What Law Firms Must Know Before Using AI | The Scalable Law Ep. 7

What really makes AI work inside a law firm, and why do so many AI initiatives fail before they ever scale?

In this episode of The Scalable Law Blueprint, host Julien Emery sits down with Eamon Graziano, CEO and Founder of B&E Management and Consulting, to break down the real reasons AI adoption succeeds or fails inside modern plaintiff law firms. Drawing from years of experience overseeing operations across multiple firms, Eamon explains why AI is a leadership and process challenge first, not a technology problem. This episode dives deep into intake automation, workflow design, and how firms can scale without replacing their people.

KEY TAKEAWAYS

• Companies that implement AI without clearly defined processes struggle to see results
• Intake efficiency directly impacts cost per acquisition and overall firm growth
• AI adoption happens faster at the individual level than at the firm-wide level
• Automation works best when aligned with how the business already operates
• Pilot-based rollouts reduce risk and build trust before scaling
• Human teams perform better when AI removes repetitive bottlenecks
• AI should augment people, not replace them
• Leadership alignment is required for AI to succeed at scale
• Results-based pricing creates stronger long-term technology partnerships

Chapters:

00:00:00 - Coming up...
00:01:12 - Meet Eamon Graziano
00:04:18 - The turning point in AI acceptance at law firms
00:08:48 - Why most intake tools fail at scale
00:11:26 - How cognitive decision engines handle complexity
00:14:27 - The limits of rule-based automation in law firms
00:18:28 - The architecture required for AI to work in real firms
00:24:10 - The power of pilot-based AI rollouts
00:27:17 - Why pilots outperform long-term contracts
00:36:17 - Why individual AI adoption outpaces firms
00:38:26 - Addressing fear that AI will replace jobs

📞 Book a 15-minute intro call with Julien

• See how top plaintiff firms scale with automation
(Click “See How It Works” to book a walkthrough)

• Download the free intake automation playbook
(“Get the playbook” on homepage)

• Join the newsletter, insights on law firm systems, AI & operational scale

📲 Connect with Julien Emery:
• LinkedIn
• X (Twitter)

🎙️ New episodes drop every 1st & 3rd Wednesday at 5am PT
Bi-weekly conversations with the operators, innovators, and legal tech leaders building the digital law firms of the future.

⚖️ About the show:
The Scalable Law Blueprint explores how modern plaintiff firms streamline operations, scale capacity and deliver five-star client experiences using automation, AI and smarter systems.
Friendly, grounded, and built for law firm leaders who want to scale without burning out their teams.

