The billable hour has survived a lot of threats, from alternative fee arrangements to client procurement, but this episode makes the case that AI changes the pressure level. We open with a blunt assessment, time compresses, clients push back, and the old strategy of “work more to earn more” stops scaling. Enter Stefan Cisla, co-founder and CEO of Ayora, who frames the moment as less a tech problem and more an operating model problem. Law firms still place P&L accountability on individual partners who carry deep legal specialization, then ask them to moonlight as revenue managers. Stefan argues firms are starting to replace that fragile setup with tools that support decision-making across pricing, budgeting, and matter management.

Stefan’s origin story is half high finance, half clinical decision science. He came out of investment banking and professional services transactions, his co-founder Dr. Gordon McKenzie came out of surgery and a PhD path tied to decision science and software. Together they pulled lessons from clinical triage and continuous improvement into the law firm context, focusing on how experts make better decisions under constraints. The hosts tease out the cultural weirdness at the center of the partnership model. Partners often take the long view for client relationships, yet short-term firm economics still take damage through write-offs, scope creep, and messy budgeting. Stefan’s pitch is reconciliation, align client-first instincts with firmer, data-backed pricing and project discipline.

A core anchor for the conversation is the often-quoted $36 billion annual “value gap,” described as preventable revenue leakage tied to write-offs, weak billing practices, bad data, and poor working capital hygiene. Stefan suggests the number matters less than the trend line. AI pushes a new kind of risk, mispricing innovation. If AI reduces billable hours, firms face a squeeze between steep rate increases and client resistance, then end up forced to express value in new ways. The show leans into a spicy idea, the push to change is no longer only client-driven. Stefan sees rising pressure from inside firms, often from the CFO and operations leaders trying to fund AI investment and protect cash flows in a higher-interest-rate environment. Greg sums it up with the line, “the call is coming from inside the house.”

Ayora’s product angle lands on two hard truths, pricing tools in legal have a rough track record, and law firm data quality has been a “25-year overnight problem.” Stefan explains why earlier tools struggled, low urgency when billable hours printed money, ugly underlying time and matter data, and products that were either too complex for occasional users or too simplistic for real-world exceptions. Ayora’s bet is that the data problem is solvable. Their system uses large language models plus proprietary approaches, including work-type ontologies, to extract signal from messy time narratives and matter metadata. The goal is consistent fields and usable categorization across tasks and phases, even when client taxonomies differ. Stefan claims field-level reconstruction and normalization at high accuracy, enough to power a chatbot-style interface that generates a pricing proposal in roughly 90 to 120 seconds by finding precedent matters and adapting them to the new scope.

The conversation closes on the part everyone in a partnership feels in their bones, culture beats software. Moving away from the billable hour is not only a finance shift, it is an identity shift, habit shift, and trust shift. Stefan describes adoption as a joint change strategy, with peers inside the firm as allies, and lots of direct conversations with lawyers to build trust in the recommendations. On the “generational gap” question, he leans toward curiosity over age. Some of their heaviest users have plenty of gray hair, and they tend to be the lawyers who care about how a practice runs. For his personal AI usage, Stefan gives an honest founder answer, meal planning for a two-year-old, automating company chores, and using AI as a sparring partner, with Notion as his favorite tool.His crystal ball point is one law firm leaders should underline twice, gross margin dynamics get messier as tech and LLM costs become part of the delivery mix, and the distance between inputs and outputs grows, driving both consolidation pressure and a new wave of innovation.

Listen on mobile platforms:  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Apple Podcasts⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ |  ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Spotify⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠YouTube⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ | Substack

[Special Thanks to ⁠Legal Technology Hub⁠ for their sponsoring this episode.]

⁠⁠⁠⁠⁠Email: geekinreviewpodcast@gmail.com
Music: ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Jerry David DeCicca⁠⁠⁠⁠⁠⁠⁠⁠⁠

Links:
Transcript:

Marlene Gebauer (00:00)
Hi, I’m Marlene Gabauer from the Geek in Review and I have our friend Stephanie Wilkins from Legal Technology Hub here. you know, Stephanie, it just seems like we just got out of conference season and now we’re back into conference season again. Can you help us with this?

