Why We’re Entering an Era Where AI Helps Lawyers Instead of Replacing Them - Wendy Jephson - S10E09

0:00Rob:Your partners don't always have the answers immediately, so how do they know? Somebody can sit down with our tech, it asks questions, so it's like the opposite of ChatGPT. It asks the questions and gets people thinking, and dives down into getting that information out of their head in a useful way to get them thinking. We've got proprietary models to analyse what they've said, to start to really pull out and highlight what is the tacit knowledge. When can people have those opportunities to start to think more deeply? That's the beauty about automation. You've got more time. You can think more deeply.

0:29Rob:You can take the time to unpack that value in your head and think more carefully about how you can share it with somebody else.

0:36Wendy Jephson:On today's Legally Speaking podcast, I'm delighted to be joined by Wendy Jepsen. Wendy is the CEO and co-founder of Let's Think. She is a dual qualified lawyer, business psychologist and serial tech entrepreneur. Wendy co-founded a fintech using behavioral science to detect financial crime, which was acquired by NASDAQ in 2017. She created and led NASDAQ's research and ideation function, developing behavioural science-led products to help financial services firms detect anomalous behaviour. So, a very big warm welcome to the show, Wendy.

1:09Rob:Hey, Rob. Really great to be here. Thank you for having me.

1:11Wendy Jephson:Oh, it's an absolute pleasure to have you on the show. Before we dive into all the amazing things that you've been getting up to, we do have a customary icebreaker question here on the lead stream podcast which is what is your favorite beverage and what is your preferred choice of footwear on a typical work day interesting um well coffee obviously

1:31Rob:and if if it's not that it's a espresso martini if it's the other end of the day also coffee um footwear interesting i mean socks if i'm at home and comfy white trainers obviously if you're what you're walking the streets of london there we go like that so coffee and white

1:49Wendy Jephson:trainers gets us through don't want to spill it mind um without going into too much of a rabbit hole let's talk all about you so to begin with would you mind telling our listeners a bit about

1:58Rob:your background and career journey sure so started out as a lawyer um decided about 12 that that's what i wanted to be so and went off into the big smoke trade in trained in london in what was richard's butler is now reed smith but quite early went in-house so i went in i had as a comment to Xerox, went in-house as I qualified, and then to Lily's Pharmaceuticals, where it was for most of my career. And that's why I got so interested in decision-making and judgment in real-world practice, because law is all about risk, that I decided to go off and study that in depth. So, went back to university, did a psychology degree and a master's in occupational organizational psychology so business really focused on how to make decisions people make decisions in the workplace and how could you help make them better and technology is a great way of thinking about doing that and this is 20 years ago so technology data people understanding how people make decisions sort of in its infancy although there's a huge amount I mean so computer science and behavioral science goes back 50 plus years clearly in these fields but 20 years ago it was you know much much harder than it is today.

3:11Rob:So roll forward, come out, looking at financial decision making first, because actually there's loads of data to measure decisions and you want to know you're going to make them better if you're trying to intervene. And looking at asset management, there was a pivot there in that we were doing some great stuff looking at decision making of fund managers and the head of compliance then, and the client that we were working with said, look, if you're looking at how people make decisions, I want that from a compliance perspective too. So I want to be able to defend them. They're actually doing what they should be doing and beating the market.

3:42Rob:And I want to get them out if actually they're doing something wrong, like insider trading. And that was a product that took off, led to an acquisition by NASDAQ in 2017, which was great. There I was head of research. I set up the research and ideation function. So pulled together behavioral scientists, data scientists, technologists, the business people working with their clients to really tackle some of the really challenging problems there by digging into the problem space. And then finally, left there with a decision to go and do it in the startup mode again, a bit faster, and with a great team of people at Let's Think, which is where we're really trying to think about technology to help the people of the world think brilliantly.

4:26Rob:So great goal. And starting in law, because lawyers are thinkers. And how do you support that decision and judgment in the world of AI in a law firm?

4:37Wendy Jephson:Yeah, and that's perfectly where I wanted to go next, actually. Let's Think, because you are the CEO and co-founder of that. So can you talk us through the process of actually building Let's Think and what inspired you to actually co-found the software company with a vision to help the people of the world, which I love?

4:53Rob:Yeah, so it really was. There's lots of technology out there that I love. So it's like a passion and a frustration in a sense that I love thinking and using technology, but also it gets in the way of thinking quite a lot. So I felt that there was something that if you could design it better, could actually help you think more productively in a better way, in a richer way, like experts, because those are the people we've been studying. How could you get to that faster and design technology to help somebody and fit it into the workplace? So there's lots of technology designed generally like Microsoft and Google.