Produced by APodcastGeek

Transcript

0:00

Companies that are looking to implement AI but don't have their process mapped out and thought through already, are going to have a tough time to do it. That's Eamonn Graziano, CEO and founder of BNI Management and Consulting. He spent years working inside complex organizations and law firms. For him, AI doesn't fail because of technology. It fails when teams skip the hard work of defining how the business actually runs. For a company, you need to really have a defined process to be able to say, how is I going to make this better or more efficient? He explains why individual AI adoption happens more quickly, while firm wide adoption requires more intention. When teams take the time to define their workflows and align around clear goals, AI becomes a powerful tool that strengthens execution and helps everyone work better together. People functionally think AI is coming for their jobs. Eamon reframes the fear around AI and explains how firms can use it to grow people and not replace them. By the end of this conversation, you'll learn how to think about AI as a leadership and process challenge, not as a technical challenge, and how firms can use automation to scale without Welcome to the Scalable Law Blueprint. The show where we explore how modern law firms scale smarter, automate workflows and deliver five star client experiences without burning out their teams. Each episode brings you impactful conversations with the innovators, operators, and legal tech leaders. Building the future of plaintiff law practice. From intake to impact, you'll learn how to build the systems and strategies that drive real growth. I'm your host, Julian Emery. Let's dive in. just to give a sense of, what the operations were like at the firms that you were overseeing. So I think you were overseeing, like, six different firms, but at least, some of the ones that we worked with, like, how many people were doing intake evidence collection and what what was like the before versus after state once super panel is fully implemented in terms of number of people, volume of cases, efficiency, cost per acquisition, all that kind of stuff, just to sort of get a sense of the before and after picture. Right. It's it's a great point because cost per acquisition is typically a marketing KPI, which you may not think intake conversion is going to have a huge effect on, although it does. Yeah. So CPG, I think at the time when we started rolling out super panel on Live leads, we had around 50 and takers doing document collection and signing cases, signing around 500 clients a month. time over the next few months, as we rolled super panel up to about 100%, of leads, I don't think we we didn't really fire anybody, but we did not hire anybody new, you know, over that that period of time and the intake team whittled down to around 15 people. When I left, they were signing 700 cases a month. So those are two, you know, huge increases in efficiency. And what's great was that was the firm that had the lowest CPA. And we had some benchmarks because that firm was running one one practice type, and we had two other firms running the same practice type. So we could really see like which CPA, you know, where the CPAs were doing well, like what was working well. So we were able to part of the advantage of having a portfolio like that was we were able to, ab test different things, both on intake but also on the marketing side of things. And the CPA at that firm. The firm where we fully rolled out super panel first was the best. lot of that was because the super, super panel was able to convert more of the leads and then over a longer period of time, those leads that maybe a human wouldn't continue to follow up on. This was lemon law. So there's like a longer life cycle of following up on leads. You know, you might have a car that's not going to be qualified now, but, you know, a year or two years later, you know, they did end up having enough problems to be a case that never, you know, the never give up of the I, really paid dividends on like a long term kind of, perspective, I guess, Yeah. Makes sense. And so it sounds like, the end result was it sounds like you had some natural there's always high turnover. Right. And intake. And so it sounds like you started with about 50 people that people sort of naturally left. And you didn't rehire people. And you I remember you guys just keeping your top performers. Yeah, that was great. The top performers loved it. It was it was incredible for the best people. They went from signing and obviously their bonus based on how many kids they signed. So it heavily incentivized the best people to stay, which was great. guess what's interesting. So when you and I met and frankly, even quite a ways, after you and I met and started working together, the conversation was kind of like, can I do certain parts of intake or not? Should it? Will humans accept it or will consumers accept it? And I feel like we're kind of turning the corner where people are recognizing like, oh, if we look at the pace of change and advancements in the last 12 months or 24 months, pick whatever time period to look at, it's almost like, without a doubt, I think if anyone is doubting the fact that I will be able to handle a lot, like you're probably way behind. The company has changed. Like we know this is coming. How do we get it to actually work in our firm? And you, since you left your role as, CEO across a number of firms and, and obviously, obviously since started working with us and helping us and consulting a bunch of clients that we work with or, are starting to work with across PII and employment law and master law and all that. And you also tested a whole bunch of AI tools. I think you tested more than anyone I ever know. Like