Beyond Generic AI: Luminance's Specialized Approach to Legal Contract Intelligence
21:05 Share

Hello and welcome to the Legal Tech Startup Focus podcast. I'm your podcast host, Charlie Uniman. On this podcast, I'll be interviewing the people who build, invest in, comment on, and use the apps made by legal tech startups.

My guests and I will be discussing many different startup related topics, covering among other things, startup management and startup life, startup investing, pricing and revenue models, and the factors that affect how users decide to purchase Legal Tech. We're not going to focus on Legal Tech per se. Instead, we'll be focusing on the startups that develop, market, and sell that tech. So whether you're a startup founder or investor,

a lawyer or other legal professional, or a law professor, law student, or commentator who thinks about LegalTech startups, sit back, listen, and learn from my guests about just what it takes for LegalTech startups to succeed. And if you're interested in LegalTech startups and enjoyed this podcast, please become a member of the free LegalTech Startup Focus community by signing up at www.legaltechstartupfocus.com.

Welcome, everyone, to yet another edition of the LegalTech Startup Focus podcast. I am so happy to be bringing this to you from not my typical perch in New Jersey, but in Key Biscayne, Florida, at the TLTF, the LegalTech Fund Summit. I attended last year's. It was splendid. I'm so looking forward to the programs that begin this afternoon.

here in Key Biscayne. But most importantly, I'm delighted to bring to the podcast as a guest, Eleanor Lightbody, who is the CEO of a scaling legal tech company called Luminance, working in the field of AI and contracts. Welcome, Eleanor. Charlie, thank you for having me. And we are here in a very large conference room where no one else is. It

but Eleanor and I, and we seem to have a nice quiet place to record. - We do. - So Eleanor, if you could tell our listeners, many of whom I believe are probably well aware of Luminance, but for those who happen not to be, tell us what Luminance does. - So Luminance was founded in 2015 by mathematicians from the University of Cambridge. And it was founded because Adam and Graham, who are the two founders,

They had a lot of friends who were lawyers and at the time they would go to dinner parties with them, they'd go for drinks and they felt that there was a consistent theme that was happening which was a lot of their friends were remarking on the hours that they were putting in to pretty repetitive work and pretty kind of manual work and so

And I really thought, well, machine learning, that's what we used to call it at the time. Obviously now it's AI, but AI is going to help humans understand data and specifically contracts at much faster rates. So they built a legal platform that could help

in-house legal teams and lawyers, private practice lawyers, understand anything that related to their contracts. So what Luminance does is that we are a specialist AI company that have built a very verticalized system that understands any interactions that a business has with their legal contracts.

What does that mean? That means that you can use our system to help you create a contract, to help you automate the process of negotiating contracts and then to help you find any key information within your existing contract database.

Right, well that's a wide spread variety of features and as my listeners have often heard me say, I wish I had tools like this when I practiced in New York City. And it's wonderful to have it all integrated because I think so many lawyers are disturbed by having to jump from one application to another application to another application.

And I think you said, if I remember correctly, a few seconds ago that the founders met some lawyers at a dinner party or at a get-together. And I'm surprised the lawyers even had time to do that because given the onerous tasks that can be done by machines...

lawyers often had to do that themselves and never got a chance to get away from the desk and have a life. So you're doing God's work at Luminance, as I've said to other startup leaders who are bringing sanity and more time to lawyers' lives. What struck me in particular as I read about Luminance

was the fact that, if I understand correctly, and tell me if I'm wrong, Luminance has really built its own language model. We speak of GPT, General Pre-trained Transformers. I noticed on your website you have LPT, which is Legal Pre-trained Transformer. Tell us what that means to have developed your own domain-specific model

and how that makes Luminance a little special when compared to some of the other companies in the legal tech vertical that are using more generic large language models. So look, for us, we've always taken the approach that when you're working with lawyers, if you're building an AI system, the most important thing is for the lawyers to trust the output and to trust the accuracy. And so how do you do that?

Well, if you were using just ChatGVT and you had a ChatGVT wraparound, you would run the risk that the answers that it was giving you maybe weren't necessarily factually accurate or weren't, you wouldn't know whether they were trustworthy or not. And so what we've done over the years is that we have fine-tuned our own models and we use a combination of different models.

to solve for a different answer, which is actually if the AI doesn't know the answer, it's okay to hand that back to the lawyer and so that the lawyer can review it because that's by far more powerful and by far more useful to the end user than having to constantly check the answers and constantly not know whether the answers are trustworthy or not. And so what we've done is that

For the first few years, we only sold to law firms. And that was key because what we were doing is that we were building a data set of the most amount of legal contracts and the largest variety of legal contracts that we could that have been tagged by some of the best lawyers in the world. So we had this very rich data set. And then we...

