High-Stakes, High-Trust: AI's Role in the Practice of Law | Dreamforce 2026

Please welcome CEO of Legora, Max Junestrand, and President and Chief Legal Officer at Salesforce, Sebastian Niles. Hey.

Hello. Wow. Max was looking for some walk-on music, so—

Oh, we were waiting for it. It's great light here.

It's great to be here with you.

Awesome. Oh, that's not us.

Oh yeah, see?

Yeah, it's just noisy. I mean, it's Dreamforce.

They're actually talking about this panel. Oh. And they're trying to get all of Dreamforce across the whole week to come over.

That's good.

Now, if you didn't know, we think of this area of law, high stakes, high trust, name of the panel. But, you know, I was talking to Marc Benioff earlier today and he was just saying how there are moments, right, in industries where everything is changing.

Yeah.

And this is fundamentally such a moment in the broader technology space, enterprise software, AI. And what I also said to him though, well, there's also this very interesting little area of the world of law. And I think this is a moment, right, where the legal industry is undergoing profound change.

Yep.

And one of the areas that we'll unpack today is, all right, who are the players? Who are the participants? Who are the firms? Who are the companies that are choosing to, right, define that change rather than just, you know, sit back and have it happen to them? So, Max, right? You've heard of this, you know, started as a very tiny, right? There's a stealth startup and the like, uh, Logora. I believe in— was it 2023 in Stockholm?

That's right.

You know, at— and he didn't plan it this way— at the age of 23, right? I know you thought he was probably 18, 19, but a little older than that.

I heard somebody yesterday thought I was 43.

Yeah, that's why I had to get— I'm trying to keep up here. So Right. So 2023, at the age of 23, you now have what, over 1,000 users across a whole slew of countries?

So we just wrapped 1,000 employees in the company and we actually did the data. So 150,000 lawyers in the world use Legora, which is about 3% of all the lawyers in the world. And we've only been around for 3 years. So it's gone very rapidly.

So 1,000 employees. So you're scaling. And you say 100,000?

150,000. Yeah.

Is it 150,000? Right. So he's got the latest stats. I have the stats from last night. It's really impressive. So, okay, the area of law. I've always had the view that if you can get right the questions of platform, technology, data, AI agents right in this field, you create a transformation playbook, right, for entire enterprises. You create acceleration, you know, playbooks, but you also are calibrating trust.

Yeah.

Right. And risk in ways that are just so important going forward. So something else that happened. So you may have heard if you were at the keynote, I believe you were, you know, Marc Benioff and Jensen, you know, of NVIDIA were talking and they said, because we can do so many more things now or just do things so differently, might that mean that organizations can be even more ambitious than ever before and that people, individuals, can be more ambitious? You know where this is going. Talk to us, Max, about what is the ambition of Legora? What is your ambition here? And what's the broader scale of impact that you hope to have?

Yeah, well, so to maybe take you back to 2023, Stockholm, Sweden, a smaller legal market than now our biggest client. What we started out with was, let's build delightful software for lawyers that makes the life of being a lawyer more fun. And we ended up with that initial mission statement because we had a couple of friends who were clerking at courts in Sweden, and they were effectively summarizing court cases all over summer. What has happened is legal has in many ways become the tip of the spear for us into broader professional service work. We've started in law and we've now moved into compliance, into risk. And to your point of, you know, when you have more intelligence on tap that you can just magically turn on and let it flow, can you go do more things? And the answer is yes, you can. There will be due diligence done from a legal perspective on every single asset before it goes to IC for an investment conversation. There would be more claims and litigation in the world. There will be more patents being filed. There will be more compliance with regulation that's updating all the time. And I think what's interesting about law in particular is that it's so supply constrained because lawyers and attorneys have this really good idea that in order to practice, you need to pass this bar exam and you get a piece of paper that says that I'm allowed to practice. And that has fundamentally put a cap on the legal services that are available. But the demand for legal services are almost infinite because if you could review every contract, you would. If you could have an agent think about your taxes all the time and make optimizations or have an agent in an enterprise look at all the incoming regulation and how that affects the company, you would. But it's too expensive today. So, I think my job as a CEO is to always raise the bar of ambition. It's to say, we did amazing last quarter, right? Like, the investors are cheering us on, the customers are happy, you know, employees are satisfied. But like, can we do more? And this goes against Swedish culture because in Sweden, there's something called Jantelagen, or the Law of Jante, which effectively means you shouldn't believe that you are something. And so you are not allowed to boast about your achievements. You're not allowed to have big dreams. Like, this is not how you should go about your world. But clearly we are a bit different. And so I like to say we're like the perfect American-Swedish marriage, and we're the latest exports from Stockholm since IKEA and ABBA. And, you know, now we're flying off the shelf here in America.