you test like 20 tools. While we were working with you, you kept comparing them against us. And I guess what I want to get into is like, I love your perspective, and I'll give you my perspective. Maybe I think the question is if I'm the head of a law firm right now and I've got, you know, 200 cases a month, I'm signing and, I don't know, 2000 leads a month that I'm working on. And I'm like, all right, how do I how do I scale this up? How do I get AI to do some or as much and, and intake as possible? The next question is like which product or team or solution do I trust? And everyone says AI everything. And like it's impossible to discern sort of what's real out there. I guess from your perspective, what makes Super Panel different and why is super panel, in your opinion, the best solution? And and then I can answer that from my perspective, because I think, I think we need to educate people on on how drastically different the approaches are to product solutions in the market. Yes, I tested a ton of different tools. The the things that actually were one of the people, probably the first person that told me this, credit to, some of the owners of the group I used to work for, they they were pretty, motivated to, you know, test and and try different things and, and see what's possible. So. But you once told me it's very easy to get to a minimum viable product. And I think that's absolutely true. Like with the large language models, getting out, you know, into the, into the, the public and, becoming more heavily adopted, it's very easy for people to come up with a minimum viable product. So I was always trying to see, you know, how deep did these different tools go. So there's a lot of tools out there that can book appointments or can answer a call and take down information. And the people that can build those tools, like I can build those tools, and I am not I've never coded a single line of code in my life, and I can build tools that can do that with, some CRMs that are able to do that. Now. What I would caution law firm owners to think about when you are potentially getting in bed with, you know, XYZ company who says they can do, XYZ, you take a look at who is the team behind that company and what is you know, what is behind it. Is it a kid in his mom's, you know, basement? Is it somebody who, you know, just like it's just like a college kid that got got out. Like, did they even have a website? Do they have an email address? Like, what is the background of the people that are doing this? Like, do they even come from the legal industry to be able to understand intake? You know, those are some things that I would look at and what I liked about you and Super Panel is you have, you know, a background in building businesses and and exiting and doing you your co-founder. It was one of was an I guy before I was cool going head to head with Microsoft and Amazon. And I know you'll talk a little bit more about that. so those those are some of the things I would, I'd say when you're, when you're in your selection kind of criteria zone next, how is the process when you're actually rolling the tool out. Most of these tools are like, hey, this is our tool. This is what it does. We can quote unquote connect to your CRM, which could mean a bunch of different things, like whether using a tool like Zapier just to connect to the CRM. Do they just have an API that, you know, has limited capability? And then it's and then is it just going to do one thing? Is it just going to book meetings? Is it just going to, you know, answer phone calls like, all those things are important to determine whether this is a long term solution for the firm. Where super panel is different is super panel. You come in, you take a look at the entire workflow. You build a customized AI experience based on essentially if you were being hired as an employee. So how does XYZ you firm do this part of the process. That is how Super Panel is going to do that part of the process. And it doesn't matter if it's completely different than anything you've ever done before, because you have enough capability now to be able to adapt to any situation. And that is a big differentiator. I've seen it nowhere else. I've seen it anywhere else. Every other tool is, hey, we do intake. Yeah. What are your nine intake questions? Let's get those answered. Boom. There's your solution. But that doesn't move the needle. And that's why that's a big part of the reason why I think the MIT study said 95% of AI, solutions are like not getting it or adoption projects fail. It's probably because people are just buying the latest tool that does one narrow thing, and then it doesn't actually fit within their workflow, that that's where you guys are different to. You. Explain this to me and kind of like my my dumb, marine mind was able to wrap my head around it. There's two common ways to build these tools. One is in one large prompt where you cram everything into a large prompt, and if anything gets out of that funnel, any, anything, any, if there's any variation that's out of what is contained in that prompted breaks the other way to do it, that most of these a lot of these tools do it are a series of prompts down a happy path. Maybe if they're more sophisticated, they have some branches to that, but a series of prompts down a happy path, and if anything goes outside of that path, they're tools not able to handle it. And I, I've experienced like having a shut tool to shut tools down and pause them and like, why is my time being wasted six months after we like rolled this out on on this thing that keeps breaking, which can be super frustrating where super panels different is, I think the you're and you're going to talk about it much more eloquently than I am. The cognitive decision engine and which I