Stephanie Wilkins (00:14)
Yes, I was thinking the exact same thing Marlene. I can’t believe we’re already on spring 2026. I mean last year was a big haul. It was a long haul when it came to conferences and a lot of people, myself included, were excited to get a break but now we are back again. So, Legal Tech Hub will be on the ground for a number of conferences this year. I’m focusing on the next few months only because you know it’s a marathon not a sprint and we’re hosting some of our own conferences and we would love to cross paths with you in whatever city we all happen to be in.

I know the first, probably the biggest one coming up first on the calendar is Legal Week in New York in March. We’ll have a booth and we’ll be taking meetings, but the day before it, we’re also bringing back by popular demand our Vendor Fest, has been renamed to LTH Velocity, which is designed as a free event for Legal Tech founders and CEOs to hear from investors and buyers about what they’re looking for in the space. We’ll also have team members on the ground at a number of conferences, including

Women in AI Summit 2.0 at Vanderbilt and the College of Law Practice Management Futures Conference in February and then Legal Week and ABA Tech Show in March, CLOC in May and so many more. So feel free to look us up or we’ll have a cup of coffee. We’ll get together and chat.

But it’s not only those events, there are so many more, which is why on Legal Tech Hub we have a whole public events calendar. It’s free. It’s not just LTH’s events. It’s all the events in the industry that we’re aware of. You can find it by looking at legaltechhub.com backslash events. know, it’s a whole, it’s set up like a calendar, like you can expect it. And so you can see what might be in your area, where you might want to spend your conference dollars. Or if you have an event that you don’t see on there and you want people to know about, you can shoot us an email and we

can add it to it because we’re just there to again give all the information to everybody that we have. but there might be things in your backyard that you’re not aware of. So check it out, check out the calendar, check out the Legal Tech Hub events and just let us know if you’re gonna be where we’re gonna be and we can all get together and talk about Legal Tech. Yep.

Marlene Gebauer (02:13)
Sounds good, thank you.

Marlene Gebauer (02:22)
Welcome to The Geek and Review, the podcast focused on innovative and creative ideas in the legal industry. I’m Marlene Gabauer.

Greg Lambert (02:29)
And I’m Greg Lambert and Marlene, you know, we usually when we talk about the business of law, we talk about the incremental changes that that are affecting us week in and week out. But today we need to address that fundamental reality of the billable hour model is being challenged like never before. So I know we’ve been saying this since like 2008, but we really mean it right now.

Marlene Gebauer (02:55)
Maybe before

that, I don’t know.

Greg Lambert (02:56)
Yeah.

So as you know, and as you deal with every day, know, clients are pushing back. AI is really challenging and compressing time. And the old way of, you know, just simply working more to earn more is just not holding up over time.

Marlene Gebauer (03:13)
Yeah. I mean, I just did a presentation on the dealt with, you know, AI valuation and billable time and, and, you know, writing an article on it and talking to people about it. So it is, it is a topic that is, is quite hot right now. Um, and you know, in this high pressure environment, the amount of money law firms are losing to inefficiency is staggering, right? You know, we’re looking at a value gap estimated at, uh, $36 billion annually.

Greg Lambert (03:28)
Mm-hmm.

Marlene Gebauer (03:43)
And this isn’t money that wasn’t earned, it’s money lost to unnecessary write-offs or bad data or poor work in capital hygiene.

Greg Lambert (03:52)
Yeah, I always love it when my CFO says, you know, billing hygiene. It’s one of my favorite. It is. It just brings a nice, nice mental image on there. so today we are joined by someone who is attacking this problem with a mix of high, high finance and behavioral science. Stefan Cisla is the co-founder and CEO of Iora.

Marlene Gebauer (03:56)
Just the word hygiene in anything is like cringy. ⁓

and Stefan, welcome to the Geek in Review.

Greg Lambert (04:18)
And welcome.

Stefan (04:19)
Hi Marlene, hi Greg, I’m delighted to be here and excited to jump in. ⁓

Greg Lambert (04:23)
We’re happy.

I’m excited to have this conversation. So, Stephen, let’s start with.

Marlene Gebauer (04:26)
Yes.

Greg Lambert (04:30)
what’s called the founder DNA that you are an investment banker and your co-founder Gordon is a surgeon who specializes in decision science. Most legal tech is built by lawyers trying to fix drafting tools. So how does looking at a law firm through the combined lens of private equity, asset management and clinical triage

Marlene Gebauer (04:46)
You

Greg Lambert (04:56)
what it is that you’re building.