Oh, it's incredible. So yes, welcome to America.

Thank you. We're loving it.

Loving it. And very soon, welcome to New York. So all those things you described, you know, you got to keep up in New York.

We're coming.

And I see you're coming. You're coming. And on that point of how you're driving forward and charging forward, let's take maybe a little more sober note on this. that may also be exciting. So, there is a point of view that human judgment cannot be outsourced, that legal insight, right, risk analysis, that subtle, hard to, you know, describe or identify question of that judgment, not only is deeply human, but it cannot be outsourced. But so, what are you seeing though, whether it's with that topic, but more broadly, As you bring AI, bring, you know, you know, agentic transformation into these core sensitive workflows, or kind of thinking about, okay, where, where do you deploy the agents? What's the handoff to the humans? Or is the handoff, you know, the agent to the supervisory agent to another agent all overseen by humans? What are you seeing in that space and what are you recommending?

So I think we need to separate sort of like theory to practice here a little bit. Let's start with theory. So I'm an engineer. I was the worst programmer in my team, which is why I became CEO. And what has happened in engineering over the last really 2 years, I think, is enormously profound. Instead of writing all of the software code yourself with hands on keyboard, you are effectively instructing an agent either via typing or with voice, and it goes off and does the work. And in order to do that work, it is making decisions along the line of what best to do, right? It is making judgment decisions about which library should I use where, or should I abstract this thing into an object, or should I leave that as a class, or like, how should I implement this thing? And sometimes the agent that's writing the code will ask you for input. It's gonna say, hey Max, I'm, you know, we have a decision here. Do you want this type of implementation or this type of implementation? And you make a decision, but the agent really does the work. What's nice about AI in coding is that you can verify the results very easily. You can run the code and either it works or it doesn't. That's not how it works in law. You don't know until you're, if you're wrong, Until you're very wrong later on and something, you know, very terrible happens. And so what we've found is in order to get real adoption, we need sophisticated drivers. And smarter, more experienced lawyers are substantially better drivers of Legora because they are able to guide both the agent in terms of which decisions make sense. But more importantly, they're very good at reviewing the outputs of our system. And they can say, you know, if you become a partner in a law firm like you were before, you become very good at reviewing the work of your associates, which is effectively reviewing the work of agents. And so you almost work up that skill of smelling bullshit or smelling, that's the right way of doing something, that's the wrong way of doing something. So the degree of handoff in law is like the bar, to give you a legal pun, the bar is so much higher. That joke didn't really fly. You know, that bar is so much higher in law than it is in coding. So where we're seeing that handoff really happen for the time being is in lower-risk use cases.

For instance?

And this is actually slightly premature, but our biggest product announcement later this fall will be how we're bringing Legora into Slack. And we've already started using Legora inside of Slack at Legora. And instead of our sales team going and pinging our legal team for input on sales contracts or NDAs, they just tag, talk to the Slackbot. And they say, hey Slackbot, or hey Legora, I just got a contract from this counterparty. Can you please review the suggestions that they've made? And instead of that agent then just escalating that task directly to a lawyer and saying, here you go, lawyer, you can now figure it out, it's going to use the company's predefined playbook and take a first stab at the problem and say, well, according to the Salesforce negotiating rules, This edit is fine, this edit is fine, and this falls outside of what we're willing to accept. I have made a counter markup. I have made some comments. Here you go. And if needed, it will escalate to a person. So I think the UX and where that handoff happens and the auditability of what the agents do is really important. And I think even if we let an agent make judgment, A person has to be accountable. I think that is really important. Like, a computer cannot be accountable, and I think that's still true for agents.

I know, very good. You know, we were having another discussion around accountability, but also responsibility, and you could still have the responsible systems.

Yeah.

Even though the ultimate accountability, you know, has to go to the people. So, You mentioned Slack.

