Wordsmith Founder Interview | Ross McNairn

Most legal AI conversations with founders start with what they are building. This one starts on the other end — how a legal tech founder thinks about buying legal AI if he were a GC.

Anna sits down with Ross McNairn, founder of Wordsmith, a legal AI enablement platform built for in-house legal teams. Ross is a qualified lawyer turned software engineer who led product teams at Skyscanner and TravelPerk. After working closely with in-house legal teams from the business side, he saw a consistent pattern: legal is business-critical, but legal knowledge doesn't scale.

In this conversation, Ross gets unusually candid about the stuff most founder interviews avoid: how demos and pilots can mislead procurement, what signals are harder to fake, and what 'the last legal tool you'll buy' actually requires.

This video was made in partnership with the featured vendor. The vendor provided product access and reviewed the video for factual accuracy. Editorial opinions and conclusions remain those of Legal Benchmarks.

00:00You don't necessarily want wordsmith to be the first product people buy, but you want to be the last. What has to be true about the product itself for Wordsmith to be the last.
Speaker change: We were very deliberate about saying, right, where do the requirements look that are quite similar? Who's got conflicts of interests? Somebody is going to be losing. If external council's making more money, you have to be paying more as an internal council. Our view is let's just pick a group where we're not going to be conflicted. Let's invest very heavily. Why do you think Harvey and Lora takes the opposite approach where they're serving both in-house and also

00:31law firms?
Speaker change: They both started with law firms because they pay a lot of money. It's difficult for people to understand the quality of legal service before they purchase it.
Speaker change: Lawyers procure software the way that they procure law firms. They rely very heavily on the perception of brand.
Speaker change: If I were to put you in the shoes of a legal leader that has to procure different tools would be a question that you would ask in the demo that's not asked enough. dig under the surface and if you can't get a clear answer on that with 1 2 3 4 then I would discontinue conversations.
Speaker change: I don't fully buy that we'll need more

01:01in-house lawyers.
Speaker change: Each step as we evolve as a society we become more and more cognitively demanding and people are going to have to reckon with the fact that they are going to have to become better lawyers.
Speaker change: One of my favorite books is AI 2041. How would you write a chapter for what a legal engineer dayto-day life would look like in 2041?
Speaker change: So you won. Hi, I'm Anna from Legal Benchmarks. Today I am sitting down with Ross Mcnan, the founder of Wordsmith. Ross brings a very rare perspective to the table. He's a qualified lawyer turned software

01:31engineer who previously led product teams at Skyscanner and Travel Perk. Having seen high growth companies from the inside, Ross realized that while legal is business critical, it's often perceived as a bottleneck. And that's what motivated him to build Wordsmith. Today I'm putting Ross in the hot seat where I'll ask him if he were a GC right now, how would he actually go about choosing the right legal AI tool and why is he bullish that Wordsmith will be the last tool a legal team will choose even if it's not the first. Let's

02:02dive in. Hey Ross, it's so good to have you. I know when we had the first conversation back in the beginning of 2025, you were eight person team back then and then you grew to 40 and now you're 80. So what has that process been like?
Speaker change: Yeah, I mean it's been nuts. like it's it's thankfully not the first sort of hyper scale journey that I've done.
Speaker change: I know you have Robbie, you have Veladimir as the trios that built this empire together. Usually we hear more about Ross than we do about the other two and I know that you had a very selective process for choosing your

02:32co-founders. So super curious to know how you found them. What is the working dynamic like between the three of you and why should legal teams trust the three of you with their legal department's work?
Speaker change: I think it's a great question. So, we've got quite a lot of experience between the three of us in working at various different stages of scale. So, Robbie and I have built travel perk or perk now together. We did several years there. I watched him as an operator. We actually went to law school together a long time ago and then he went off and qualified in New York and did all sorts of things

03:03and I went into tech. So we split and then we we came back together apart and man was nice working with him but Gibbs was I was very keen even though I'm an engineer or retrained as an engineer I was really keen to get somebody who was spectacular as an engineering lead. So when I was at Leto, one of the previous unicorns that we built, um I was running product and engineering and I moved them all of my team over to to meta um when we were going through a merger at one point and then it was just a few years later when I was starting Wordsmith that I pinged those guys who had moved into

03:34BAI stuff and doing a lot of ML work who is the who's the best who's the best person at Meta Meta Europe that you know of and the two of them came back and said hey you have to meet this gigs guy he's he's crazy you know he's brilliant. I started coding with him in the evenings for a while just to see what it was like and and then the three of us went and did a kind of a pressure test on the community of Santiago in the the west coast of Spain where my objective was to try and put us in a very uncomfortable position to see what it

04:05was like enough of these teams and actually I really didn't want a team that was just me or that that I became a sole founder of after two years because I had to fire all the co-founders or the dynamics to work So, I wanted to pre-qualify that and frontload that fast. But we did. I took them for a week to the commamino. You didn't have to plan anything. It's like sitting at 11:00 in the middle of the night with no accommodation trying to work your way out of a massive raining storm and kind of get to see pretty quickly what people's character is like and how how we worked as a group. Um, so yeah, I