5:30Rob:And then there's lots of work specific technologies. We wanted to actually have a combination of the two. How do you actually really understand people and how they really think in the workplace? And then, of course, you have a layer on top. So what is it lawyers specifically need? What might accountants specifically need? And then it's about who do you do it with, isn't it? It's a hard thing, as you will know, it's a hard thing. Setting up a company for nothing, especially being something new. You know all of that. So you've got to do it with the people that you want to run that journey with. and my co-founder Anna Leslie who worked with me in NASDAQ is the best behavioral scientist on the planet you should have her on the podcast and I kind of left now that she left NASDAQ at the same time as I did and I said look I'm I'm going to set up this thing there's no guarantees in it there never are and so people who want to have to want to be able to want to come and work with me and I want to work with them and so it's your it's a free choice to everyone except for Anna Anna had no choice.

6:29Rob:She had to come and do it with me. And so there was Anna, first of all, Claire Fisher, who's also our COO, who's also amazing. She got us our B Corp, which is, you can see it behind me. And NED that I worked with before as well, Victoria Raffi, who is also fantastic, because actually having that structure, you know, wisdom and structure and governance is really important when you're a messy hack stage. And then Mike, our CTO, who has got a fantastic background in data science and technology, has been in healthcare, done lots of cool things with AI. And you need the right kind of person, again, as a tech person who can get that, roll their sleeves up and start building and can cope with all the ideas that are flying around.

7:14Rob:So, you know, that core team was really important and continues to be as we kind of forge our way forward. so that's that's kind of let's think and we have other members of the team who are also fractional

7:25Wendy Jephson:who are also brilliant obviously and fractional is a growing trend and it makes a lot of sense for for businesses as well and you know that team is important we is greater than me isn't it and i think the more you can kind of bring all these great minds and great experience and come together and gel was brilliant you talked about behavioral science wanted to get a little bit deeper because let's think it's a very different approach from the traditional legal ai tools out there so can you explain to our listeners how but your behavioral science ai actually works yes absolutely

7:51Rob:and and that is you know the the most frequent comment we get when we show people our tech is i haven't seen anything like that before and so it starts from and the behavioral science methodology that we use has been proven in multiple domains over the past we've been applying it for 20 plus years but the kind of the evidence base is at least 40 years old now but it's been manually done so it's people interviewing and elicit you know using different methodology to pull information out of experts heads the stuff that makes them really good at what they do but actually at that point it's hard for them to articulate it so it's that stuff that's hard to write down and there's various different methodologies you know you kind of people go what questions do you ask but it's a method rather than a set of questions because you have to know how to ask when to ask quite a difficult thing to do which is why i'd leave it to anna most of the time but we'd used it to great effect she's used it in government law enforcement intelligence agencies i've used it in financial services it's been used in health care and the military and aviation high-risk industries where if you get it wrong people die um and what it's really good at if you think about um kind of person-centered design design thinking at the problem space analysis end it's like gives you the tools on steroids to go and really understand that domain and the problems and how the experts navigate that usually quite well but they can't tell you and a junior person doesn't know what they don't know so they definitely can't tell you um so it's really good and we that's what made us successful entering government me designing technology because you get amazing requirements from it in a way that you can understand how to fit it into somebody's end they can use it whether they're an expert or a junior.

9:38Rob:And that's the point. You have to be able to support somebody without interrupting them when they know what they're doing, but give the scaffold and support to somebody who's junior. That methodology, you know, is a manual process. It's really slow. It takes an age to analyze the data you collect. So along come LLMs and we think, can we automate ourselves and what we do? So it's not identical. We get people in a room and interview them for two hours on work that they've done. Not practical, hard to get at. takes us days to analyze what they say. So we thought, okay, can we train LLMs to ask the questions and probe, provide the memory probes, prompts and probes that we would do in order to help them remember things, to dive deeper into what is it they know and what they were noticing via an LLM in a 10 to 15 minute chat.

