The AI Dividend: David Bushby on Corporate Teams Going It Alone
44m 35s
The speaker begins by reflecting on a personal "aha moment" with AI agentic solutions that autonomously perform tasks, revolutionizing their work process. The conversation then shifts to legal technology, where the initial question for new demands is whether to build in-house solutions using frontier AI models rather than outsourcing. A significant portion of the discussion focuses on the lack of cost savings passed from law firms to clients despite widespread AI adoption. Legal operations professionals note no reduction in invoices or evidence of AI efficiency gains, prompting considerations to enforce such benefits through billing guidelines, quoting processes, and detailed invoice reviews. Concurrently, law firm rates are rising sharply, further incentivizing in-house teams to leverage AI tools to insource work. The speaker emphasizes the growing capability of foundational models and legal-specific AI, advocating for in-house counsel to take more responsibility for preliminary work, using AI to complete tasks before involving law firms primarily for validation.
When you see the whole crew, Claude architecture roll together, including web actions, it's like, oh my word. Yeah, but where are we going? I mean, it's definitely been this past, the way I explained it was, you had the ahamau when you first used an LLM world, and then this past three weeks has been that ahamau, ahamau moment all over again, with these true agentic solutions that actually do things for you, where you have something broken over here, you're getting it to perform a task, then you move to something else and you get it to perform another task. So it's like giving it to an assistant or a colleague, but it works in the background, and then it gives you a little ping, when it's ready, 15 minutes later, and panbettles say, here's the page in that I draft directly in-conference for you, formatting and everything, and then you say, "A quick look, and it's maybe good to go, or maybe need a couple of edits." So the way of working has been really, it's changed for me personally, in a massive way. I've always thought there's a ton of value in deep work, but this is like multitasking on another level. Yeah. When we're flipping it to the legal option, legal text site, I guess when we get a demand in now, the first question is probably not looking at the market, is probably more a question of, "Could we do this ourselves? Could we build something here with a frontier model?" That's sort of interjecting its place, and then I suppose if that is the thinking, how are you seeing sort of, you know, I'd say you guys have got a pretty mature text stack, you know, a diverse text stack as well. Is that something you're thinking about? Yeah, all the time. Hello, and welcome back to the latest episode of "Law What's Next." The podcast where Tom and I speak to people, we think are super interesting and informed on the topics of legal tech and all things happening in the world of law. And we've got really interesting guests today. Tom, do you want to do the honors? Yeah, we welcome David Bushby, the head of legal upset camera, and also the prominent author behind "In Council Weekly." Very much a bite, says weekly newsletter for in-house counsel that's wildly popular. Welcome to the show, David. Great to have you with us. Thanks, Tom. Thanks, Alex. Big fan of the show myself. So I could be here. David, I'm conscious you run the legal ops function at Canva. So maybe just for quickly for the benefit of the audience, if you could just give a bit of a scene set of like, not just the role but the scope and responsibilities, that'd be great. Yeah, sure. So, yeah, legal operations at Canva. There are four of us dedicated to the role, including myself. So myself for legal ops manager, we're called systems engineer as well as a parent legal. So, yeah, we're in a fortunate position to be entirely dedicated to legal ops function. That's all we do. Day in, day out. Supporting a global team at Canva. We're approaching about 80 or so in that team across quite a number of jurisdictions. So a lot of the lawyers or attorneys or team members are based out of Sydney or other parts of Australia, the major cities typically. But we've also got a very large presence about a fifth of us, I think. Also, a based out of the US. So primarily Austin and also San Francisco, by area. But also team in the Philippines and the UK in London, where I think you interviewed Saham at the time as well as China. So, yeah, pretty global mix. And I guess we do all things systems and process related. And a big part of that at the moment, of course, is AI with the generative AI. All that jazz is very much at the forefront of what we do for it. Blessed to work in a place like Canva that really promotes that. And we've had access to all the great tools for our best part of probably two years now. So we've been on that journey for a while. And we do all the