love is based on, you know, the Ooda loop framework, observe, orient, decide, and act. Super panel has a mission at each step of the way. Like in this stage, our mission is to collect initial evidence or collect initial information about this case so super panel can autonomously act to achieve that objective. Because of the way you set up the cognitive decision engine under the hood, other great thing is the new voice model that you guys rolled out a lot of most of these tools are using 11 labs that be maybe going directly to GPT or Microsoft for their for their voice models. And you guys built your own, which is the best AI voice I've ever heard. I've heard a lot of them, at least to this point. I'm sure there's people who have heard more, but I've I've heard quite a few. And yours is the best. I'm not biased, at least in that statement. That's helpful. Yeah, I guess I guess what I want to, emphasize is just the approach that we see a lot of companies taking and, and some of the problems that arise from that. So, Okay, let me share my screen, and I'll show you like, so when you're looking at different AI solutions again, like you're saying even right. Like there's, it's so easy to build a, an MVP and you can come at this MVP from different directions. But to make an AI a generic system work accurately at scale requires a totally different approach to the problem. And this how the majority the market comes at it, why it's problematic. And I want to kind of show where we come out the problem from. So can think about any job right. Think about like a job to be done and not not just a task. I'm talking like an end to end job function, like intake, which can involve multiple steps. And I'll get into how we think about that in a second. There's different categories of jobs or categories of work. There's categories of work that involve, you know, reviewing documents, summarizing information and documents, drafting documents. There's a lot of stuff in legal that's a job category around reviewing, organizing, rewriting documents. That's a different category of work than interacting with customers, making decisions, moving people down a customer journey. So we're focused on customer journeys and building journeys or experiences that we take people through end to end and we can do a little bit of that, or we can do the entire thing end to end. But all of this fits into a job category. Now, if you look at all products on the market and all products that have come before us, there's been limitations on what type of complexity can be handled. So think think about like building a workflow in Salesforce or in companies like, like ServiceNow or whatever. They've all got these sort of like automation builders and historically, every automation that you try to build every year, you try to get technology to do something autonomously. You're you're completely reliant on building out rules that determine how that product should perform. And so you have to write a lot of code, and you got to go deep in an industry to understand how things are done. You got to write all these like if else statements under the hood to be able to automate a job. And the result of that is, if you look at this continuum of like a simple to a complex job, if you're relying heavily on these deterministic rules to make decisions, the complexity of job that you can do autonomously has a cap to it. You can't really go much farther than that. It's supposed to be a little checkmark, so you can do anything in here, but you can't do much more. And as automation gets a little more sophisticated, you can do a little bit more. But relying on rules, it gets really complex to build a system that can do a lot of work, because you have to write so much code and there's so many different kind of custom use cases that you have to, that you have to solve for. And it gets really complex. So what happened? I started coming one long before ChatGPT. So I, we started working on this with AI long before ChatGPT came out. And so I basically allowed us to make probabilistic decisions where it's not a rule you're giving, you're giving a decision to, a reasoning engine to make a decision, but it's not based entirely on a rule. So there's some probabilistic distribution of the outcome of that decision. And so what that does is you can build solutions now over here, right. And those solutions can allow you to now tackle problems that are maybe a little more complex. And you start to expand your solution space here. And when Lmms came out, that was a big leap forward in the probabilistic decision making. And so what happened is a lot of people without a whole lot of grounding in how to build these systems, a lot of people ended up building these these sort of AI only decision engines. And what's interesting is if you just use a big prompt and you just use AI to do a lot of the work, you can actually you can actually get higher than this. You can actually leapfrog some of the complexity that you can handle with sort of the old, AI models and so you get these solutions that can handle certain tasks quite well, but they're also very unreliable because it's all probabilistic decision making is quite unreliable. And so depending on the job that you're trying to tackle, you need this combination of rules based, deterministic decision making and probabilistic decision making. If you're all rules based, you're limited in what you can handle. If you're all probabilistic based, you're also limited in what you can handle. And it turns out intake is like up here in terms of complexity. So it's very complex. There's a lot that can go wrong in a customer journey. There's a lot more needs to be understood. There's a lot of touch points. There's a lot of interactions over different channels. There's a lot of