Yes.

And it's so incredible how, look, how you run your business, right, on Slack. 9 out of 10 of the top AI-native companies all run their entire business, you know, on Slack. And at Salesforce, right, we're on a path to $61+ billion, you know, in the next couple of years. We run the whole business, right, on Slack and our incredible, you know, global teams They're building, right, skills.

That's right.

Right. They built a Contract 360 skill to enable the sellers and the AEs, the business folks, to just get that insight. What's the customer's whole footprint? Right. And then different areas where with Slackbot, with services and the like, it becomes a real force multiplier. Now, one thing you noted is that, you know, is there an infinite demand potentially, right, for legal, for compliance, for risk, this type of work? But there are supply constraints.

Yes.

Now, one way we tackle that, you know, even a large company, is through, you know, we call them self-serve motions. So you might have high repeatable, right, or very repeatable, high-volume, you know, work that still needs to get done. You know, it's important, but you can scale it. And then you can have high-stakes, high-trust work that perhaps is the best province, right, for legal professionals to really be applying the area. One of the areas, noted is that law doesn't work exactly like, you know, binary 0-1s, right? It's not— can't really rely on probabilistic systems. On the other hand, it isn't entirely deterministic. You know, our chief AI scientist Silvio and I had written a paper around the absence of an agentic trust architecture.

Okay.

And that we're going to enter a period where you know, you're going to have to move from rules, right? So you don't need rules and then standards, right? It's a very common element in the law. But how are you tackling that combination of, let's call them pure rules versus not entirely open textured?

Yeah.

But this concept of standards that either you're raising the bar on or standards that either the clients have to meet or the firms have to meet or your customers. After me.

So, I imagine that most of you know about vibe coding. I'm also sure most of you don't vibe draft. There is a reason you don't vibe draft a very important clause in an important share purchase agreement. There, you want to use precedent and deterministic clause libraries to extract the precise text that has been proven in precedent and in market that that type of language works, right? You don't want to generate a new type of contract every single time, but you want to allow an operating system and an agent to tap into those deterministic assets and then bring them back, right? So in a way, you know, we're building this sort of Jarvis for the lawyer. And they say, morning, Jarvis. Daddy's home, or mommy's here. I don't know. And I don't know. That was an Iron Man reference. And then they kick that off, and they start to tell Jarvis what to do. And what's important there is Jarvis should— it's a fine balance between going out Too far and like going into areas where it's unexplored and you're not basing things in your precedent, and other cases of just being like too unuseful, right? So what we've had to spend a lot of time with is like, what's the right level of guardrails on the agents and on the systems that we build? Because if it's totally open, you get something like OpenCLAW and it just goes completely crazy all the time. Or if you constrain it too much, it's not very useful. And what we've also spent a lot of time on at Legora is actually enabling and working with our customers on this change. Because I think it's a misconception that rolling out AI or these systems is kind of an IT rollout, like it's another IT tool. That's not the case. It's a workforce transformation where you really have to get an understanding of like, where does that frontier lie? Where can I use it today and where can I not use it?

I love your point that this is in some ways a— is it a workforce transformation, right? Is it a process transformation? You know, at Salesforce, we often say that the AI, you know, transformation, the journey of being an agentic enterprise for us, for our customers, is a people transformation, right?

Very much.

One of the key, you know, skills, capability, and mindsets that we think is so important and particularly going forward in the years, decades to come is curiosity.

Yes.

Right. And you're having that beginner's mind, you know, the element, you know, as I have said, you know, the greatest risk, sometimes the greatest risk is the unexamined status quo.

Sure.

So from that lens of curiosity and, you know, Max has grown a lot since he was 23. What have you learned? Like, what have you learned about workflows? What have you learned about, yes, lawyers? What have you just learned about, I mean, just any of these areas, right, during your time? And what do you expect to learn more of going forward?