04:36think the reasons that I'm bullish on the team is that we've done this again and again, right? Like between us, we've scaled. I've done three unicorns before this. Um, when you wrap that together, it's a good team and and more importantly, just a great culture. Teams only really work that out, but you know, two, three years in, I think it's actually most common case for post series A companies failing is that the founding teams just don't jam cuz you may get quick wins or short things over, but any type of startup is a decade of your life really. To spend a decade with anybody, it's like a form of corporate

05:06marriage. You just do a little dating. There's so many points there I want to double click on but I want to first learn a little bit more about your background because you mentioned that you have exited multiple unicorns like you started out as a lawyer. So I wonder what has been your experience coming back to the legal industry like how do you compare the legal industry then to what it is like now?
Speaker change: You see all these these people on LinkedIn side coding different AI products, different legal tech products. That was kind of me before AI about 10

05:37years ago. I just I can resonate with that desire just to get your hands dirty and start constructing and building things and applying it to the domain that you're working in. But that was kind of how it began is I started off as a lawyer and I was I was sent a chunk of work up for us to do. It's like a bunch of DD work and the idea of printing out a tree of paper, literally four sheets of it. They sent DVDs up and they wanted us to print them out and then they wanted me to go through with a sticky thing and put little stickers on the relevant pages. I was like, what? Why did I go and do seven years of training

06:08and qualifying and learning about juristprudence and hum and all this stuff to sit here and do this work? Didn't make any sense to me. But there was a lot of pressure.
Speaker change: How did you end up being in legal in the first place?
Speaker change: Well, I was attracted to the bits of the profession that I think most of us idealized a bit. Robbie and I were in the debating team together, the muting society. So, we did a lot of legal debate. I love that. I like Jewish prudence. I like I like reading. That was a book nerd. Just enjoyed sitting there. I took all sorts of weird modules at at law schools. space law and international conflict law and medieval

06:38law. Just one of these people who was curious about the academics of it, I really loved that. And I think when I arrived in private practice, you know, that's not really what you do. You print a tree and you start doing DD, at least I did back then. And that drove me quickly to want to find something that was much more scalable as a profession. And so what has your observations been after you've left the legal field, did a bunch of cool things, helped scale a bunch of startups from 0 to one and one to 100, and now you're back here again. What are the changes that you've seen

07:08that surprised you?
Speaker change: I still think lawyers now feel a pressure to for the first time ever use technology. I think that it's being driven by for many lawyers by the wrong incentives. I think a lot of them are being dragged into this out of fear of missing out or out of a structural nervousness about what's going to happen to their jobs. sort of existential stress and I think there's a much smaller cohort of lawyers like you know yourself and many others that are beginning to emerge now who I think are a little bit closer to the the what what draws me at the beginning which is there is a group of people who just want to

07:38build with law and can see the potential that sits within here and I think that that that bit still I kind of forgot that that cohort of people existed with 10 years out of the profession that there would still be practical impractical resistance to it. you move into the tech world and everybody there is such an optimist about how you can use and leverage technology and so moving back into law it's been a bit of a shift to then remember oh wait actually you've got two groups of people that you want to manage here and helping this group through that transition is a real thing so that's a big one the other thing I found kind of interesting is

08:08that lawyers almost at times procure software the way that they procure law firms in that it's diff it's difficult for people to often understand the quality of legal service before they purchase it so they rely very heavily on the perception of brand or is it a magic circle? Is it this? And when you watch them try and apply the same purchasing methodology to software doesn't work as well. So I think it's interesting watching professionals not used to this go through it.
Speaker change: You said that when you were choosing your founding team, you had to pressure test your co-founders in many different

08:40ways. And I'm sure for legal teams as well now inundated with so many different tools in the market, all of them branding themselves as being capable of solving all of the problems out there in the world that they also need to pressure test the vendors as well. Do you think chat is wordsmith's biggest competitor?
Speaker change: I don't think that that it is. I think it can look like that early on when people are just building the initial bits of their product. But it's almost impossible I I think to have something that will be superb for a dentist and superb for an in-house lawyer and superb

09:10for you know a high-end litigation lawyer in in links. I think ultimately over time you'll seize just naturally those people do have very different jobs and as the technology matures and those specialisms will become more and more apparent. And what do you think are the differentiating factors between the legal text and the general purpose AI assistance? Because what a few studies have been published last year or the reports that when legal AI is compared to general purpose AI including our own research that the legal AI platforms are not necessarily always more accurate

09:41than let's say chubby for example. In that case, a lot of legal teams will think, well, why do I just spend way less and get a general purpose assistant rather than a legal AI assistant? What would you say to them?
Speaker change: Oh, I would say nearly every single client that we sell into has a horizontal product as well. They have something that the the software engineering team will use, etc. I think accuracy is one important component of these systems, but it is one of 50 different factors that we need to invest in and build out. And just to be concrete and specific, you use chatd.