things that need to be done internally on systems and ways that behavior needs to happen within internal systems like CRM, case management tools and to use heavily probabilistic decision making or heavy rules based and not enough probabilistic decision making. It's going to break down like the performance just breaks down. And so when you look at these these tools, you're seeing either a pretty heavy line based tool that's going to break down and fail. And so basically it's only going to be able to handle what's say up to here in terms of complexity. And you're like, oh I it doesn't work. It's breaking down and able to handle this. Or you're seeing a lot of CRMs or case management system built on some probabilistic decision making, and they can handle things over here. Or, you know, maybe up here they can handle some additional tasks. But what's required is a specific architecture that can handle enough rules to check off use cases here, but also not be so rules based that like, it's really hard to build for every customer and make it work for every customer, and also not be so AI based that you can't trust it. And it turns out the only way to actually do that in and to solve for the intake use case is every customer, almost invariably, to actually get AI to perform well at scale, it requires some amount of custom code in their environment and senior platform. They can handle complex rules using custom code on a customer by customer basis, as well as have that coupled with AI to make probabilistic decision making the architecture. To handle that, we've built the platform that handles this. We've built the architecture to handle this. We built the internal tooling to allow us to get to that level of customization very quickly on every customer, such that our business is scalable and we're not building custom software for everyone, but that requires a specific type of architecture and that type of architecture, just to kind of go a little farther here, And so that type of architecture is is multilayered. So we've got this hierarchical architecture under the hood where the sort of top line organizing layer in the decision engine is we have some work outcome or output that needs to be done. This could be evaluating a case, qualifying a case like in case this is a customer journey. And this is a it's an end to end customer journey. And there's an outcome we're trying to nudge people to. Just like if you hired a person, you have to teach them about your customer journey and then guide them down that path. Now what we're doing here is under the hood. We're then breaking this down into different we call it sequences. And so a sequence is a is a. Think about it like a workflow that has a starting to finish. And you can chain multiple sequences together to accomplish a customer journey. Within every sequence we have we call it an agenda graph. Think about that as like a macro step of like, hey, at this stage in the customer journey, we need you to gather this information and do this. And here are the tools you got to use. And here's the access you have. And you got to get to this outcome. And then depending on the outcome there's another stage or kind of macro step they need to go to. And so you can have this really complex complex mapping of of these sort of macro steps along our customer journey within AI workflow. And then even below that there are substeps. So when when an autonomous system is making decisions on what to do, not just in a sequence, but then in a macro step, then it's got to go down a sub steps to be like, all right, how do I make decisions here? And to to what Aman's point was earlier is what's happening is it's it's it's constantly checking a state. It's using an Ooda loop. Observe, orient, decide, act. It's observing the situation. It's orienting itself against the goals that it's been trained to accomplish. It knows the tools that it has at its disposal, and then it's executing decisions and then coming back. And as the situation updates, it's in making new decisions based on its orientation, what it's trying to get to. And the only way it can do that with high accuracy at scale and repeatedly, is to have this multilayered architecture where it can move autonomously through all of these different stages on a customer journey and be able to execute code functions and use different tools. It can use a browser, it can use, you know, web forms, email, the CRM, the CMS, it can use an e-signature tool. And it it has a combination of custom code for a specific law firm based on how they do things, as well as the the reasoning engines, the most powerful LMS models that have been honed and tailored to this use case. That's how it's able to work. Its way through. And one of the things even mentioned is, is, is voice. We found that we had to build our own voice engine because voice is like this, how a voice sounds. But then there's also all the little intricacies of of how you interact with the voice, like we naturally, as humans kind of know when someone is done speaking. And maybe they're not like thinking about something else to say. We we understand, you know, people's tone and can be empathetic based on what they're saying. You have to get to, like, really micro inference levels in a voice model to be able to accomplish that level of human like interaction. And we just weren't able to accomplish that off of any of the off the shelf, voice platforms for, say, AI tools. And we ended up building our own that's been outperforming everything that we tried off the self shelf. We had we had to build it just because we're just focused on these complex use cases. There's so many other use cases for voice. Yeah, it doesn't require that level of complexity. But this multilayered architecture, using that Ooda loop framework to