So when we started, we were 3 engineers. And in our first Y Combinator interview, this is a bit embarrassing to admit, but in our first Y Combinator interview where We did not get accepted. Um, we got asked the questions, what type of lawyers are you building for? And my co-founder replied, are there different types of lawyers? And that was not the correct response. Um, and we later reapplied to Y Combinator and we got in. But what that showed me was, wow, we really don't know shit. So what I did was first I went on LinkedIn I found all these partners and I started writing them and I said, hey, can I buy you lunch? And I'll pay for your like hourly fee to have lunch with me because I want to learn about what you do. And I'd ask these patent attorneys and the litigators and the, you know, transactional teams. First off, all of them were way too nice to charge me for lunch. Most of them actually ended up paying, which was lucky for me because we didn't have a lot of money at the time. And I would just ask them so many things about how they were working, where were their pain points. And what you realized very quickly was like, wow, tech savviness does not exist here. Like, that is not a skill that most law firm partners have built up over the years. They know a lot of things, but they're not very tech savvy. And you realize quickly that the opportunity to build software in this space is enormous. There's so much greenfield. Every way you try to run, you fumble over a rock and you're like, wow, this piece of software makes $50 million a year and looks like it was built in Windows 95. Let's rebuild that. And you run in enough directions and you hit enough rocks and you're like, wow, if we just build all of this, we're going to be a $100 billion company one day. And what I also learned was nobody was going to take 3 software engineers serious. So I went to the biggest law firm in the Nordics. They're called Mannheimers Wackling. And I tried to get a meeting with their managing partner. And it was impossible. Just impossible. I mean, one of our early angel investors said it's harder to get a meeting with that managing partner than to break into Fort Knox. And at the time, it felt like it. But I thought of something. If I send him an email and say, I've organized a lunch with your second and your third biggest competitor, and unless you have time to meet, I'm going to go see them instead. And suddenly a slot opened up. So at 11:00 AM, I go to Mannheim Assoziation and I sit down with the managing partner. This is like pre-vibe coding. So we don't even have a vibe-coded solution. We have like a non-working prototype. But in these situations, it's important to have a lot of confidence. And so I walked in and he comes in and I commend him on his office. And I said, wow, Jan, you have such nice offices. They remind me a lot of McKinsey. And he says, oh, did you used to work there? I said, yeah, for a while. And I had done an 8-week internship 2 summers ago. But he didn't have to know that. And so we sat down, we had a conversation for 20 minutes, and by the end of it, I basically said, Jan, I don't know that much about law, but you sure as hell don't know much about AI. So we're going to need each other. And then he said, Max, I'm a guy who likes to take decisions on the spot. You should move into our firm. And then we shook hands, and sure as hell, we moved into the firm. We got a conference room at the top floor. No windows, but that didn't matter because we were looking at the screens anyways. And we spent the first 9 months of the company in that law firm. And I would just harass every lawyer there by bringing my computer to the lunch canteen and be like, hey, can you try our software? And they'll be like, no, I'm trying to have lunch. I was like, please. You know, you put the computer in front of them, you watch them try it, and then you just learn. You realize where is it working, where is it not working, what type of prompts are they trying. When they open the system for the first time, what do they think that they can do? Because remember, this was like pre-ChatGPT for a lot of people. This was before Microsoft Copilot. Like GPT was so early when we started. And the things that I learned from then, which we still have with us, is one, you have to build with your partners. You have to have this, super quick iteration cycle even as a bigger company where, you know, you observe the problems, you go back, you build it, you solve it, you repeat, right? And so having super tight connection is important. Secondly, there's so much green space. And the things that we started building was like fundamentally different from anything that they had ever seen before. And I think that really paved the way for let's not just rebuild the old artifacts, let's like fundamentally reimagine what the future of legal work can look like. And that has really been sort of north stars for us, um, so far. And when I look ahead, one of the most interesting, um, changes that we're observing is the demand and the pull from very large legal departments. So initially in the first 2 years of the business, we did a lot of, uh, law firm contracts. And what we're seeing now is actually chief legal officers and GCs going, okay, the tech is tested, it's battle-tested, and we see an opportunity to both bring in more interesting work in-house and we see an opportunity to amplify our teams to do more interesting work and to accelerate things like contract review, legal research, employee claims and things like this. So that we can do more stuff. And legal, even within a company, is often supply constrained, right? Like, you never feel like you have enough lawyers to handle everything that's going on, which is also why—

Is this true? All the teams in the audience?

Yeah, yeah, it's true.

They all need to be augmented.