Stefan (04:58)
Yeah, that’s an amazing question. I think, to be honest, I think what we’re seeing is a bit of an organization failure as opposed to anything else. I think a lot of these questions boil down to what’s the split of responsibilities for the P &L between the different parts of the fam, right?

In many cases a lot of that rests on individual partners and that is frankly weird because you you’re expecting someone to have a very narrow legal specialism and at the same time you want them to have a broad quite generic business acumen. That doesn’t necessarily go together but look I think more and more firms are recognizing that right and I think

The interesting bit is that historically that basically meant hiring more people, especially in what I would call like middle office, so people who essentially help with decision making. Right, but today I think increasingly that’s actually about tooling. And I think increasingly firms recognize that you can bridge a lot of that gap with the right set of tools.

Greg Lambert (06:12)
Well, I’m curious about the two of you and your background. What’s the story of how you two got together?

Stefan (06:20)
We met online like many happy married couples these days.

Marlene Gebauer (06:24)
Hahaha!

Stefan (06:26)
And that feels like eons ago, but that was only three and a bit years ago. Now look, ⁓ we both reached a point in our careers where we want to change. I worked in finance for many years and I’ve seen a bunch of things that worked and a bunch of things that didn’t. I probably had a bit more insight into the inner workings of professional services than the many other people I spent a bit of time doing transactions in that space.

So he left medicine by then and he transitioned full-time into software. So when he was a surgeon, he did an academic track at the Imperial Hospital here in London and he then ended up doing a PhD, which as is often the case with scientific PhDs, it was a blend of science and coding. And so he basically caught the software bug and he hasn’t moved back.

Greg Lambert (07:20)
⁓ man. ⁓

Marlene Gebauer (07:21)
So a decision

scientist, what is, what is that? What does that mean sort of in this context?

Stefan (07:27)
Yeah, it’s a great question. look, I think if you ask Gordon, know, he was a practicing doctor, but he also really enjoyed the sort of, let’s say, continuous improvement aspect of being a clinician. And a lot of that is all about, you know, making highly specialized professionals make better decisions and empower them to do that. Because obviously, you know, we’re dealing with highly intelligent, very motivated people.

So how do you do that? How do you enable that? And many of these learnings, we essentially transport them into the legal world, which is obviously very different to medicine, but there clearly are some parallels.

Marlene Gebauer (08:09)
Interesting. Yeah. Very, very, very unique. I, you we had to, we had to dig into that. you’ve described lawyers as, I like this description, so I want to, I want to hear more about this moonlighting revenue managers. So please explain and you know, why is it that these sort of highly intelligent professionals, you know, who are fiduciaries for clients, you know, are, you know,

Greg Lambert (08:09)
It’s quite a combination there.

Stefan (08:13)
I know.

Marlene Gebauer (08:34)
often are really challenged in that similar role for dealing with their own firms’

Stefan (08:43)
Yeah, totally. And look, I alluded to that at beginning. I think that’s part of the thesis that we started with when we started thinking about these problems.

I really do think that lawyers have a really hard task. Because in professional services in general, and especially in big law, the traditional way is that you essentially run your own client relationships and you run your own practice, your own P &L, even if you’re part of much larger organization.

And I think, know, inevitably, if you’re also the practicing attorney, you know, all those, all those halves will end up, end up leading to the conflicts, right? And I think, to be honest, I think in my experience, many lawyers actually take the long-term view on the client relationships. And, know, that sometimes shows up as, as perhaps things which in the near term are not like the most optimal outcomes for the firm. But, you know, at the same time, if you look at the growth trajectory of law

maybe there’s something to be said for taking that approach. I think where we come in is…

And we’ve been saying this for a long time, is like, maybe this can be reconciled, right? Maybe you can make somehow square these two priorities. As you alluded at the beginning, for a long time when we were talking about these things, that perhaps wasn’t the highest priority among many people within the industry on their list. And that is changing now, because I think the entire commercial model is being challenged. And I think it’s a great time to rethink a lot of these fundamental questions.

switch and very much baked into how professional services convince and how big law firms operate.