make decisions that, that cognitive decision engine and then the tooling that we have to be able to quickly spin up custom code on every customer's account to make sure we're making the AI system drive a customer journey and behave in a way that is specific, to affirm that we're working with our ability to execute that quickly and implement that with that level of, of sort of bespoke ness. That's even a word. That's the only way that you get AI to actually perform at scale. we can talk about like the pilot approach and how you experience that as a customer, but also have how we approach that to make sure that people, you know, trust the system before they scale up. Yeah. Let's talk about the pilot. I love the pilot approach for a couple different reasons. One, it allows you to work out all the bugs with the firm before going live, without the firm being on the hook for some sort of long term partnership. And typically not too much of a time commitment. then rolling out in the pilot, like the way we did it, was we built it out, at least in our use cases. We started with just, you know, responding to form filled leads. We then tested it internally. We then tested it on ten leads and then evaluated how it did on those leads. Then we got we made some tweaks and we came back. We tested it probably on another 1020 leads. And then we evaluated that. Then we started with like 5% of leads or something like that, and we continued to monitor it and see how it performed. And then as time progressed, we got it to handle 100% of the leads. And I, I was very eager. I remember being very eager to get it to 100% of leads. Due to some of the objectives that I had. So I was very happy to get it to 100% of leads quickly, but there was really no reason not to like objectively speaking, the data never said no. The data always. Yes. Do more because won't speak for every single super panel client, but at least in my instance, being a client, the data, the data always pointed towards super powerful performing better. And I think that's probably true overall, because humans or humans, humans can have bad days. Humans, you know, can forget things and humans have biases. So if you just follow the data, you're going to end up getting to 100% of your leads are being handled by super panel. And that's what happened with us. We then started to add more pieces of the process to it. We went and added, you know, inbound calls. We added, you know, document collection. There's probably some other things that we added that I can't even think of right now. We started live transferring calls to our end takers. self-serve signing right. The huge one. Can I sign a client without, a human ever being involved? The answer is yes, it can do it. Super panel does it? They do it. You know, every month there are, probably many sign ups that come from that. I don't know the exact number of percent. But we certainly had them signing, super panel signing cases. And I think what's important to to think about, if you're a law firm and you're looking to adopt a tool like this, the most other tools are going to say, hey, sign a contract, you know, we're going to implement and, you know, you're stuck with us for a year. Or if you're lucky, it's month to month and you're going to get what it is. But with Super Panel, you do a pilot, you pay a fee for the pilot that covers, quite frankly, super panels, probably losing money on every single pilot. But it gives you the ability to say, okay, this is working. We're excited. Do we? We believe. And at that point, Julian comes in and says, now it's time to switch to a long term contract. And I don't think there's been one pilot that hasn't converted because the customer service and the quality is so high. Correct me if I'm wrong about that, but yeah, I love that approach. It is a much more intentional and, individualized experience for the law firm. And it's funny, I was on a, one of my clients is implementing a law firm CRM. I won't say their name, but I was on the implementation call last night and was like, the guy. The guy probably ran the same implementation call 100 times, you know, exact same way. And I asked a couple questions and completely threw him off because they just have their playbook and it's, you know, every law firm's different. So for small law firms that's probably fine or to run things like that. But for, you know, medium to large size law firms where there's a ton of complexity and there's a way of doing things, you can't expect a tool out of the box just to fit in seamlessly and then be, you know, a resounding success. So obviously there are tools where that is the case, but not tools that are interacting with clients and are the sales mechanism for the firm. It's just way too important of a use case. To leave that up to chance. You know? One of the things I credit you with as well, I'm revealing that we've done a ton of implementations and it's it's different. But especially at the beginning, we were especially, a bit probably more cautious than we even needed to be then because we wanted things to go well. We were monitoring everything. We still do that. We're monitoring everything and trying to make sure things are doing well. And I remember, when we were working with, with with you and you're at the firm is you said you're like, Julian, I've got, you know, 50 people that are doing this job day to day. You're like, we have a certain error rate across the entire team. Like, there are so many mistakes every day. And so the faster you can get your system to make some mistakes and then fix it, and then just keep iterating until the error rate on your system is lower than the error rate. Our human team, the better. And your approach was like, let's just keep running cohorts until we get that error rate down. And