It's so important that we use that and fine tune these models so that the models know when to give you the answer and also when not to give you the answer. And that's really key. Yeah, I remember when I was a young associate working for a partner.

I was always afraid not to have an answer, but perhaps I should have been more worried about having an answer that I was not confident about and then giving it to the partner. And I won't go into too many details about the instances where that happened, but had I been more confident and smarter, frankly, and it sounds as if your large language model has been smartened to do this, I would have said, "I don't know."

And if the wrath of the partner was directed at me, at least it was because he or she knew that something had to be found out rather than that something was wrong. Exactly. And what you can do is when the AI system tells you, so when Luminance's platform tells you it's not sure, you can then teach it.

So the next time that you were to see that same example, the AI would know. And so that's really, really key is that out the box, the AI comes with this pre-understanding of what's typical for you as an organization and what's not. And it kind of highlights and flags key areas of risk or key information within your contracts.

But in time, the more that you use it, the more that it understands your own unique positions, the way that you like to negotiate, what you're interested in looking at in your executed contracts. And it gets smarter and smarter. That's really, really key for two reasons. The first is that we are building institutionalized knowledge within these systems. And that means if someone were to leave tomorrow, you don't run the risk of them leaving with all that information. Like that is there in their...

Yeah, that's there in their systems and in their platform. And second, it just gives you that edge because this is all about how can we automate those high volume, low value tasks and then how can we augment what's left over? And by the AI understanding who you are, it will know the areas that it can start to automate and the areas that it can start adding efficiencies or increasing productivity. - Understood.

Now, you had mentioned, and I understand the benefits of this, of having worked in the early years of Luminance's existence with lawyers. Highly valued knowledge applied to help you train the AI. Is the law firm your market now, or is it more enterprise, in-house, or both as far as Luminance goes?

So it's both. I would say that a lot of, we started with the law firms because we wanted to build legal precedent and we wanted to build trust and validate what we were doing. But in the last three years we launched a product that was much more kind of pointed towards the in-house legal team and that has accelerated so fast that you know we've got the likes of AMD using the product and

like Coke Industries using that product, DHL using that product, all the way through to your kind of mum and pop shop who want to start understanding their contracts in a faster manner. So a lot of our growth is coming from that in-house legal team market.

If our listeners were to visit the website, you would see the testimonials from some of the particularly enterprise users. And it's quite a wonderful roster of testimonials, as well as a list, a sort of chyron running across the bottom that lists

the enterprises and I know having spoken to in-house counsel and at the top ranks, they're under incredible budgetary pressures and yet their work is exploding because in the face of those pressures the financial team at the enterprise is saying bring more work in-house and if they can do that and do it in a fashion that

allows them to be sure they're getting quality results and saving money. They're that much happier. And I think this is a good segue to the next topic I sort of talked about with Eleanor in emails that we might discuss, and that is whether it's a lawyer working at his or her desk in a private law firm or a GC and that person's more junior staff or a business person.

who deals with contracts more often than not, far more often than not. They don't want to learn software, unlike yours truly, who's sort of a nerdy geek and loves to play around with this stuff and did so when he was practicing law. Most of them just want to get their work done.

and yet you have a product that does a lot of complex things. How does Luminance try to get its users up the learning curve or conversely flatten the learning curve so the users can get up that curve more quickly, onboard people so that they can get value more quickly from using the application?

Yeah, look, I think that's a really key thing. You don't want to change the way in which anyone's working. You just want to meet them there. And that helps drive adoption and usage.

And so the first thing to say is that we run what we call a proof of value. So two week free trial that anyone can test the product and in those two weeks our customers are saying whether it's 90% of time savings on the negotiation of things like NDAs or MHAs, whether that's identifying key information in their contracts that they didn't know existed.

A great example would be one of our customers the other day using our AI in the repository went in and saw that there was something that was unusual in their supplier contracts

And they noticed, or AI helped them notice, that of hundreds of very similar suppliers, they had one supplier that they had negotiated an early payment discount five years ago. And it was a 2% early payment discount. And they hadn't been receiving it for five years.