Marlene Gebauer (10:13)
Yeah, think firms are doing well, you know, with sort of the old model, but I think leadership also recognizes like how long is that going to last? And so we need to be thinking about the future that way.

Greg Lambert (10:27)
Yeah, yeah, the old adage of it’s hard to tell millionaires are doing it wrong, but I think the millionaires are actually waking up and realizing, especially with AI, that they know they’re on limited time, I think, because I’m actually hearing it from people a lot more than I’ve ever heard it before, but from partners.

Stefan (10:27)
totally.

Marlene Gebauer (10:49)
And it’s like, it’s like up and coming partners. I think it’s, it’s the people who know that they are going to sort of inherit and, they need to make sure that, you know, there’s, there’s something, you know, there’s something there to do to, to inherit.

Greg Lambert (10:53)
Yeah.

Yeah,

I also hear a lot of I’m glad I’m retiring in five years.

Marlene Gebauer (11:06)
hahahaha

Stefan (11:07)
Hey we’re living interesting times for sure.

Greg Lambert (11:10)
So Marlene alluded to this in the intro, but Stefan, you mentioned that there’s this $36 billion revenue leakage figure. And then when you dig into the data, what do you think is the biggest contributor to this gap? Is it just a ostrich effect where everyone’s got their head buried in the sand, or do you think it’s an actual process failure?

Stefan (11:38)
Yeah, I mean look, it’s great question again and I think to be honest the answer is changing quite rapidly.

The number we came up with is essentially an estimate of how much money gets left out of the table for preventable reasons. So things like scope creep and perhaps slightly less approach to things like budget management on matters and so on. These are well-understood problems to some extent. That said, they’ve always been really difficult to address. And that’s probably why they persisted.

I think that the reason why I think this is changing now is that as the commercial model changes, I think it’s going to be expressed in a different way. So perhaps it’s going to be less about individual write-offs on matters and perhaps it’s going to be actually about like mispricing innovation in a way.

What I mean by that is AI will reduce billable hours. I think that’s increasingly clear. And I think people have a choice. They will either need to start increasing the rates pretty steeply. And I think that in itself, because of AI, will only get harder. Or they will need to find other ways to express value here.

And look, think the billiard hour is here to stay, but I think that it also will all get much more messy, right? And we’re already seeing this when we’re working with our clients. And I think if you don’t get that right, I think that’s where you will see huge value destruction taking place.

Greg Lambert (13:04)
⁓ Yeah, when you’re talking to your clients, because I know you’re talking to them, they’re feeling the pressure as well, like you mentioned, like never before. So how are these external pressures that are coming in from clients in the industry giving you the opportunity to convey this message to the law firms that use

and that is, you know, is, you doing this as, you need to, you know, bone up on your good hygiene or do you see this as a very critical survival strategy that they need to take going forward?

Stefan (13:45)
I think it’s the latter. look, it’s really interesting in terms of what we’re seeing as a business ourselves. We’re actually seeing like incredible market pull for what we’re doing. And we can go into more detail as to what exactly it is that we do. there’s like a huge uptake in the industry to look at products like ours.

And look, I think that’s partially because, as you say, obviously, AI is yet another toolkit for clients to potentially push back on rates or on bills in general. But I think what’s interesting, and I think this is very much something that we’re seeing across the board, is that maybe for the first time, it’s actually the pressures coming from law firms themselves to move away from the billable hour.

And think that’s actually what’s like, you know, that’s driving the real change. That is driving the real kind of appetite to look into this.

Marlene Gebauer (14:36)
So.

Greg Lambert (14:37)
The call is coming from inside the house, Marlene.

Marlene Gebauer (14:39)
Yeah. Yeah.

So, you know, we’re also like in this post zero interest rate policy world, right? Where the, cost of capital is, is real. So now that interest rates are up, are you seeing CFOs exert more pressure on partners? You know, is the CFO like the Trojan horse for adoption?