sort of outside council spend and management. That's another big, big focus of ours. So I'm sure getting into that conversation through the podcast. Maybe less games are that straight away actually, David. Because I think when we caught up a couple of weeks back, one of the things you wanted to get our sort of taken insight on was, are we seeing reflected in whether it be our fees or even our fee arrangements or our billing or even down to the itemized line items on our invoices, any sort of inkling or idea of productivity enhancement inside law firms. And it was clearly so in a passion for you. Have you discovered any in the last intervening couple of weeks? And yeah, maybe let's dive into that topic straight away. Yeah, Tom, unfortunately not. And yet to get to hear of someone either in our team or any other colleague around the globes as they, hey, listen, we got a quote for X amount of dollars on a piece of advice or piece of legal work and we got it back and it was half price, for example, or substantially cheaper. And when we asked our phone why that is the case, they said, oh, we used one of our new AI tools. And so therefore it was a quicker and therefore we changed to less. So I'm yet to hear that anecdotally. And also, we reveal a lot of invoices of course in-house and sort of pouring over those line islands. There's nothing in it, the line out of level. So I mean, I asked you guys, have you heard that type of evidence or anecdotal seen it in the flesh and maybe in the listening? Seriously, if anyone has seen or experienced that, please let us know through whatever channel you can. Yeah, definitely. It's that definitely, I don't think I've seen or heard anybody talk about it as well. So it was funny when you were talking about it, David, on our pre-sort of chat. But no, I haven't seen that. And yeah, I definitely would put that call out to the listeners, maybe we even put it in the post when we push this out there, a call to action from them. But no, I think it's interesting. And it's such a funny juxtaposition, isn't it about the how the conversation is couched and what we expect to see versus what we are seeing. Do you think that will change into the point of, I guess, in-house teams or legal ops teams that help look after the external council management? Is that going to find its way into, I don't know, if billing guidelines is the place because they're, you know, people aren't pouring over those, but is it going to come into maybe the way you have to then instruct law firms, you have to prompt them to give that as part of their quote? Could you see it going like that? I probably can. That's probably the next step now that we're fairly convinced that's not happening naturally. I mean, this particular question around the so-called AI-dividend pretty obsessed with since about the start of last year, early 2025, in preparation for SAP out sort of we have a summit where everyone in the legal team comes to the headquarters here in Sydney and as part of that. I, by spoke to a number of law firm partners or consultants or innovation folks in large law firms, largely locally as well, just to really ask that question, say, what are you thinking here? And this is early 2025. So, adoption was starting to pick up, I think, the tools were certainly available. So that was when I first started to ask that question and it culminated in a bit of research and not taking it on my side and then we did panel of outside council to come and join our summit to talk to this very issue and at that point it was, there was no clear yes or no, there was probably a lot of hope and enthusiasm that that would be the case that surely the average cost of a matter ought to come down. If you're charging by the hour and we're seeing double-digit percentage reductions in time spent within that ought to reduce the cost of a matter. Maybe not in all cases, but it must be, it's got to happen at some level and so it was maybe the direction of travel that they would be hope and then yeah as time went on and adoption kept on increasing and seeing all the headlines at law firms adopting the major players in their workflows. We're just not saying that evidence come true. I just started to feel by the end of last year, okay we just really need to go at a loan in terms of we can control what tools we use and we can somewhat control the adoption rates on those tools how effectively we use them. So therefore that AIDVD and maybe it's just up to us on the in-house side and we can't just sort of sit back and wait for a external council to lead on passing through that economic benefit. So that's kind of where it's added and to your point what's the next steps like how do we start to really enforce that is that through the quoting process or the RFP process getting some sort of something in writing, some undertaking some promises to pass on. Is it through the the billing guidelines as well like it's another mechanism potentially where you can have firms and we do have firms agree to our billing guidelines they click a button at the point of uploading their engagement with a 12 system where they do acknowledge they read and then our guidelines will take precedent so their engagement letter. So do we put it in there and you know we're yet to do this year's version of the billing guidelines but that's another mechanism and then probably at the back end as well at the point of invoice review start to look at line item little narrations and say well he is a clear candidate