10:13You want it to complete a document to help you to fill out the placeholders within it. It's going to it's going to destroy the formatting and it's going to rewrite the whole document and it's going to give you back just something that's a plain text bit of text you need to copy and then go and redo the whole thing. That's not per se an accuracy problem. But if you try and take practical use case where a lawyer wants to go and do some basic templating work, it's completely useless for most people if it doesn't have that that that instance of there. Or if you want to go and make, you know, redlinining and you maybe find something that could vibe

10:43code or insert a few changes here or there that you may be able to make it very accurate, but there's a huge amount of work that needs to go around understanding the context of who the agreement sits with. What happens if you've got two agreements? It needs to merge them. What happens if one's a word document and one's a PDF? What happens if you know they've got comments that are relevant to that that don't? So I think the devil really comes into the detail with these things and you know chatbt is is great for a little bit of everybody but rapidly if anybody wants to do anything meaningful you'll hit the limits with it very quickly.
Speaker change: So you mentioned before something quite

11:13interesting which is you don't necessarily want wordsmith to be the first product people buy but you want to be the last. What do you think has to be true about the product itself and also about the market in order for wordsmith to be the last? So when you when you build products with these you know most of the companies that I've come into as a product exec have a lot of product debt which means that there are a bunch of decisions that have been made that you then spend a very long time unpicking and usually the people that then become problematic to you as you start to grow people that then pick out a subset of your user base and they

11:44solve a problem for them much more comprehensively than you can. So I think one of the risks early on in this cycle is trying to be too applicable to everybody because early on that works well when the tech is immature but over time you're going to find yourself super vulnerable if you're trying to sight battle on the high end of litigation at regulators and then you're trying to also cater for a sole GC who's basically you know doing labor law or those are very different use cases. So I think for me what needs to be true is you need to have you need to it's a balancing active you need to now guess what is the right

12:14cohort of the market that you should be doing after do they have a similar grouping of requirements when the technology gets more mature and how do you set yourself up so that in 2 3 years time you are by far the best for that group of people versus trying to cater for everything under the sun because as much as anything the advantage of being tight on who you're catering for is all the stuff you can say no to that you don't need to build that you don't need to bundle in to make your product more complicated that the the means that you can keep it simple and precise and excellent for that group. And so for me,

12:44we were very deliberate about because I've seen this journey a few times now when you go to sprawling, we were very deliberate about saying, right, where do the requirements look that are quite similar? Who's got conflicts of interest? How can you both cut internal spend on external lawyers and make external council more money? This just struck and also make money as a legal tech provider. Like the calculus of that just long-term just doesn't work. Somebody is going to be losing. If external council is making more money, you have to be paying more as an

13:14internal council. It doesn't work otherwise. So our view is let's just pick a group where we're not going to be conflicted. Let's invest very heavily and be very sharp and let's also be clear about who we're going to repeatedly say no to and who we're setting up to conflict with in the medium term.
Speaker change: What do you think is the target segment that Wordsmith is actually serving? I know that there is already a very clear messaging that Wordsmith is serving in-house legal teams, but even within the in-house legal team umbrella, there's quite a lot of different in-house council teams. There are the solo councils on one end and then you

13:45have the mid-market and then you have the huge enterprise clients that could be multinational and they all may have different needs. So, where does Wordsmith fall in this spectrum?
Speaker change: We started in midmarket and now we're moving up into enterprise with the increased feature set that we've been building recently. That's where we're best. Now we do have people that sit as sold GCs and small teams that use us definitely because we can be as much as anything. It's like where are we focusing our road map on and that's that's the sweet spot for us and our objectives to help them with their operational efficiency. You know getting

14:15more for less to help reduce external spend. We're explicit. We have clients that have saved 200k in the first week of using us at external council. And three and most importantly is to improve how they interact with the rest of business. A lot of legal techs right now just focus on being a co-pilot. Now a big reason for our integrations and I'm not sure if he's had a chance to look at a lot of the Slack stuff and things like that that we've gone very deep on but that is all around creating a big surface area with the rest of the business because the biggest you know I think it was actually the receive a conversation he said hey the the

14:45biggest difference between private practice and being an in-house lawyer is in private practice you have loads of clients now you want to keep them all but you know if you lose one or two here that's not the end of the world when you're inhouse you get one boss it's the company and that boss boss has 500 or 5,000 stakeholders. So even the way that the products are built, you can at a glance when you look at these legal texts, you can see are they built for raw or are they built for in-house cuz the more fun ones will have projects where they're bucketed per client so that you can have masses that are siloed that you can start to say have one for

15:16Acme, one for Disney, one for this and you sort of have these different whereas the in-house ones tend to have one corporate context and then it's designed more for lots of individual stakeholders across the business and so that today that small rift is getting more and more pronounced over time. Why do you think Ben then that Harvey and Lora takes the opposite approach where they're serving both in-house and also law firms? Do you think that's a smart approach?
Speaker change: I think they both started with law firms because they pay a lot money. I think in house is a opportunistic thing because

15:46nobody's claimed spaceship and I think that will be much harder as people become aware that there are structural differences between the two. And also as there starts to become a clear conflict that's apparent if your law firm is pushing you technology to lock you into that law firm, you're going to increase your dependency with that law firm. And how why would anybody do something that is going to be economically damaging to them? So one of the major advantages that I see us having over time, we don't have a conflict like you can work with any law firm, you can collaborate with them, you can be agnostic, not locked to whoever else is using the platform and

16:17we are explicitly going to help you drive that spend into the ground. Whereas if you think that the conversations that they would be having with with the law firms is, hey guys, we're going to take your revenue and get rid of 80% of it. It's just the early stages of this market. There are, you know, tens of millions of corporates around the world. It's a massive market. And I think that they there is going to be a huge shift now to insourcing a lot of work into them. In-house teams are going to start to set up very differently. They're going to get a lot more demand from across the distance as AI starts to become more prevalent. they're going to get much more pressure