Stefan (15:00)
Yeah, mean, totally. And it is a very valid argument. think, you know, when I say that the pressure comes from within, I think that usually means that it comes from the control room, So essentially, you know, think firms see this as potentially a way to capture some of the expected efficiencies that they expect to realize from the AI investments in the coming months and perhaps even years. Like this move to, you know, something

in the bill of law, probably some kind of project-based fee or like a fixed fee policy or something of this sort. That’s really interesting, but I think that is like a management decision to some extent, but it’s also a big change to a large extent, right? Because you need to make sure that this trickles down. And I think people both want to capture these efficiencies, but I think in general, people expect that…

the continuous sort of AI adoption and the entire kind of like transformation of the industry will require quite a lot of investment and like CapEx. so obviously, financing that investment is something that’s in people’s minds. So it obviously contributes to how people think about, where can we maximize our cash flows?

Greg Lambert (16:10)
Well, let’s jump into the product itself, the solution that you have put out. So, and I think you’ve admitted this before and we’ve all seen it. There’s been a lot of entrance into the pricing world for law firms and there’s been a lot of exits for these pricing tools. It hasn’t exactly been a high success rate in legal tech. So, can you…

Talk to us more about why you think products that haven’t succeeded in this market and why that is, and then kind of fill us in on, if you see all these failures, what makes you think that you’re gonna be different?

Stefan (16:51)
Yeah, so our pricing product is not the first product we’ve built. And when we started looking into this space, probably like in late 2024, a number of people that I really deeply respect and who know the space really well, and they really know the Legal Tech space really well, they did warn us, this is like a graveyard of good ideas. And many people tried.

And you know, look, I won’t name any of the existing products, but if you think about the great success stories of this industry, they probably aren’t that great.

And I think there are a couple of reasons why that’s the case. One is, in the past, I think there was no real impetus for law firms to really adopt it, and especially lawyers. The billable hour is an amazing mechanism in many respects. And one of those is that it gives you a little viewway in how you think about this upfront quoting element, because it just caters to life changing halfway through a matter.

And I think secondly, this is a hard category to build in. You probably want to leverage the existing data within a law firm, especially things like time sheets to figure out how much time things take. And frankly, the underlying data that you may want to use is pretty poor quality,

And secondly, you want the product to be both sophisticated enough to cater to a variety of scenarios and also super simple so that people who are not everyday users can still make use of it easily. And I think, know…

be honest, the first point, the market appetite is changing and I think that is definitely getting easier. I mentioned we’re seeing a real pull from law firms and our early adopters have been some of the biggest firms in the US and in the UK. And I think that really speaks for…

how this industry is moving and the people who tend to be like early movers on emerging, successful emerging tech are the ones that also partnered up with us.

That said, I think that the building that is still pretty difficult. Now, we actually refuse to accept that we can be sort of constrained by externalities, like bad data. People kept telling us, there’s nothing you can do about us. And we just refused to accept that. And we said no.

And we have a very talented team and that team is like passionate about solving really hard problems. Most of us on the team are like ex, like scientists, mathematicians, and we just like attacking this kind of conundrum. And so we found ways to actually like clean and seriously augment those internal data sets and so much so that we actually can use them to produce pretty clean quotes for legal matters.

And sorry, and then the last thing, guess, you we talk to lawyers like every day. It’s one of my like KPIs personally is that I want to speak to at least one lawyer a day. My dad is a lawyer, which does make it a bit easier to manage sometimes. But you know, I try not to cheat. Yes, exactly. I try not to cheat too often.

Marlene Gebauer (19:45)
If you’re having a slow day, you can always call dad.

Greg Lambert (19:49)
Yeah.

Stefan (19:51)
or at least not double count things. look, we end up building products that lawyers actually like using. And our pricing product, doesn’t even look like a pricing product. It’s a chatbot.

Yeah, look, it’s not that entirely different to the vast majority of successful ⁓ legal tech products of today. ⁓ And the way, you know…

Marlene Gebauer (20:20)
It looks like one I

use every day, whose name shall not be mentioned.

Greg Lambert (20:23)
You

Stefan (20:23)
Absolutely

right and like but look there’s the point right like if we learned one thing over the past 24 months is that this is the format that really resonates with with people in legal And I mean look, know attorneys literally spend the days reading and writing and then suddenly have a piece of tech that Not only reads what you say but also writes back which is just pretty amazing And you know the way this works is that you can basically ⁓

ask the child book to build a proposal for a new matter, can describe this matter, can give it a document that describes this matter. And then what happens is the agent behind it, goes into that data that we cleaned, know, that matter data and that timesheet data inside the law firm, it finds the right precedence and then it basically uses those precedence as like, as a starting point for a new quote. And it basically gives you a full client.

entire proposal in something like 90 to 120 seconds.