honestly, that kind of influence, how we started building out our onboarding in general was like, how do we call it a stabilization point? Or like how do we get to the stabilization point? And let's just continually ab test and I'll pull up another graph here just to show what that looks like and how we do this. Now. I appreciate you never told me that before. I appreciate that. Yeah. That was, that was you pushing us in the early days. I remember pushing you really hard, and that's why I went to other schools, because I was like, you guys need you. You want to win this horse race? You know, you guys got to have you have to deliver. But, I mean, I made the strategic bet that this was, you know, when you hire, when you hire 300 people in like X amount of months and then churn, you know, 80% of that, you know, it's not sustainable. Those weren't my numbers. I mean, that was that was a little bit of an exaggeration, but There is high churn and intake. This high churn intake. Yes. The approach we take with pilots now is, is it's quite structured. And so we look at, you know, that that customer journey. Right. So what's the entire customer journey? What's the journey we're trying to build? You don't build the whole thing and launch the whole thing at day one. Not at all. We break it down and we say, okay, what are all the steps here? What are the different workflow sequences that we need to set up? Again, our product, we call them sequences. And then we say okay for that part of that, for that part of your workflow as a firm right now, what is the current success benchmark? How are you measuring that today, whether it's conversion rate or connection rate or, speed to first contact with a lead, whatever it is, what is the current success rate and how do you measure that? And so we'll build out a sequence that then does that part of the flow. And like what you said before, like we we just run it on a small cohort. So we just run it on ten first. We run it internally and then we might run it on ten leads. So if I come here like we'll just run it on ten leads and that's it. And then we evaluate like how did everyone on this go? Ten things mess up and more often than not is like, especially when we first launch a script like this, there's things that aren't quite right. And so we make some iterations and tweak to the system, and then we run it on another ten leads, maybe 20. This time we check it, we evaluate it, and then we run it another ten. And we don't yet have a significant sample size to really compare from an AB testing perspective, but we're iterating to the point where we think it's it's like performing as well, if not better than the current way of doing things. And then once we get to a point where it's running pretty well, we'll run it on, let's say, 10% of all leads going through a given funnel. And if it's still performing well, usually it's like 1 or 2 iterations. Here it hits what we call this stabilization point, where it's consistently performing at or above the way you're currently doing things. The human team is doing things, and you kind of reorient your workflow to be like, all right, AI is not going to handle this part of it. Let's now ramp this part of the workflow from 10% up to 100%. And now let's add another piece. Maybe let's go deeper. Let's have it do more. Let's have a do more. And we take the same approach with every chunk of the workflow that we add as we build out that entire customer journey. That way we're not just throwing something on and hoping it works. It's this iterative approach where we're testing against the current benchmarks and making sure that it's working. And I think that's the only way you can get AI systems to work accurately at scale, in such a way that consumers prefer that path and go down that path consistently and and you're able to scale up your, your, your operations. Otherwise every problem you're afraid might happen can happen. And like, you don't even know about it. Like you need this iterative approach and and then it also even once you're full of shit up and running, then like maybe you change some of your operations, maybe start taking new types of cases. We're going to launch this again with a new type of case. Or maybe you open up in a new state and we got to make some tweaks. And so this is an ongoing relationship where we're helping each firm deploy this agent system in their environment and test it, benchmark it, make sure it's always working with this, this, this iterative approach. So another thing that differentiates Super Panel is most of these tools out there are like hey, pay X amount of fee for X amount of usage. Nothing is tied to results. Or most of what I saw is not tied to results. But the way super panel price is, is based on results. So Super Panel truly is a long term partner in the firms success because you are financially incentivized for the firm to succeed, which is one of the big things that I like about how you guys approach that. as a firm, I'm paying for the amount of signups that you guys help get across the line. I obviously am monetizing those cases that are signed up, and you are financially incentivized to sign up more and better quality cases as well, which is, you know, it makes a long term relationship and partnership make a lot of sense. So that's another thing. You know, most tools out there will say, hey, it's 10,000 bucks a month based on your volume X, y, z. You know, no responsibility, accountability on results. You know, as long as you know your check clears, that's all that they care about. So that's where super panels different. I'm sure there are other tools out there that are going to move towards that model. The better ones, I would imagine will, but that's a really good way. If you're a law firm owner or ops guy or gal who