16:47to get rid of the $3,000 an hour fees that they're having to pay externally and they're going to have to become a lot more technical as they become capable of leveraging and deploying this stuff internally. As for the future, it's the only thing that we're focused on and I think that that long term will work out very well for us. So for a very busy legal team leader that probably isn't that technical or doesn't have a lot of time to explore all of the different options out there, what do you think would be your advice to them on what kind of proxy metrics they can pay attention when they're looking at, let's say, a vendor's website to see whether

17:18it's a good fit for their team? One of the things that you mentioned before that I think is super helpful is looking at the logos. What logos is that vendor already serving? Do you have any other good proxy metrics that can help these teams decide? And what do you think is a really bad proxy metric?
Speaker change: Well, I think I think that's a really good question. I think when I start with the bad, I think a bad proxy metric is just money. Like how much money have you or haven't you raised? If that was a metric with that that we were optimizing for, you would play the game very very differently. I think a good metric to look at is definitely logos, but also

17:49almost informally, you want to find people that have chat. You want to find in these things like how they've procured the software because ultimately the thing that really matters with this is what's your usage going to be like over time. When you buy this stuff and you commit to spending any amount of money to the CFO, you're really going to have to make sure that is a product that gets high usage for you. Otherwise, you're just wasting cash. So, I would strongly recommend anybody should y start with the logos, but then actually just go and try and find a way of engaging with the people that are using those products for 5 minutes that you're not talking to. you're cutting through

18:19all the noise and you're just doing immediate references cuz speaking to two or three real customers will give you an incredibly clear, you know, it's much much more effective than I would say a twoe trial where you get to look at 5% of the functionality where everybody's just benchmarking the systems against each other. It's very difficult to do that. So that would be if I was shopping right now, that's what I would do is I would take two or three vendors, make sure that I'm like looking at ones that are specialist in my segment, more generic, and I would go speak to one or two of the customers and try and get an idea of where things. And now if I were to put you also in the shoes of a legal

18:50leader that has to procure and compare different tools for their team, what would be a question that you would ask in the demo that you think would be really good that's not asked enough?
Speaker change: So I think one of the ways that you can tease out is the product built for your segment or not is really how well the team so I'll give you a specific question but I think theory first. Is this a group of people that understand my pain points deliberately? Are they talking about random stuff like that other lawyers do and trying to hawk um

19:21solutions that aren't that room or are they coming in and attempting to do they understand the organizational structure of an in-house team? Do they get you know what a privacy council does, what a commercial council does? Do they understand the dynamics of my or within that can they show me specific targeted use cases in the demo that actually really apply to what we're doing? And then I would dig in a tiny bit more and say, "Well, that's interesting." If you're doing that on an NDA review, what happens if comes in like practical stuff? What happens if it comes in PDF? What happens if it comes in and you know

19:51the the web pages have half the agreement in them and I don't have I don't have access to that. You know, they're sitting on the page and I it's only half of the contract that I have because I just have the order for and the terms and conditions are sitting on the website. You know, these are very specific in-house problems and so I'd be quite targeted about just asking that kind of stuff. dig under the surface and see is this just a surface tool that is being used by everyone or is it actually got people battle testing it in my segment. What the reality of what's happened for the past couple of years is there was a big incentive early on here to build stuff that demoed well and looked cool that you could stick you

20:23know I think if you look at the first versions of these products they were useless like two years ago. So one shot click it get an output and it was just about trying to get vibes and hype early on and then people are trying to backfill that stuff. So you really want to work out are there actual structural hardcore applied use cases that I'm going to be able to put into production within a week. And if you can't get a clear answer on that with 1 2 3 4 then I would discontinue conversations.
Speaker change: Usually what happens is a sales team would say I'll come back to you after I check with my product team. They

20:53wouldn't answer it in the demo.
Speaker change: But that just tells you what you need to know, right? Whereas if they're selling to your segment all day, every day. So for example, the way that we set up is we have we only do in-house. We set them up by geography and by segment. So we have specialists in SMB, mid-market, enterprise. Each one of them has a different set of use cases and they have an organizational structure of a typical in-house team for their segment. When they arrive on the call, they'll sit down, map out the people on the say, "Right, I'm dealing with a multinational that has four, you know, 40 lawyers. They've got three different or four

21:23different sub teams within that and I've got two representatives, one for commercial, one for privacy, which means that they've sort of pre-qualified it down to these are the three or four use cases that are most frequently used across the words user base that are probably going to be relevant to that. If you've got somebody that's just sending everywhere, you're not going to get that kind of time to go to market sales motion. They're going to have to go back and check with the lawyer on their team. They'll have to come up with something and then send you an email.
Speaker change: So that's another proxy metric. How well does the sales team answer questions? How well does
Speaker change: Sure. I mean,
Speaker change: yeah. Well, we have a so for example again like we we have a driving test. They're not allowed to sell until they

21:53until they can actually. So we make them go through and learn you know they have to become familiar with a lot of legal terminology. They can't actually explain fully what you know super cap or liability cap is they actually understand a degree of the the domain before you can even engage with a lawyer. So otherwise you just are talking. So, I'm wondering in the pilots that you have won when you were compared head-to-head with other legal texts or have lost, what were the reasons that you've gotten for why Wordsmith won or lost that deal?
Speaker change: It changes month for month, right? Because the space is moving very