Marlene Gebauer (21:28)
So I mean, I want to explore a little bit of the logic about ⁓ sort of relying on historical data. ⁓ You’ve mentioned that, yes, we go back to historical data. You mentioned sometimes the data is not great. And AI is sort of changing matter economics continually and not just.

It’s, you know, it’s changing the process completely in terms of how we may have done something before. So, ⁓ shouldn’t we be looking at that as sort of the, the, the, the baseline, but you know, here, this is how we’re doing it now with AI and, and, and how do we improve from there as opposed to historical data? You know, isn’t, isn’t that, I mean, I know that’s, that’s traditionally what we use, but isn’t that irrelevant in a way?

Stefan (22:21)
Yeah, I look, I love this question personally because I think there’s something very true about it. But I also think that the answer is quite nuanced and it’s a way for me to show off about how deeply we fought for this problem. It’s true, of course, at the point where…

AI or AI plus re-engineered workflows, the moment they change the way that matters are delivered, really, then yes, this historic data might become irrelevant. But the good news is that our agent can actually take that into account. That’s kind of covered. And the reason why it can do that is…

Because we clean the data, because we actually can see what a given matter which meets certain criteria, displays certain features, how it behaved five years ago and how exactly the same kind of matter behaves today, we can actually begin helping people figure out to what extent is AI actually changing things on the ground.

And I think it’s a bigger question. think it’s not just the pricing question. It obviously has big implications on pricing, but there’s more to this. And we actually are spending a lot of time with our customers, like digging into this.

And the conclusions are interesting. I mean, this is a space which is changing really, really fast. Adoption inside big law firms is like skyrocketing in terms of AI in general. So if you look at things that happened three months ago, the reality might be very different today. But what we are seeing is that

sure, some things are definitely changing and I think that’s especially true when you think about like allocation of tasks within the process but I think overall, matter economics, know, they’re not changing dramatically just yet so it’s one of those where I say usually like, you watch the space and we’re basically ready for this to happen.

Greg Lambert (24:15)
Well, Stephan, this morning I was actually doing a presentation for a group of law students and we were talking about knowledge management, I think things like data quality and knowledge management, one of the 25 year overnight problems was we’ve got bad data that we’ve worked with. And when it comes to, especially to pricing, to time entry, to billing, we’ve got

Every individual lawyer has their own method of how they enter time, how the client wants that in there. so that garbage in, garbage out kind of problem has been another 25 year overnight problem with AI. And so with the…

I understand that you’re leveraging the large language models to help you kind of clean up the unstructured time narratives. So can you walk us through kind of how you’re able to take the individuals, everyone’s doing it differently, and then creating an output that makes sense on the other end? Yeah, consistent.

Marlene Gebauer (25:22)
consistently.

Stefan (25:23)
Yeah, yeah.

mean, look, that’s basically the foundational question, right? And like, we actually spent like an awful lot of time looking into this way before we even built a pricing product because we knew that this was going to be a major problem, right?

The way that I think about data quality inside law firms, I tend to focus on two things, which I think are definitely not the entirety of it, but those are two very important aspects to it. One is your Matlab data set and the other one is your time data set. They’re related, but they kind of describe two different things. So at Matlab level, you really want to be able to somehow look at a Matlab entry in a database and be able to tell what it actually hits.

someone will work with, he leads business development at a really large international firm. He said, look, frankly, if you go to most law firms and you ask them what it is that you do for your specific clients, there’s no practically good answer to those questions without going back and talking to individual partners who own those relationships.

And that metadata is kind of the foundation for that. So if that doesn’t exist, then you can’t really compare apples to apples. It’s very difficult to find the right precedent. And so we need to fix that. And then the second thing is the time data sets. You write time narratives can be very messy.

they definitely are not consistent. But I guess, you know, the whole idea behind task and phase is that that breaks consistency, right? And if only people use that properly. So we had to deal with this as well. And so our system actually, and look, this is quite technical and sounds a bit boring, but I think it’s just so important. We spent probably two years building a system that creates metadata about matters and about time entries.

Greg Lambert (27:06)
We’re huge fans of metadata here. So

you’re talking to the right audience.