that ought to have been used by ought to have been. There should have been AI involved in the production of that particular line item. And it seems here that this line item has taken X number of hours and we do require our firms to use AI in our billing guidelines in all sort of reasonable applications. And it's clear here that this hasn't been the case. So do we sort of create that as a additional flag at the line item level? But hey, is all the traditional billing guideline breaches, but then do you start to overlay an AI analysis and we've started to experiment with this and say look there they ought to have been a certain percentage saving on this line item because clearly could have been done faster more effectively with AI. So those are maybe three touch points where you put in the practical you make it happen by practical mechanisms. Yeah, I don't know if this is the the squeeze point, I guess, of some of the height because there's an element if I assume the best of intentions and maybe the negative is just law firms are getting the savings and they're just pocketing them. But if we put that out of the picture for one second, the maybe a more likely reality is for a lot of firms, they just haven't got it in slickly enough. They haven't deployed it enough well. They don't have the adoption rates well enough. They haven't got the training well enough and actually maybe they are double checking or triple checking or whatever they're doing the outputs. So actually the gains aren't there for them to pass on. But the reality, the squeeze on that is that's not what's being pushed out there in the PR machine because obviously they want to sign up to these pretty high profile deals and the sort of get the up shift, you know, the halo effect of being seen as being a progressive firm. But actually then, let you say this is a choke point or a squeeze point where their larger clients, at least are going to be starting asking the questions that you're asking, which is, OK, I see these press releases, I see your name sponsoring these events, I see you talking about case studies with, you know, a large provider. But I definitely don't see the line items. I definitely don't see the reduction. I guess that's the double, I've sort of the nature of the beast at the moment. Well, yeah, I mean, it almost gets even worse because not only that we're not seeing the savings with we're seeing the incisive exponential climb of rates coming through. We see them come to our e-building platform firm submitting these rates and it's double digit increases typically across the board. It was a really interesting report put out of one of the major e-building providers on all of the sort of the RFP data that was coming through and granted that it's more RF probably more RFP style work streams that have been quite. I mean, the average, the averages for for the US, I think, rates are up about 12.6% UK rates up just under 16% Europe of 8.2%. But we do a lot of work with US Council and UK Council and you know, double digit increases of, you know, 12 to 16% I mean, sure, we, but we get plenty. I just pulled up, I just pulled up our latest sort of batch of increases from one large US firm that there's stuff in there is 25 to 35% increases on already really expensive rates. And that's, it's just, it's been, it's really tough because you're trying to contain your legal spend and budget within a particular envelope for the fiscal year. And all things being equal same amount of work, you're looking at 20% or at least a double digit increasing your budget for no particular gain. So you've got, yeah, rates rising really quickly on top of already expensive rates and then, then we're not seeing any, any evidence at all of a I say it's coming through. So we're, we're so interested. What do you do? What do you do? I mean, I mean, and you guys, I love to know you guys feeling that similarly in your neck at the words of your company. We've come to things immediately jump out to me, I'm sure they do to you as well Alex is like, I half wonder if some of the rates rises are almost anticipatory of something which you flagged early on in your discussion, David about the potential for in house teams to start insourcing more work. So, you know, our rate increases in some sense, almost anticipatory, my next thing I jump to is other demands signals already there like Alex and I sat down with Nick West in London, who's the chief strategy officer at mission. And he talked to certain