22:24quickly, I would say. So, the interesting thing with us is a year ago, we actually didn't we were kind of building a different business. you know, we were building something that was sort of an autonomous agent that would sit in Slack and respond to queries and didn't really have a lot of the co-pilot infrastructure of an assistant and all that stuff that sat there. So obviously back then we lost on that background but anybody that had our technology stack which is a typical like SharePoint, Google Drive, notion that kind of stuff, we would immediately win based on those integrations early on. Then we started

22:55to become much more competitive in the assistant space and we started winning a lot of deals that people were looking for something that could sit there. We were losing a lot on contracting. So most of last year we invested very heavily in the contracting tech so that we could do headless contracting which means you don't actually need to be in word. You can review multiple agreements, single agreements, any type of agreement. We spent a lot of time on playbooks. We were losing for that for a while. We lost a lot to some of the point solutions. If you got somebody that really wanted a commercial contracting tool about seven or eight months ago, we weren't very competitive

23:25in that space. That's changed significantly now. So, we don't have an issue there at all. We win a lot of deals on that now. Then we started to lose deals as we moved up market in people that wanted more due diligence stuff. They wanted to do these big tabular reviews and stuff like that. We built a product there which is now extremely competitive. We're winning a lot of deals as it sits there. So, it it it depends and it depends a lot on what the client's looking for. Now, sometimes where we lose is when people want a really specific CLM solution and then also long-term obligation tracking and

23:55stuff like that. We'll see how much longer that's that's going to be an issue for us. So, it depends and it's sort of dissolved over time, but our attitude is that we have a phenomenal engineering team. Most of them, they come from, you know, they're ex Sky Scanner guys that have built two or three companies with me. They're from Meta with Gigs or they're from Fanduel or another Unicorn. Those are kind of the three lineages of our team. And when we set our eyes on a new feature or a new area that we need capabilities within sort of 6 to 12 weeks, we can become incredibly competitive very quickly, which is our unique strength, I would say.
Speaker change: So, what's on the road map in 2026 that

24:27lawyers or people that's on the lookout for tools should be excited about and the competitors should be scared about? Well, with without telling you something that would give our competitors a head start, I do think that in about eight weeks time, there are a few things we're going to release that are going to structurally change everything about how our product works. Like it's a very fundamental shift that we've been working on for a while now, and it's going to mean that all elements of it become significantly more compelling. But some of the specific things that I can tell you is that we have and will be

24:57very soon to be releasing a massive upgrade to contract management, life cycle management. So how you search, analyze, how you work with repositories, how you take masses of messy data that you've never had the time to organize. You know, everybody just has these big folders that they just stick everything in. And then they come around and it's like, I have no idea how to organize and group that and work out the clusterings and what's the related DPA to this MSA, where are the amendments, all that. And then from that, how do you get intelligence out of that? And how do you blend that seamlessly into a system so that you can use that and draft off it?

25:28And that's a huge area that we'll be releasing very soon. And doing that in an integrated way that you don't need to shift stuff over, you can do it with one click and just connect up to your existing systems and then get that intelligence appearing in words split is going to be the big thing that lands in the next couple of weeks.
Speaker change: Am I hearing three letters? MCP somewhere in there. [laughter]
Speaker change: So MCP is huge. We've already got some really great MCP integrations. The bottleneck with MCP is often the systems that people want to use haven't yet exposed great MCP endpoints. The minute that happens, it's going to be wild.

26:00Speaker change: Yeah. So we talked about how legal teams should choose, demo, test their tools a little bit after they get a tool in the door. Let's say Word Smith gets in the door. What have you seen that's very interesting about adoption and usage? Because that's one of the KPIs, right, being measured for the legal teams and also for the vendors themselves? What I've noticed speaking to some legal leaders is that there's desperate usage across different kinds of lawyers. So can you speak a little bit about that and maybe like what are the ways that can drive adoption that was surprising or would be helpful for legal teams to

26:31know?
Speaker change: That's such a good question. So actually our internal northstar metric beyond you know revenue and all these other when it comes to product perspective is the number of what we call legal interactions that they have within Wordsmith in in a in a week a month. The way that we define that is just a general Q&A isn't a legal interaction. leg fresh is when there's a specific or bespoke bit of tooling that we've built that doesn't exist on chat GPT that gets triggered and used. So for example, advanced contract review or drafting a a template or a blueprint or you know

27:02performing one of these things. So our metric for that we were at about 14 of those for an average WordSmith user at the beginning of last year in in the space of a week. Now I think we're sitting at about 75 something like that. So it's the number one thing that we focus on and there are there are several bits to it. So one is just making an incredibly simple to use product because most teams do not want to sit and learn how to configure the desktop before they can get any meaningful values. They don't want to have to build hundreds of workflows or bring in one of the the big four to get them up and running. So what

27:32we tend to work on with them is early on focusing on the one or two or three power use cases that's going to unlock that awareness across their whole team because you need to have a hero use case where the GC can sit and point to that person or that group of people and say we have shifted this number from here to here. They are the the hero they are the champion and then naturally people are competitive and it will get rid of a lot of the skepticism and the adoption will flow across the whole team. Where you tend to fail on these rollouts is where you're too broad. You don't have anything particularly sharp. You just