Marlene Gebauer (27:09)
Yeah. Yeah. It’s like, totally have a follow-up question when you were just talking about phases

Stefan (27:09)
Yes, hi, this.

Marlene Gebauer (27:13)
and tasks. Cause it’s like, okay, what about clients that have sort of whole different, you know, schema about these things? Like, how do you marry that?

Stefan (27:18)
Yeah.

I mean, look, this is why it takes so long to build products like ours, because, you know, like the high level problem is quite hard, but when you actually start going into the detail and figuring out all the like exceptions and edge cases, it becomes a nightmare, But fundamentally, you know, the concept that we had is pretty simple, and it’s actually not even that new. We reckon, know, consistent or not, these time narratives tend to be a pretty good descriptor of like what’s happening overall.

So if you get really good at reading time narratives, then you might be able to find out quite a lot about, the very least, the matter itself, and maybe enough to then go down to individual time entries and deal with them. the way that we do it is we essentially extract signal from those data sets. And I think the difference in our approach versus things that have been done in the past is that

A, we do leverage frontier models in extracting that unstructured textual data, but we also developed a lot of our own proprietary technology on the back of that. And this is largely…

predicated on our ontologies. So for every work type that we cover, actually pre-build a set of our own ontologies that describe exactly how the system should interpret what it’s seeing in the time sheets. And we then have a host of quality control.

processes and part of this is bit more traditional, part of it is agentic to ensure that what we produce is actually correct. And the bottom line is that we can produce between 10 and probably 30, 40 fields about a matter and then we can reconstruct task and phase using your own taxonomy or any taxonomy that you want us to use. And in the tests in the wild, we achieve around 95 % accuracy rate, which is staggering.

I mean, one of our clients said this is close to science fiction. So we’re super proud of that.

Marlene Gebauer (29:22)
you know, you’re saying moving away from the billable billable hour is, isn’t just a financial shift. mean, yes, of course it is, but, but.

Really, it’s a massive cultural shift. How do you get people to rethink how they charge for their work when this has been the norm for so many years? How do you wean people off the clock? And dealing with, and I’m sure people in internal firms, they’re thinking about this.

The cultural friction that’s inside firms when they try and change the models and how do you deal with that? Is it the hardest part like actually building the AI? And I think I know your answer to this, or is it convincing the partner to trust the recommendation to think about how they’re going to use AI and change their whole billing practice?

Stefan (30:18)
Yeah, that’ll just not. Look, mean, in all honesty, it’s a bit of both. know, building the product is not easy, especially if you want to make it work in the wild, like we just discussed. But I think, you know, that cultural change piece is, I mean, it’s huge and it’s like really, really important.

because frankly, yeah, exactly, because you can build the best products in the world, the most fancy stuff in the back, and if nobody looks at them, then you’re lost. And we spend a lot of time with Lovers to help navigate that, right? And I think there are a couple of things. think there’s…

Marlene Gebauer (30:38)
Get right.

Greg Lambert (30:39)
Yeah.

Stefan (30:56)
Some element of this is about building trust and about showing people that they can actually trust a computer to come up with something that makes sense. And part of it is about, as you say, changing habits and helping people adjust to a new way of working. And again, this comes back to conversations. We spend a lot of time with the lawyers who are our users.

Marlene Gebauer (31:18)
Do either one of you have a psychology degree?

Stefan (31:21)
I wish I did sometimes. funnily enough, our chief of staff, he’s actually an ex-Mayor Brown, corporate finance lawyer, and he spends a lot of time talking to our users. Oftentimes he jumps on those one-on-one calls with partners that we work with, or with others across the firms.

think that helps for sure, this kind of you know, cultural affinity and this shared experience of, you know, he’s been in other people’s shoes essentially. ⁓ Yeah, for sure. And I think, you know, I think increasingly what we’re seeing is that…

Marlene Gebauer (31:51)
Being a peer goes a long way, I think.

Stefan (32:00)
you gotta make sure that you leverage everybody across the firm who can be your ally. So I think we touched on at the beginning, a lot of these initiatives come from somewhere from within the center, from the CFO or someone who’s in charge of operations or pricing. And you may as well use the resources that they have at their disposal to help you.