I'm not saying cut Sebastian's budget because now he has AI, right? And so what I'm most excited about is you know, what are the ideas that are going to pop up as we enter this next era of the company? And also as we get better and better models and better and better systems on top of those models. So coding really had its breakthrough moment in Christmas with Opus 4.5 and GPT-5.2. Like, that was the eureka, like, cross the chasm, like, now it is more pragmatic and more productive to work with AI than to not work with AI. Like before, you had to correct the system more than you actually got value out of it. And working with Legora back in 2023, it was like working with a really shitty summer intern who didn't really understand— understood what you wanted. But now, honestly, it is so Impressive. And the first draft of every task that we go out and do, I think, is actually better than what most people would, would end up doing themselves.

I hear, you know, on that theme of curiosity, how much curiosity you bring to the customers, right? Whether it's when you were at the firm, you know, hounding all these folks, when you were, you know, sending notes out, right, to the law firm partners and the like.

And I think it's important to not just say the customer knows best. Like, the best relationships are not where you try to be like mushy cushy, you know, each other's best halves. Like, you want to be each other's jetpacks, and you want to be— you want to be slightly uncomfortable, but you're growing together and you're accelerating.

No, I think that's very inspiring, right? How do we build together, right? You know, grow together. Everyone's challenging, you know, each other. By the way, just for the record, I never received one of these, you know, notes from Max Where he's offering to take me to lunch or you know get a coffee. I couldn't. And yes, I would have you know.

You would have made me pay for it.

No comment. So one other somewhat lighthearted question. I don't know if you were just making Doctor Who references with the TARDIS and the Daleks, but when you're back in England, maybe. So who in the audience thinks that? either of us still look like the photos up there. Anyone? It's unclear.

So I was thinking about that. You have a nice beard, you should get it on the photo. I couldn't do that, you know.

Can Lagora do that for me? So, okay, let's kind of continue to build on this though. And you just touched on it. You know, when I spend time with customers of all sizes, Right, including all the folks we serve in the professional services advisory law firm space. But, you know, regulated industries, you know, broadly. And when I spend time with investors and analysts, you know, there often comes up the question of what's the chasm that we're trying to cross, right? You have companies, is it, you know, folks like you, investors, a lot of them are living like over here. They want to become agentic enterprises. They want to transform. But there's a big gap, right, between where they are and where they want to go.

Yeah.

And so in that context, is there a gap that you see between the companies, right, the in-house with the companies and these global, right, advisors, law firms who seek to serve, right, these, these companies?

I mean, you have a fundamental mismatch of incentives in a way where, you know, The law firms are thinking about how existential is this technology to my current business model, which is, you know, hourly based, and that is more and more moving to outcome-based outcomes, outcome-based pricing. And, you know, enterprises are going, how much of this can I take in-house? But I think the reality is like, again, by building together, you kind of figure out like which parts should live on each side. Side here because you're not going to take in bet-the-company litigation. Like, you're still going to work with professionals on those types of things, and it's too expensive to have those professionals in-house. However, you can do a much bigger part of the initial dispute analysis before you even engage external counsel, right? And so I think what's like fundamentally happening with the technology is like it's accelerating capability everywhere. And I'm sure legal teams internally have experienced this as well, where an engineer or an investment banker or somebody has like done some vibe drafting, sends it to legal, and it's like, hey, does this look good? Or like, here's my own legal research memo Claude gave me. What do you think? And it's just like, oof, it takes me more time to review this than to just do it myself from the beginning. So I think as we see that gap crossing the chasm, we realize that it's not only about technology, it's about people and change and management. And I think we're actually the only startup in the world who has more lawyers than software engineers and are this successful. And the reason is we have these legal engineers, tech-savvy lawyers who partner with our customers to both give them use cases, but maybe more importantly, hold them in the hand as we sort of transition through this chasm and actually build out all the use cases. And some firms, some law firms are going to make that transition very well, but some of them are likely not. And, you know, it's the sort of innovator's dilemma. Like, you have to be very ambitious right now if you want to keep at it.

No, it really resonates for me. Um, you know, we have a saying that this is a moment where It's the end of software where humans do all of the work.

Yeah.