28:02roll the product out. You do a hackathon and everybody just has a few goes and you never get one or two use cases that are 10x more impactful. And so as a result, it's just a detected. So I'd say use case identification and you can start that even at the trial process. You should start that at the beginning of the sales call. So our philosophy is pick the two or three killer use cases and then just focus on getting those right but that alone will give the team ROI and then everything else is upside after that.
Speaker change: So for those lawyers that are very busy and want a benchmark of what some killer

28:33use cases are for mid-market and enterprise teams. What would you say they can reference as a starting point?
Speaker change: Well there there are three or four that I would look at very carefully depending on the team that you have the most pressure with. contracting simple very easy to go after got a lot that you can do with third party paper review is usually the big pain point not on your templates all the mess that comes in the basic commercial things like there are everybody's trying to do it in some design very few solutions will actually save you time on it so I'd say that's a hygiene factor you should probably check that out do you have a playbook can you

29:03put it in place can you send in any type of documents and can it mark it up and save you 70% plus time on the tail end true I would say about a quarter of what we find to the the sand in the system for in-house legal teams is actually just responding to the rest of the company. So, it's low hanging fruit of just getting repetitive Q&A off your table and doing that in a way that allows the legal team to approve and audit what is going in and out is can most of these sort of autoresponders are useless for legal because if it sends

29:33out any bad advice, it doesn't work. So, I would go after something like that. It's easy to measure. you know that each one of them is 10 minutes of time saved and you can just look at a number of how many responses you cleared in a month or a week and you can add it up. The third one that we found to be big is repetitive form completion. So COC work taking PDFs or employment or labor law where you're just repeatedly completing these things. Nearly every subset of a legal function inhouse does this all day long in an invisible painful way. They have a Google folder with a few templates somewhere and people are digging them out manually going through

30:04them. And so a big use case we found is standardizing that and then exposing that to other teams with very strict guardrails so that if you want to get it off your table, you can give it to teams to complete within very strict parameters under your approval. And so this is where I think the use case is very different. If you're in a big wall where you're trying to give creative or nuanced or value ad advice, this stuff is really operational. This is the stuff that you just want off your desk. It's on the opposite end of the legal function that you should be focusing on. And that's where I would be spending most of my time looking for for the value is the boring stuff that you can

30:35delete. I think the end state is as legal becomes faster and quicker as a service in the company. I think people will want more of it. I think it's like mobile internet. The cheaper and more available you make it, the more demand goes up and I think that there is a huge amount of latent demand inside companies for legal input that people don't go to legal for because you take too long or it's too high friction or it's just annoying. So, I think that the job of in-house lawyers is going to go from just get this the universe of what they do right now, paint it up a bit and then

31:05you'll find that there's about 300 claims more stuff that people need and want everywhere. So, I think the vision for us is to create the tendrils that an in-house lawyer can deploy this across the whole business and then have the little air traffic control center that they sit in where they can have transparency and control of all of the legal advice that's going on. they can see 10 times more. They have the levers to properly change risk in places that right now they can't get anywhere near. You know, there's so much untouched and unmanaged risk in in a business because

31:36it doesn't come to you because you get perceived as slowing things down. Whereas, if you were actually accelerating those bits of the business, it would all flow through you and you'd have a lot more control and you'd be much more critical. Our view is that if we can become that invisible spider's web for an in-house lawyer and we could become their infrastructure within a business. You make them we want to be the first tool that a lawyer goes to procure when they arrive in a new job. They're like right this is infrastructure that we need to run a business like everybody needs a lawyer.
Speaker change: I don't fully buy that we'll need more in-house lawyers because everyone will

32:08get redeployed into more highle work and I just want to share with you a little bit about what I've seen in house. So when I was in house, I noticed and from speaking to other in-house lawyers that in-house roles have already been gradually reduced full-time in-house roles throughout the years, even before AI, I mean, it's been happening already with contracting and outsourcing. We all know that it first gets into like part-time work, then it gets kind of outsourced to cheaper jurisdictions. And so the same with AI, being able to do a lot of the work like playbook creation.

32:38I know that in one of my previous companies I worked at, it was someone's full-time role until all of the playbooks were created and then that person got let go. And so for sure in this automation process and all of this productivity efficiency driving age there will be roles that will be eliminated and there will be roles that remain important. So I wonder if what you're seeing is redeployment of you know certain roles that will become more important in the future but at the same time there is going to be an elimination

33:09of certain in-house roles that people do today.
Speaker change: I think your observations broadly I would agree with them I think a good analogy is software engineering and product. So how I've seen that change between and wordsmith. So at Perk we had squads that were six to eight engineers with a dedicated designer, a dedicated product manager and this like Spotify squad model and they would sit there and it was quite you know heavy-handed and it would run a twoe sprint and do a very classic software development cycle. At Wordsmith the perfect squad size is

33:39between one and three engineers. They have about half a designer and they have about a third of it as a product manager. So, it's much leaner, much much much more efficient, and the output of one of those teams now exceeds what I was getting from a whole squad before. Now, does that mean that I need fewer engineers in total? I don't think so. I'm taking on way more stuff. Does it also mean that there that that we're not going to have as many of these roles globally? I don't think so. You're not finding the fewer software engineers. I think what you find and the two things we need to break apart is two things can