elements of work. There are some demand signals that suggest in house teams have enhanced their capability with AI to do more work. So what might have used to constitute two hours of a part of science to produce a document for XYZ client is now five to 10 minutes reviewing a document for a certain client. And so, I half wonder if these increases are at all anticipatory. That's one thing I'm thinking about. And then the other thing Alex, we had Kevin Cone on the CEO at BrightflaG and he talked to something really interesting in their 2025 data, which was there seeing an increase in part the time and utilization on matters, which was fascinating. Because the whole model and the pyramid model of law firms was there, they're much more focused on business development and the idea that you delegate down and I found that equally fascinating. I don't know about you Alex, if you had anything to build on there. Yeah, no, no, same. Yeah, I think that's it. And yeah, to answer your question, David, I guess, yeah, we're exactly the same as you in terms of seeing those high rates increases. I guess the backdrop of inflation has been a bit of cover for that as well, like it's inflationary, but way beyond inflationary in terms of the actual increase. So there's a bit of a laugh at that as well. But no, yeah, definitely to your points, Tom, I think that's maybe a bit of sandbagging, but a preparation for the future. And like you say, rather than seeing a senior associate or a junior associate leveraging the tech and giving back some value. It's actually sounds like partners trying to leverage the tech, but still charging you out at a pretty high rate. You don't see any of that benefit yourself. So yeah, it's a it's a messy, a messy mess at the moment. And it feels like everything's pointing towards in how seems just trying to take it back themselves, right? I think that's maybe David, you want to see stuff come through your doors first before it goes anywhere near law firm. If that's the way they're going to play it. I mean, I've always wanted to keep as much that's kind of how my default to mechanism is look likely that's let's keep it in house. This is prior to this environment is hyper inflationary rates environment that we're in. But now I mean, there's just even more incentive to not send it and that just your point there, Tom with Kevin saying that if this if partners are putting more time on the sheet. And that's that that's so scary because I mean, I'm just looking at a partner rate that's $2,935 USD per hour. And you go, well, okay, well, if this if that's the direction of that, I was going to do proportionately more work on the matter compared to sending it off to junior is even more frightening. There's even more incentive for us for us to say a lot. If it needs to be if this is the firm that we need to go to let's do as much as we can in in house. So it's an interesting one. I mean, I feel for the firms on on that front where they where they where they must be thinking, well, what's the lesser evil here is a is it just not getting is a. Passing the benefit of still getting the work but demonstrating and minimally passing on the benefit of the idea to them. But you know, earning a little bit less revenues versus not earning the revenue at all because your client has done the work internally more and more. I mean, it's generally so interesting to see that shake out and we certainly encourage our team as much as possible to do as much as we can with with the a. We have the legal specific tools that we have. We've got tools from kind of the so called legacy era. I mean, these tools are only 10 years old. If you're doing pretty old now. Then we've got the new legal text that had more AI native born in this generative AI era. Then we've got the foundational model to make we're lucky to have. Claude, Jack G. B. T Gemini and others clean as another good example, like all on enterprise license for at least a couple of years. And on top of that and again, we might get into it. You've got the likes of Claude rolling out co work skills called code, Chrome extension. And there's just the the aperture of use cases and utility is just blown right open again in just the past few weeks. But I think on that there's never been a better time to actually ensure small work. As they're in house like in terms of what each of us have access to in terms of whether it might be legal research databases. You can now intelligently pass through your context on that information. So you can tailor it to your risk profile, your organizational objective. So like really do feel like there's a responsibility on us to shift some away from the law firm. Pass through to them partially complete outcomes, you know, they become more of a validation step in you know multi chain process that is much ever increasingly technology driven. I think it's it's getting so interesting. We've chatted offline David about the capability of legal AI tools, which is obviously incredible.