10:25Rob:And that's what we've designed at one end. So somebody can sit down with our tech and it asks questions. So it's like the opposite of chat TPT. It asks the questions and gets people thinking and dives down into getting that information out of their head in a useful way to get them thinking and then it can analyze it so again then we've got proprietary models to analyze what they've said to start to really pull out and highlight well what is the tacit knowledge what are the cues in each moment that they're paying attention to I was speaking to somebody today and it's like you know a partner you they can be on a client call and they know exactly what to pay attention to and what a client's saying a trainee's desperately scribbling notes and has no idea what they're listening to because they just haven't viewed into it so that's what you can start to reveal so it's about revealing what is it that somebody knows and what are the important things about what they know that make them effective and then you can turn that into useful artifacts and knowledge assets like a file note or a learning guide or your can make you sort out your sra competence records trade you know all sorts of materials like billing narratives can all come out the same reflective session so you can start to you know use that content in multiple different ways which of course is what we do as a human you have the content in your head and then you reframe it it's just right now you have lots of point solutions that do that whereas if we can get the content out in the right way we can also reframe it into lots of different useful things super powerful and super interesting so thanks for

11:56Wendy Jephson:for sharing that i want to talk about investment in ai now because there's been a real significant growth in global investment in ai especially in legal services we've seen some huge headlines in in recent months even sort of coming off the blocks um you know in 2026 so when did you realize the biggest risk isn't that ai will actually fail but it succeeds in the wrong way

12:16Rob:oh that's a really interesting question i'm going to ask you to say a bit more about that so you said investment first and then you said yeah like there's been significant investments

12:26Wendy Jephson:in legal services particularly firstly do you have a viewpoint on that and some of those sort

12:31Rob:of investments um generally yes well i suppose it's been a an industry like many professional services that's been massively under invested in in terms of technology so the fact that there is this influx means the industry can change rapidly and on mass and the fact you know one of the one of the key things about you know trying to get innovation and transformation is nobody wants to go first and the fact that this has kind of been dropped on is like well okay we've all got to do this haven't we I mean what is it there's a Clio stat isn't it 96 % of law firms are engaging with AI so everyone's doing it so that's you know everyone's okay to do it then aren't they so that's that's one part of it and then the the risk I think of it is and this is where I suppose we see the risk particularly is if you're the massive focus you will have seen is on automation so what happens about developing people's cognitive skills that's where the real risk is and there's lots of you know research anyway about how what the impact it has in terms of your cognitive functioning and development and different ways you can address that but what we'd already seen from hybrid working in covid is this lack of transfer of experiential know-how and knowledge from partners to juniors that's been going on for you know centuries isn't it i sat in the room with a partner and try to think what's going on.

13:56Rob:But that had already been noticed. And you're at risk of that happening even faster, because lots of these AI products make it look super easy to come to a conclusion as if it was always done. Whereas most of the valuable work in law is figuring it out. And there's an awful lot of invisible thinking that you need to get access to if you're going to learn that your partners don't always have the answers immediately. So how do they know? in fact I was speaking to a partner today and she said you know something she finds useful is often you come to a conclusion really quickly and then you have to actually take a few steps to figure out how have I reached that conclusion and that's sort of unpacking your own expertise quite hard when you haven't got somebody and that's what she likes using our tech for is to actually go okay so I know what to do but how do I know what to do and what are the steps that actually have got me there let me unpack that and then I can see it written out and that's quite useful because I can use that as an explainer to a client as well and that's the value bit that actually sits on top of the automation piece um so that's the biggest risk I think is that you're automating away the opportunity to develop your future partners yeah and that was exactly

15:08Wendy Jephson:where um you know I thought you would sort of highlight and I know you talked about that and just going a bit more on automation because many law firms you know have embraced AI we talked about that's that you know to improve their speed and efficiency but what are the consequences when automation is introduced without considering how lawyers learn and develop judgment exactly well

15:29Rob:so judgment is always about making decisions under risk or under uncertainty which means that you are trying to understand what do i know what don't i know and actually i can't cover all risks you never can so which one can I live with and which ones can't am I not going to and how do I explain that to the client because I'm not making the decision as a lawyer I'm handing that to my client and that's what I felt when I was in-house that's what was interesting is that I was always advising the clients within the business of what they could go and do and where the lines were and where the gray was and they then had to take on the risk and you that's what was interesting in seeing how different people cope with that so actually and so often you just kind of go well here's the answer well how did you get to that what is it I really need to take into account what am I likely to hit up against and what was interesting so we as we were obviously we're licensing with law firms so we needed an agreement to do that you know I could have written that myself but I thought this is a much better idea to get somebody who's you know currently practicing to do this to write this and give me help on how am I going to negotiate with law firms?