So that's return on investment instantaneously. That's an ROI boost right there. So whether it's time savings, cost savings, finding business opportunities within your contracts or mitigating against risk, people can see that within that two weeks and that builds a very good business case of why they should be deploying it. And why is it such a seamless process from starting the trial to then becoming a customer? Well, because

The first thing is that we meet the lawyers in Microsoft Word. Who doesn't love Microsoft Word? And so all you have to do is press a button within Microsoft Word and instantaneously any contract that you're receiving, even if you've never seen it before, you press the button and the AI will color code

the clauses in the contract that's come in. Green, anything that is acceptable. Yellow, things that the AI doesn't know. As I was speaking about earlier, the like, we're not sure, hand it back to the human. We don't know whether you'd be okay with this or not. And then red, absolutely no one would ever agree to this. You should not agree to it. And then it'll give you language that you should use instead. That's all like meeting you

in the place that you're used to working. It's just augmenting it massively. And then likewise within the platform, all you have to do is, it's almost like Google search within the Lumines platform. You just ask a question and then the AI will surface up that information. So it can be as simple as,

Marketing have asked me to review any contract that was signed last month that has a marketing clause and has written consent in it. And you just ask your question and straight away you'll be shown where those contracts are, who those customers are and exactly what that language is. And then you can obviously get much more granular, much more complex. But it's really...

It's a drive adoption. It's really finding a way to make it as simple as possible to interact with them. So important and remarkably

You're the first startup leader that I've interviewed who has used the words that I often use in my own mind when thinking about what I would like from a legal tech application, and that is, "I just want to push the button." Literally, that is the way I've thought about it. And what an incredible marketing tool that is to be able to say and mean it truthfully. "Hey, you have to push a button. That's it." And you get results. You get benefit.

And to be able to see that you're, as this one enterprise did, one company did, that you haven't been getting something to which you're entitled under a contract for so many periods, years, and immediately know that you can claw that back is, as I think I said, immediate ROI, return on investment. So good stuff, very good stuff. And I encourage my startup leader listeners to

to try to get as close as you can, although it's a tough, tough step to take, to where your application is just pushing a button. Because that's what lawyers want. They don't want to fiddle with software. Leave that to me and to the good people at Luminance. I for fun, they for profit. We're going to...

look at one more topic. You've been in the business for a while. You're leading a company that has raised a good deal of money, I've seen. When you're sitting down with your staff looking at selling a legal tech product to law firms and

a growing market with the enterprise, what are the metrics or a metric in particular that you find helpful to monitor and be mindful of, whether it's a KPI or an OKR or whatever? Look, I think it's really important that everyone has a few key metrics that they monitor the health of their business on. And so that will be different depending on who you are and what's important to you. For us, we focus on a few things. And the first one, which is

How many of those proof of values, those trials, are we running on a monthly basis? We have really great conversion rates. And so if we know that each account executive has to turn up with two a month, if they continue to do that, we know that a chart that reflects our revenue will continue going up to the right. The second is how...

Are our customers getting return on investment? What return on investment are they getting and what parts of the platform are they using? Constantly monitoring that is absolutely key. And really, as a business leader, there's so much noise, there's so many distractions, but for us, it's...

are our AR/RHs growing? And are they growing in a sustainable way? And so the fact of the last two years of AR/RHs growing over 5x, that for us is like a metric that we're constantly focused on because that means that we are doing something and our customers are using us and continuing to use us, which is obviously key. Yeah, I can't encourage leaders of all sorts of startups, particularly legal tech startups,

in line with what Eleanor just said, to find the metrics applicable to their business and the business model that they have used to earn revenue and to derive revenue and earn profit.

And I think I'm going to be posting later today to the Legal Tech Startup Focus community an article that I happen to have come across by coincidence, even though I knew we were going to be talking about this topic that we just discussed, metrics. A very good article by a seasoned...

CEO of startups about just that, how metrics can be lagging indicators, leading indicators, and a nice flow chart describing in very general ways, but helpful ways, what metrics can mean to a startup. So be on the lookout for that. I'll try to get it posted later today.

I'm always jealous of my listeners time so we try to keep it nice and neat and quick and I know you have a presentation to make at the conference later today so I'll let you save your voice although I want to say I'm happy I gave you a warm-up thank you so much and thank you so much for for joining me as a guest on the legal tech startup focus podcast people can reach luminance where at

They can reach Luminance through the website. There is a information request page, but also just info at luminance.com. And the fact that you offer that free trial period is certainly an encouragement to people to give it a try. Eleanor, thank you. I usually say at the end of podcasts that I conduct over Zoom, hope to meet you in person, but they don't have to say that. We've met in person. Enjoy the rest of the conference and good luck at your presentation and

Once more, heartfelt thank you. Thank you. Thank you so much. Thank you for listening to the LegalTech Startup Focus podcast. If you're interested in LegalTech startups and enjoyed this podcast, please consider joining the free LegalTech Startup Focus community by going to www.legaltechstartupfocus.com and signing up. Again, thanks.