34:09be true. companies can reconfigure how they work to be more efficient and to hit certain problems more aggressively and the aggregate number of people in a particular domain can continue to go up. So I think the idea that the the old model that we had 3 years ago is going to be exactly the same in 5 years time. Yeah, I would agree with you. There will be a huge reconfiguration of the whole legal services. That's separate to the idea that we're going to need fewer lawyers. If you look back at the evolution of legal services over the last you know 50 years when we bought in

34:40email when we bought in a word processor what happened to the per capita number of lawyers in both cases almost it almost doubles within 5 years of that so I do think legal services and specifically in the corporate context is something that we need way more of than we're currently getting from you know supplied to us and I also think that the teams are going to fundamentally change how they're organized and the types of work that people are going to do is going to be very different
Speaker change: one common push back I hear about that is a lot of the legal AI startups they promise that they will eliminate the

35:10grudge work so the lawyers can focus on more higher level judgment task but then there is also the saying that most people are just ordinary people not everyone is a superstar and you can only make so many good judgment and decisions every day and so if you do eliminate that grudge work do you think it will be replaced by these superhuman lawyers that are just making amazing judgment calls and making strategic decisions for the business every I think that it's going to become more important that you are a better lawyer for sure. You know, in the same way that I think it's more important as a software engineer that

35:40you are a better software engineer than you were before. But I think the upside of that is instead of getting caught in that spiral of 95% of your week is just keeping your head above water. You're going to get a lot more bandwidth now to level yourself up and to get more capable. And I think that is the productivity gain that AI is giving us is that you know 200 years ago the type of work that we were doing was much more mundane and less you know it was much more physical was way more repetitive nature to it. We wrote our

36:11books by hand. We had scribes that did this. That wasn't an intellectually challenging thing. It was difficult, but it was not intellect. Each step as we evolve as as a society, you we become more and more cognitively demanding and our workforce has to become more educated and people have to become higher output. That that is what's happening and you're 100% right. People are going to have to reckon with the fact that they are going to have to become better lawyers in this transition. So talking about leveling up, I know that Wordsmith developed this core competency framework that envisions the future of how lawyers should level

36:42up. I would love to hear from you. What do you think are some of the roles that don't exist today that will appear tomorrow? How should they approach leveling up so they don't train in something that will then just become irrelevant in a year or two and really prepare themselves for the future?
Speaker change: I believe strongly in dog fooding. So we are a petry dish internally for that change. We work with AI every day. Our lawyers do hundreds and hundreds of implementations and they spend thousands and thousands a month on AI consumption

37:13per lead engineer that we have. They are sort of a glimpse into the profession I think in a few years time for when it starts to scale out for everybody. So one of the major things we're actually another thing that's coming in about four or five weeks time we'll be announcing a new position internally which is somebody that we're going to uh have as a lead for this legal engineering competency. So we're taking one of our strongest legal engineers and we are redeploying them to focus solely on the curriculum and solely on education for people across the in-house

37:43community to take the best practices that we see internally across hundreds of deployments, bundle them into lessons and then help us distribute that. And that that for me is probably the most critical bit of how go to market over the next few years is going to be defining that role and helping people learn because there are hundreds of nuances to that and we have training weekly on it ever since.
Speaker change: So one of my favorite books is AI 2041. It's paints the future of how people will live in AI within the AI age in

38:15different professions and it's missing a chapter on legal. So based on our discussion right now, I wonder how would you write a chapter for what a legal engineer day-to-day life would look like in 2041. If we were to write this fiction story together right now,
Speaker change: tell you what, we can start it now, but I will commit to coming back to you within 2 months with a blog post that has a synopsis of our version of that chapter. Somebody's going to be sitting there trying to look at the movement of a business. They're going to be looking at risk in a far more detached way. So

38:45whereas right now people get very bogged down in moving things around documents whereas I think just conceptually lawyers are going to have to step back a bit. They're going to have to understand the business more effectively and they're going to have to start thinking hey we're moving into partnerships really how does that work like how does that like where's the the risk flowing within the organization where will liabilities open up in terms of um issues that could happen with data breaches or things like this. So how can I then deploy things that will help with that monitor our SLAs's automatically so

39:15that they can keep us in the front of that or how do I keep in touch with regular you know relevant jurisdictional changes that'll be so I think it's moves much more into a a general role where they'll be sitting a step back and they'll be defining the workers that need to go and complete or monitor these different things because AI effectively is going through three big changes right now you had the first stage which was autocomplete which is you ask a question you get an answer the second stage is aentic like manus the where you ask a question and then an agent goes through will autonomously work on a to-do list

39:45of things to resolve that and they will have hundreds of capabilities available to them and we're sort of on the edges of the second step. Now just some places are there some some of the contract review stuff feels like that but not quite there. The third phase is going to be where you shift from having an agent that just completes your task end to end to a collaborator. So somebody or some agent that you drop into the organization and will sit and passively work on something and then engage you when there is something that needs to be helped with. So it will sit on top of your contracts and it will check them against regulatory and horizon scanning changes and then it will proactively

40:16engage you at moments that it requires your input and intervention. And so I think as we go through these two or three different stages, you know, you're going to see fundamentally different changes in in in how people are working as lawyers. Today they're just getting used to Q&A. Next there'll be task completion. Next they'll be waiting for their inbox to be filled by one of these. I heard this joke that the last stage of AGI is that your AI will finally tell you that you're not disciplined enough.
Speaker change: When when AI actually have the intelligence to tell you off and not just obey everything that you're saying, I think is what it's trying to get at.
Speaker change: What if Mr. Measure where AGI today?