16:44Rob:Because clearly they're quite good at this, aren't they? And so they gave me the usual, here's your contract. I've put some comments in the side and, you know, that came over to me. And then I asked them to use our tech as well to think about what they've done. That was the most useful piece of information. Clearly I'm biased, but it was a really good test because what I got from them were, okay, before it was, have you just taken the precedent and just tweaked it a bit? and how much am I paying for that? And by the way, you've got two lawyers on here. We've got a technology lawyer and a data protection lawyer.

17:15Rob:Why have you got two? Why can't one do it? Both of them kind of thought, spoke aloud to the tech and pulled out what they'd been thinking about. And they hadn't written that stuff down to me. So what I then got was actually the technology lawyer had been really thinking strategically about me and the power, the position of, let's think at the moment, compared to law firms, where I was going to have to negotiate with and what the law firm is going to be thinking about in terms of covering their own risks and how I could deal with that, where I could go with that. And the data protection lawyer was really, you know, technically thinking about the detail because it's all in the devil's in the detail of the wording, isn't it?

17:53Rob:And so that actually, to me, you know, for them both using it for five, ten minutes each was massively valuable and made me think, actually, I could go and have a conversation with them again. You know, I would want to do that and i'm okay paying their bill now as much as anyone's happy paying a legal bill and that's the challenge isn't it i think that's the real challenge right now is it's so easy it's like doctor's google moment isn't it hello doctor i've googled my my symptoms can you prescribe me x y and z and people are i talking to a cio yesterday saying clients are saying look our bees are bees bees are fees are going to come down aren't they because you're using AI so you must be faster and it's not all about being faster so I can produce the asset you know the document quicker but actually my thinking time is probably still going to be what you want and that's what you're really paying for and that's still going to be in the mix and that's how do you explain that though?

18:51Wendy Jephson:Well yeah I remember people saying to me look Rob we're drowning in information but we're craving wisdom and it's those years of wisdom isn't it that we're looking for that expertise um we touched on this but sort of adopting ai early you know there are adopters reaching sort of 18 24 months since implementation so what are some of the consequences now of you know that's becoming more visible versus positive versus negative what have you

19:15Rob:been seeing generally for those early adopters i think i think it's um an appetite to try as much as anything it's uh okay this stuff is in in my workplace now and i will see what it can do and that's the hardest thing especially with super busy people who's every you know six minutes is accounted for just getting somebody to go let me all right then i'll click on that thing it's pretty hard isn't it so i think the acceptance that these these tools are out there and some are better than others and they can be surprisingly good I think and that you know the fact that there is that delight moment that some of this stuff does make your life better is really helpful and the fact the most powerful recommendation is your peers talking about it isn't it you know it's not companies it's not us lot going this is brilliant I'll change your life because of course we're going to say that and you shouldn't listen to us you'd listen to the people who've actually used it who do the job you do.

20:18Wendy Jephson:Today's Legally Speaking podcast episode is proudly sponsored by Clio. If your legal management software feels more frustrating than helpful, you're not alone. Many solicitors across the UK delay switching because moving client and case data sounds like a headache. Clio makes it simple. As the intelligent legal work platform, Clio combines context-aware AI with trusted legal research to power your practice. And when you switch their dedicated migrations team supports you every step of the way ensuring a smooth transfer of your data you'll also get award-winning support 24 5 via live chat phone and email so help is always there when you need it ready to leave frustration behind visit clio.com forward slash uk to learn more and discover why so many uk solicitors choose clio now back to the show once you see the result your eyes light up don't they you really get the buy-in you can genuinely believe this is going to save you time or make you a much better lawyer or whatever part of that workflow it is and that that's the key point and building capability where i want to go now because you've shared and i like this firms that use ai to build capacity rather than hollow it out will gain a real competitive advantage so what does this look like in practice for junior lawyers