Buying tools is honestly not the answer, it’s just part of the answer and it should be delivered as a joined up change strategy.

Greg Lambert (32:31)
And I’m curious because we’ve talked about a lot of the culture of the firm. And one of the things that I stress a lot when it comes to like AI adoption is it’s not a generational thing. It’s more of a curiosity thing, the ability to the willingness to be to change. But I’m wondering when it comes to changing something as pivotal as the billable hour.

⁓ Are you seeing a generational gap? Are you seeing a creativity gap? What’s kind of the demarcation on acceptance versus defiance?

Stefan (33:05)
Yeah, mean, we think about this a lot because we’re obviously interested in who are the people that spend the most time in our platform. And maybe there’s some generational aspect, but I honestly don’t know. mean, many of our super users come with quite a bit more grey hair than I mean. And I think it is about curiosity. And I think it’s also about…

Probably in general, the broader interest in what it takes to run a practice. think at Everfam there are lawyers who are much more interested in that kind business side of things than others. And I think if you can identify those and make sure that you take them on the journey early on, they can be really powerful allies because they are those peers that you can then…

Marlene Gebauer (33:50)
So ⁓ before we get to our crystal ball question, we’d like to know kind of aside from your use of AI at work, what are you personally using AI every day for to kind of help you on that individual level? What kind of cool thing are you doing with it?

Stefan (34:08)
My god, mean, yeah, a lot of meal planning for our two year old, but it’s, I honestly feel like my life is increasingly run by like a bunch of sister prongs that I wrote for myself. And I mean, it sounds a bit scary when I say it like that, but there are so many great use cases. And, you know, I think when you’re running like a lean tech company,

Greg Lambert (34:15)
Thanks for watching.

Stefan (34:31)
They say that necessity is the matter of all inventions. And I think it’s definitely true in my case. I’m not an engineer, but I really enjoy playing with these things and getting to understand at a slightly deeper level how they work. And so I actually automated a lot of the things I used to spend a fair amount of my time on. So that could be things like…

dealing with law firm procurement processes or like record management within the company name, but then it does seep into your personal life. you know, whenever I’m dealing with something that’s difficult, whether it’s at work or somewhere else, I do end up using AI as a bit of a sparring partner. And I also learned to use agents to deal with a lot of the treasury that comes with running a leading startup.

Greg Lambert (35:21)
Do you have a tool that you’re using?

Stefan (35:26)
Notion, hands down, is by far my favourite.

Greg Lambert (35:27)
nice.

All right, well, Stephen, we have reached the point to where we ask everyone our crystal ball question. So I want you to pull out the crystal ball and from your perspective, take a look in there. And what change or challenges in the legal industry do you think we’re going to need to start addressing sooner rather than later?

Stefan (35:50)
I’m not sure if I’m going to be particularly original when I say this, but I think this is something I think about a lot. people have to figure out how to live with very different gross margin dynamics. Right now it’s a relatively well understood concept, but I think it is going to get messier and I think…

it’s not going to be just the lowest time that goes into direct costs in the future. You will increasingly see tech being part of the mix and especially LLM costs. And I think that is going to create lot of complexities. And I also think that there will be a much greater disconnect between ⁓ the inputs and the outputs, which again, will probably cause some consolation and I’m sure a lot of innovation as well.

Greg Lambert (36:34)
Well, Stefan Cisla from IORRA. I want to thank you very much for ⁓ coming in, having this fascinating talk with us. We love talking metadata and all the nice geeky stuff.

Marlene Gebauer (36:47)
Yeah, thank you.

Stefan (36:48)
Thank you. It was amazing to be here.

Marlene Gebauer (36:51)
And thanks to all of you, our listeners, for taking the time to listen to the Geek in Review podcast. If you enjoy the show, please share it with a colleague. We’d love to hear from you on LinkedIn.

Greg Lambert (37:01)
And Stefan, if listeners want to learn more about you or Iora or read the report which talks about that $36 billion value gap, where’s the best place for them to go?

Stefan (37:14)
Yeah, you can find us on ayora.ai. that is a-y-o-r-a.ai. Hope to see you there.

Marlene Gebauer (37:23)
And as always, the music you hear is from Jerry David DeSica. Thanks so much, Jerry. Bye everybody.

Stefan (37:28)
Bye.