And I think that in the area of law, you know, is this the end of like lawyers doing all of the, you know, of the legal work? And what prompted me to kind of pose that question is when you were telling us your, your very sad, unfortunate Y Combinator story where the investors said to you, Well, you know, who's the lawyer that you're pitching to, these different set of lawyers? What I thought your answer was going to be, you said to them, well, we're not really trying to sell to lawyers, we're trying to solve problems. And so when you come at it that way, and is this the end of lawyers doing all the work? Do they get capacity? How— I mean, what is your view on that? And what is the view that you think the actual customers that you have, whether it's lawyers or companies? have on that question?

So, I mean, there's sort of a question of like, do you sell the shovels or do you like go dig for the gold, right? And we're in the shovel business. We want to make every single lawyer in the world more productive and happier and, you know, more productive in the things that they want to go and achieve. And, you know, at the end of the day, you're right, like you're just getting to an outcome. But I think that designing the systems that will drive those outcomes is best done by the people who have done that work. And I think it's an opportunity for legal teams to instead sort of live in the place of let's review all the work by AI. you move upstream and you design the systems and the rule set and the guardrails under which AI will operate. Like one example from our own engineering team is when a bug report comes in, we have an agent who picks it up and starts to fix the bug. When we have a contract that comes in, we have a Legora agent who takes the contract, starts to review it, and escalates if needed. And, you know, the complexity of the tasks that that system will be able to manage continually going to increase, right? You know, a year ago we were at the level of an NDA. Now we can do a really good pass on a data room and the first due diligence. Um, you know, with one of our latest acquisitions, we're able to look at an entire sort of discovery database and run an initial chronology and fact extraction, and that becomes the initial, you know, data that we then make a determination about a case on. And so The complexity will increase, and I think professionals will have to continuously sort of move up with that. The change or the difference between what's happening now to before is that it's just moving so much faster, right? Like, this has been happening all the time. Now it's just happening at an explosive rate, and it's very tiring and exciting to continuously keep up with it.

As you know, I share you know, the excitement around these areas. You know, you and I were talking about the open letter, you know, I'd issued to really push, because I think it's an incredible opportunity for the partners, right? The associates of these law firms, especially though for the companies, right? All the in-house legal, you know, privacy compliance risk teams. Last question. Right now, what is giving you the most energy? Like on a personal level, what's giving you energy? You might say, well, here are the things that are draining my energy, but what's giving you energy? that you want everyone to be thinking about?

I had a recent experience of going back to my old college and doing a presentation for like the full class of computer science students and all of the business majors and like the applied physics people had like snuck their way in. It was supposed to seat like 300 people. There were 1,000 people in the room. Like there was no oxygen. Nobody could like breathe almost because it was so packed. And I was, you know, talking about Legora and what we've done so far. And then the line to like say hi was like 400 people. And people came up to me and was like, Max, Legora is my dream job when I graduate. What do I have to do to like get an application in? And I was like, wow. Like, I remember being there and my dream job was Spotify. I made 7 applications to go work at Spotify and I never got an interview ever. And I told that to Daniel Ek when I met him. I was like, you missed out on me, man. Like, I could have done some good work for you.

But if you had done that, maybe you wouldn't be doing this.

That's true. That's true. So I'm happy he, like, you know, rejected me on my resume instantly.

We're trying to spoil it for you guys.

He took like half the resumes and was like, you know, oh, unlucky. We don't hire unlucky people at Spotify. But no, that was like such a boost. That was incredible. And I was like, wow, if I'm going to work for something, I am excited to work for, you know, building category-winning and generational tech companies outside of the Valley. Like people in Stockholm need a And people in Milan and Munich and Paris and Madrid and London and New York, like all our big hubs, they want to work at generational software companies that like define what AI will mean for humanity. And every single of those companies today are in the West Coast, Seattle or here. And there's so much talent in other pieces of that world with slightly different values, slightly different morals. And I think it's important that the power of this technology gets more equally distributed. And I really hope that we can play a big part in that.

Love it. Well, this is an area that's high trust, high stakes. Really appreciate and respect that you are seeking to bring, right, that lens of high trust, high stakes, and also velocity to all of this. Well, let's give Max a round of applause.

Thank you for Thank you so much.

Hey, this is great.

Thanks for having us.

I think, you know, trust is a growth lever, and I wish you all the best both in your transformation and maybe some other folks that you're meeting with in the months to come.

Thank you, Sebastian.

Very good.

See you soon.

Thank you.

See you at the afterwork.

Yes.