21:41Rob:and trainees yeah and i think this is really key i mean in professional services businesses where your people are your product aren't they yes very much competitive advantages how do you make your people as good as they can be how and being as good as you can be in a way that you enjoy it in the workplace so you stay so you don't have the cost i mean is it two two to four hundred thousand dollars to train a lawyer it's a lot of money invested and then they go write them off because they haven't enjoyed it or you haven't got value out of them yet so really important to think about how you get value out of them is gets quality quicker as we like to say and what so we start from the premise having understood you know done a lot of work of how does expertise develop and traditionally it's developed and it's always domain specific you have to focus in on a specific area so you can that's why somebody can be brilliant in in you know in sport you're brilliant as a ping pong player but you're terrible as footballer it doesn't transfer and it's the hours of practice you put in so that's where it comes from doesn't it and same is true in knowledge work um and the way it develops is by exposure to lots and lots and lots of different instances of the same thing and you know that's where the experiential work comes in what we know from the science is that if you can get at the mental models for people who are expert who've had that exposure you can share that with juniors and they can learn faster because they start to get their eye in on what to look for you know it's like um so i think in tennis that you know when you're the people who are on the receiving end aren't looking at where the ball is thrown they're looking at hip movement and stuff like that because that tells you where the ball's going to go to much more than looking at the ball um and it's the same thing with lawyers you know what is it they're paying attention to in a risky conversation they're tuned into it so if you can give them the mental models of what are the key things they're looking for then they can start to look for them it's the signal in the noise that you can help them with so and that's what you know was being lost or not transferred it was it was that's why there's this kind of learning biosmosis supervisory models learn alongside which has never been perfect because you can have somebody who's as good as a chocolate teapot as your supervisor somebody else's phrase again which I liked um so you know some people are better than others explaining things aren't they they're not trained to be teachers so so it's that premise well can we get out examples that are really good from people's heads in a way that helps them also see their own expertise because experts undervalue what they know but in a way that those juniors can start to see what those experts are seeing see through their eyes start to look at uh those that you know those kinds of things that they're looking at in client meetings in the work that they're producing and and and if they see work as it plays out as well they see how messy it is and where things go wrong and how people get tripped up without having to go through it in real time which takes years so you can build up that exposure to this is what real work looks like i can start to develop my own expert mental models because those lawyers of the future are not going to be like the ones today they'll work in a different way they'll use different technologies they'll think differently but they still need to get exposure to you know those key signals that experts have reliably used to spot the risks in the first place so it's how do you get that out and transfer over in the new way of working with the new technologies well that's it you know they are going to be different

25:14Wendy Jephson:careers you know putting my legal recruiting hats on you know nothing changes nothing changes they're going to be a new evolution of you know new things that are going to be out there and ways and i'm excited for it and let's stick with the human then the human and ai collaboration because what does good human ai collaboration look like in a law firm and how should law firm leaders be thinking about redesigning workflows to maximize efficiency and expertise yeah so so we look at it and we very

25:40Rob:much look at well how do people currently think and work because that's where we want to get in the support um so and it's quite hard you know people have very varied days don't they so it's not like you can go well and the previous tech i designed you had a compliance officer sit down in the morning and they'd work through a list and you know you see that was easy you had a pretty structured workflow it's very different when somebody's jumping from lots of tasks so you're really looking at the okay what are they thinking about in any moment in time how do you make it It's quick and easy to use for that sort of thing.

26:15Rob:But then it's, okay, when do they think, you know, when is it that they are perhaps have come off a call and they might have said certain things in the call that are captured in a transcript. So technology is great for capturing that, but they probably thought a few things without saying them, that they actually want to kind of unpack in the risk. And that's true in any meeting, isn't it? Not everybody speaks, but they've probably got quite a lot going on in their head. So again, how can you help support people to get that out of their head in a safe space for themselves, first of all, documented in a nice structured way, in a way they might then feel comfortable sharing it with somebody and discussing it with somebody.

26:53Rob:So I think it's really thinking carefully about how does this work really happen when you're in a thinking environment? When does it happen? When can people have those opportunities to start to think more deeply? And that's the beauty about automation. You've got more time. You can think more deeply. You can take the time to unpack that value in your head and think more carefully about how you can share it with somebody else. We had, again, a conversation today with a partner who was saying, OK, so actually now I've produced this. I did a think aloud because I wanted to unpack a knotty issue. And I wanted to, like I say, get to how did I get to that conclusion that I got to very quickly?

27:30Rob:actually now I can turn that into a learning note and I could share that with my supervisory trainee before I go and meet with them and I could give them something to think about maybe that will improve our interaction and that's exactly what we want it's actually can you get the kind of thoughts out in a better way so instead of running to that meeting and neither of you you know in the meeting you're trying to figure out what you're talking about and structure it you both come better prepared with better thoughts and you have a better conversation in that meeting and I think that's where the opportunity for technology can really lie in getting things ready faster so you can the person-to-person time is more effective.

28:08Wendy Jephson:Yeah you're enhancing the overall quality and you're making it more efficient aren't you and I think that what's there not to like we were sort of touching on before around billable hours and client pressure you know clients are increasingly questioning why they should pay for work that AI can allegedly do. So how do you see AI reshaping billing models and how can human expertise continue to add value to the client outcomes?