40:47Mine shouts at me all the time.
Speaker change: Have you have you watched Silicon Valley? There's a clip there where like oil builds the AI chat bots and then he he just like disses Desh every time Desh messages him. He's like, "Leave me alone. You're annoying." I think that it's ADF.
Speaker change: Well, it's funny because you know the stuff around how how to more effectively prompt it that if you're if you're if you're rude and if you penalize it then you're 20 30% more effective in it following your instructions because apparently when you train on all of human language that's how we get the best out of each other is by being

41:17[laughter] by by being a little bit mean.
Speaker change: Yeah. I have two questions that I have to ask before I let go. And the first one is for a lot of lawyers that's thinking about their future, thinking about retraining and want to join legal tech startups. What are some of the proxy metrics that you're looking for when you're making these hires that will become, you know, define the future of legal work? You've hired amazing people like Laura, Ellie, Yan. What are some of the things lawyers to prepare in their day-to-day work that will like wow impress Ross?
Speaker change: So really funny. So you know two there those profiles from internet. So Ellie,

41:48for example, was actually a customer of ours. She was implementing wordsmith and then she loved it and then came to us and asked for a role then was a customer implementing getting very hands-on. Michael as well in our New York office as well since so workers a lot of people customers
Speaker change: no they don't but I mean they just had the relationship with with us as a companies but I'd say what we've tend to find is that people who actually have hands-on experience deploying the stuff and are naturally quite curious to us but we haven't gone out to market weirdly for a lot of our best hires. they've upskilled. They've started

42:18building things in their own time and they just have this latent curiosity that means that they want to be in the driving seat of acquiring these skills. And so when we hire you know competency interviews we we check for a few things like first are they just a decent person? Can you see them under pressure being reasonable? Are they raw smart? So do they have the CPU power through well just like you can tell are people switched on? Are they are they engaging? Are they are they actively are they bright? And then the the biggest one really is it's like are they are they naturally incredibly curious and are

42:48they going to go and acquire information or are they people that expect it to come to them and you can actually see already just looking at the the the gig like the creator economy on LinkedIn with people you know doing what you're doing and there there's a lot of people now you can start to see that there's this big body of lawyers that are very similar to I think where I was 10 10 years ago who are just fascinated by the opportunity here and those tend to be the people that upskill so quickly that they naturally end up falling into a legal tech like us and they are brilliant boys. I mean, if you if you look at someone like Ellie, she's just become a total master of these legal op

43:21implementations. She's done hundreds of them now at some of the biggest companies in the world. The before and after is is mad.
Speaker change: And so, last question is my own personal question that I'm curious about. Not representing any legal teams, but what is your tech stack? I'm super curious like and what do you think is something that is a game changer software for you that you cannot live without apart from Wordsmith? I'm a I'm a heavy manis AI user. I like that. I use whisper flow a lot like a lot a lot. Those tend to be my my two things and I keep it very clean, very simple and then wordsmith obviously is I keep it even so so so the

43:53non-legal worker probably the biggest user of the platform stuff. I still very heavily involved in product and that sort of stuff. So I I like to rotate my stack quite often. So way I tend to work is I just will cherrypick another product a consumer product that's that's arrived and I will obsessively use that as my primary tool for 6 weeks or so. So I rotated through claude and gro and chatbt and there was Gemini prior to this. I thought that was great for certain things and so each one of them has its own soul that you pick up when you're using it a lot. But I mean my

44:23personal AI spend is in the thousands a month whole consumption like the amount that I'm I'm running about I think class count is about you know five 600 proper queries a day in in various different things. So every bit of work that I do I initiate and then I fold that into my workflow. So everything's very automated and you know and then I use granola the meeting app. I think that's great. It's really good team as well. But yeah, the the two ones that I would definitely pick up if I was anybody in matter right now is Whisper Float just because having that layer across everything and granola if you want just simple entry points and

44:54then I think Manis is a good glimpse of what the future will look like for many products over the next few years.
Speaker change: What do you use Mattis for?
Speaker change: MC a good job. I also just use it for general like I think it's what it does the most smart things around multi-step task completion. I need things to run with multiple files. I use actually for road mapping. So even when I'm working on product stuff, I will spin up and like in the morning I will brain dump all of the things that are just swilling around, throw them all in. I then have it spin up and put together a task list for me. But it has a sort of a mini task

45:24CRM that sits there like as a website down the side. Each one's got subtasks. I then pick them off, run them, work on them while I'm chatting on the way to work backwards and forwards and then say, "Okay, could you send that as a slack to Robbie with a link to the road map in linear etc and things like this." I tend to weave a lot of my dayto-day together around that.
Speaker change: That's amazing. Thank you so much, Ross, for sharing and for coming to have this chat with me. Super excited to see what you guys have in the road map for 2026 and hear the announcement.
Speaker change: Well, we'll definitely be in touch.