is evaluating tools like what is their pricing structure? You know, pricing on results implies a certain confidence in hitting those results. Whereas just, you know, sticker price is, you know, maybe not as assured on the, you know, results as maybe it should be. Yeah, it makes sense. That makes sense. What do you think about switching gears to, like, just the adoption of AI in Workstreams or like, broad job functions that can involve multiple people because, you and I, I read an article recently which really resonated, talking about how AI adoption on an individual basis is very high. People will adopt tools and it makes them better. And they're interacting with AI. And like everyone's adopting. That's great. But to to change an entire workflow and like reorganize how your business does things, that's different. And there's also risk involved, right. Where like There's even a bit of a disincentive. Maybe if, let's say, you know, an intake right here, maybe an intake leader and like. Well, if I implement AI to do this and I'm on the cutting edge of doing this, like, am I putting my own job at risk or my friends jobs at risks or whatever it might be? Yeah. How do you think about that? And, like, how have you seen that encounter that overcome that? Because there's such a huge benefit to the companies for figuring it out first, but then like. I've seen a lot of bad AI projects. I don't just work with law firms. I work with other companies as well. And the one thing that I'd say about that is companies that are looking to implement AI but don't have their process mapped out and thought through already are going to have a tough time to do it. I think what that article hit on, people adopt AI individually a lot faster than companies is because people know what they're doing, and they're the decision maker on how they're going to execute their task. But for a company, you need to really have a defined process to be able to say, how is AI going to make this better, or more efficient or whatever. And then the leadership of the company, the that, you know, the mid mid-level management and then the, you know, on the ground folks all need to be aligned to execute on that objective. So you can't do that if you you don't know if you think AI is not, you know, a silver bullet that's going to come in and just save a company. You know, there's poorly implemented AI in fortune 500, fortune 100 companies. You have to really have a dialed in, mapped out process. And then when you implement AI, it's, you know, it's more of a journey than just like a, hey, roll out this one project and then there we go. have you seen the fear, I guess, from teams that are worried it might replace some of their work Oh, yeah. have you how has that manifested and how have you handled those conversations. Almost every time, everybody hears the stories and the predictions from the AI gurus that people aren't even going to be working in the next ten years, and we're all going to be, you know, on the streets or in a utopia because I is going to solve all our problems. But people functionally think AI is coming for their jobs. I've seen a lot of small businesses are just scared of AI because of that. How do I deal with it? The way I always try to approach things is by saying, hey, AI is here to help, to be a partner for you, for an intake or for example, let's use an integer. Hey, this AI is going to make you more effective. You are going to be able to sign up more clients. You're going to be able to make more money because of this AI. And if we don't end up needing this amount of in takers to do this job, great. We're going to sign up more cases where some of you maybe will become, you know, client service reps or some other job in the firm where the AI can allow you to do more, and the whole firm is going to grow as a result of that. That's how I always try to frame it. You know, I'm. But then, of course, there's thing, you know, you always see on the news where, there's CEOs that, you know, adopt AI and, I won't name any names, adopt AI, cut. You know, fire thousands of people and then realize they need to rehire thousands of people when the AI adoption doesn't work. So I've never adopted AI to sole purpose of reducing headcount. I prefer just to maybe not backfill if that's the case. Or if that, if that's necessary. But it is also a good augmentation. Yeah. Repurposing into other areas is great because if you opt for a law firm, if you're optimizing intake, you're going to get a bunch more cases. You're going to need to handle those cases, you know, and one of the biggest problems with law firms, you know, after somebody signs up with a law firm, they don't hear from them. And, you know, for months, which is incredibly frustrating if you're a client. So there's plenty of other jobs to be, to be done, and especially if somebody has that firm culture and they're a good cultural fit and they work hard, like there's no reason to just get rid of somebody because I made things more efficient. AI. That's how I like to frame it. That's how I like to get everybody from the senior executives down to, you know, the the tip of the spear, you know, onboard. So. Yeah makes sense. Well anyway, this is awesome. We covered a lot of ground here. There's a lot more we could talk about, but we should save for another episode. Let's wrap it there. Thanks for checking out the Scalable Law Blueprint. If today's conversation helped you think differently about how your law firm runs. Share this episode with a colleague. And don't forget to follow the show so you never miss what's next. To see how automation can transform your intake and operations. Visit Super panel.io and discover how leading plaintiff firms scale with confidence. I'm Julian Emery and I'll see you in the next episode.