying in a very specific lane. It sounds like that's intentional from how you're strategizing. Am I representing that correctly? I mean, again, not to take these models and just bastardize them and put them into every application where there may be an input. You know what I mean? Do you look at the market that way? Do you see other people doing it that way?

So the nerd hat's on. And what I'd describe is there's two types of investments that every company makes in AI. We make infrastructure investments and application investments. For Supio, we're pretty infrastructure heavy. We're always investing in how to understand data. So even that complex workflow that you're talking about, like we just talked about for understanding data, KCON data is mainly like medical billing financial data. So all of those different processes step by step is to just distill down into a single ledger that firms are producing today through so much time. And we think about that as infrastructure because once that piece of data is there, now every workflow, whether you're doing a mediation brief or a demand letter or some type of complaint, it has access to this type of data in a way where it could selectively use that as the agent or the user feels necessary.

And that part is application because the application itself is utilizing the latest foundation models where using reasoning, right? The user might say, "Okay, go into this ledger, think about what are the unrelated charges and pull those out and then add this again." That's all workflow stuff. And you could in theory build applications that talk to clients and things like that as well. So for us, it's always a balanced investment. I think about the client facing, collecting feedback. It's all on the application layer and that does require a lot more nuanced understanding of the situation and we try to make that really flexible for lawyers to be able to do what they need to do.

Well, I think that it's a good lens for folks that are out there evaluating because it's not just in this space. Obviously AI is overwhelming if I'm a law firm and the number of point solutions or also folks who claim to be a holistic. So it's pretty intense what's coming at them and there's almost a paralysis in some of the observations I've made. So a good lens to put it into what you're doing. And that's why I think a framework is important for any firm to evaluate. That's why I think the efficiency versus the quality versus the outcome versus maybe insights or intelligence gained by the humans in the system, like all four of those are fair game and you might get all four with one model, but put it in its proper swim lane if you're going to evaluate something and ultimately what you hope to gain and measure success or ROI against, right?

So I think the ideas that you present about what Supio brings to bear can be learned in what I might call AI center of excellence at a law firm. So I think it's pretty powerful stuff and you guys should keep doing what you're doing. Well, love it. And you nerded out. We never really hit on the fact that the nerding side comes naturally because we share an alma mater. We're here in March. I don't know when we're listening to this in the launch recording. So hopefully we're not jinxing here on Friday the 13th, but a Blue Devil. That's where all ... Is this Duke and Durham, did you start down this path of computer science and sort of getting into what ultimately led you here? When did you start to nerd out on this side of business?

I've always loved building, especially when I was at Duke, but I think it wasn't until later in my career in COVID at Microsoft when my coworkers were remodeling bathrooms and I was sitting there thinking about what I wanted to do with my life and I was like, instead of staying home and being in lockdown, let's go build a business. So I called up my co-founder. We sat at a Starbucks and we wrote down all the problems we wanted to solve and being from what we called the mothership and I was at Office 365, we were able to really see all of the different types of products that were integrating against Office to try to automate productivity. And Legal Tech came up always front and center and it was an enormously hard proposition. And as we were trying to build that ecosystem, it was always complicated.

And one of the things we always left that company with was understanding that if you can't understand all the workflows, if you think about attorneys, the long tail of complexity, then starting with the dataset is always the truth. So as we approached Supio, we held that and it's always just about understanding the facts and data. And if you think about that, that's an enormously complex value proposition given how different every case is and how much data and nuance there is.

Yeah. I think that it's a good origin story, Jerry. And everybody thinks Duke's a basketball school, but it's a nerd school, right? All those Carolina fans out there recognize that yes, we are a technology and an innovation. So go Duke. I think it's really cool to have met you after the fact. I'm a little bit earlier on the campus, but I love the work that y'all are doing, you and your co-founder and the whole team and look forward to keeping an eye. And again, on the leadership of competency, I know you do a lot on the confidentiality. Keep up the good work there too, because again, we're only as good as the work that the legal professionals are able to bring to society and I think you guys are part of that next generation, which is really cool. So what closing remarks do you have?

Any last advice or public service announcements when it comes to how firms are operating or what they can do in 2026 when everything's coming at them so fast and the world seems to be changing? What's your guidance since you're sort of leading a lot of this change?

Yeah. When I think about technology, I think about technology as like a relationship just like you have with people. We talk a lot about trust and the way you form trust with technology is you build a relationship with it. Even if you think about something like accuracy, which we hold really dear to Supio, it's not accuracy that is breaking trust for attorneys. It's the consistency of the performance, right? You could have a really junior staff member on your team who's just not very accurate, but at the same time they're consistent. They do the same thing over and over again and that becomes a place where you can form workflows on. So what I'd encourage attorneys to do is clearly go try Supio, but also start building a relationship with AI. So as it comes at you and you fast forward two or three years and everything is like getting streamed to your brain or something, it's not a surprise because you have that relationship gradually built with the technology.

Yes, I think it's good advice whether you're working with Claude, build a skill, treat it like an employee, build its foundation. I agree with that. I think that it then becomes understandable in the sort of world that we live in, Jerry. Well, appreciate it. Again, I think you played in a lot of the areas that are relevant to our title. Jerry, thank you for taking the time. Look forward to sharing more information about Asupio and the things that you're doing, but keep an eye out Jerry on the road. If you get a chance to see him or hear him speak, take that opportunity. So thank you for those of you who've listened to another episode of Ethical-ish. We look forward to bring you episodes like this and other topics that are intriguing in the world that we are living trying to bring legal services to the everyday consumer out there.

Feel free to subscribe. We're pretty much on every channel that has podcasts. You can go to Spotify, you can go to YouTube and subscribe and you'll get all episodes like this coming forth. So thank you and look forward to seeing you on a future episode. Have a great day.

Lawyers and legal consumers, you are our audience and our biggest strength. If you find Ethical-ish a good investment of your time, please spread the word until next time sponsored by Case Status, powered by LawPods.