more and more the capability of the front-end models and particularly the way the front-end models can utilize tools to deliver outputs to you is it's fascinating right now. And in terms of mandate to your team David, I think we spoke about like any tips and tricks for how you get your in-house counsel to leverage the technology more, to stop and think before they go to a law firm for advice. Yeah, I mean like I said we were very it's I think largely the function of being at Canberra that is a AI company so that being the case and with the messaging from the founders right through senior leadership all around using AI within the product but also for our own productivity to get our operational work done in running the company. I mean that's been the strong themes or over 12 months now so when you have one of the founders maybe two and a half years ago maybe even three years ago sitting in with us at our legal summit helping us use chat GPT which was brand new at the time to draft letters of demand showing us how to do that using chat GPT. I mean it's just such a strong signal that as a legal team we need to be using it so we've advanced very quickly because the culture that we have in our team but we've had the available resources from the IT level all the way down we've just had this access and enable at the at the Canberra level lots of training and enablement and then that's enable our small teams in say legal ops we've been able to run courses on custom GPT's building clean agents we wanted to do the same thing with Claude skills in co-work same thing in Claude code and so the lawyers on the team are already building things in Claude code they're already building Claude skills so they're just there's a handful of those that are definitely at the forefront and pushing the tech within weeks of its release like pushing it to its legal use cases so so there's we've got that sort of tailwind with us and then the next piece is I guess how to get people to sort of think about this a whole team to think about this when we look to outsource work we do need to get a proof of for that we have a request form and there are a bunch of things in that request form that are just really to set up a matter of now um building system but maybe a third of those questions what we call to own as mindset questions and and they are literally questions of say like why does why do we need to outsource this why can't this be done in house so you do need to answer that question for every piece of work that we outsource if they if they picked a top tier or you know large more firm then the next question is I have you considered the smaller firm if if it's going to be on hourly rates the next question will be like what have we discussed uh it's fiendies is there a discount have you asked for a discount like it's really common to see 10 to 15 percent like and so all of those every time someone says I I need to outsource some work I need to spend some company money those questions are mandatory and they do have to answer those and then there's an approval process that there are thresholds we don't get everything you know penny and dime approved but there are certain thresholds where it goes to the next level or go to lead and at one point at some point it does go to our CLO depending level of spend and so the answer is those questions need to be there they need to be thoughtful and we have a record so I think you know having folks put their names to you know request money answer those types of questions that make you think how hold on could we do this ourselves do we need to spend money could I narrow those go in some way I've thought about we haven't implemented this but a specific question I like how have you used its tool AI wherever it may be to to help answer this question we don't have that yet but anecdotally we would definitely have have the team fairly consistently used those AI tools because the availability is there they're pretty clued up and skilled already prior to considering outsourcing so so there's I think there's more there's more things that we can press on for the behaviors but that's what we add at the moment yeah that's cool and you mentioned there then there are some people in the team who are looking at using Claude code to make something to create a piece of tech I guess or to create a solution of some kind which is super like really cool and I guess that they're vibe coding it because they're not necessarily devs or they're experts in the law but not necessarily in building but they can use this tool to make something really cool I think a lot of people we interested about that because obviously we've seen a lot about vibe coding but as far as the challenges taking a vibe coded idea and then bringing that into production I guess within the legal team are you seeing that like are you able to take some of these ideas and make them useful or is it something that's an individual lawyer might be able to build something cool for them and then they just deploy that to their working practice or just be interested if you're able to share David though how how is that going and maybe because you're a you know modern techie focused company maybe it's a bit easier for you to do that but we're really interesting to know as anyone sort of cross that chasm in your team yeah it's been an evolution from probably the first things but the custom gpt you can create those and then share them across the team but that was maybe the first step in creating some sort of scalable shareable solution that's AI paste and then we we use Glean as well so you can create agents with Glean that's maybe one step more structured compared to a custom gpt to creating an agent and again they're shareable internally but you do sort of operate within some constraints and chord skills is probably the next sort of step up again you can create in chord co-work that is you can create a skill you don't really sort of share that by URL like we when someone creates a skill like we share it in a Google Drive folder and you can download the skills file but the dot skills file upload didn't that into your skills menu in co-work and then then you can file that skill like trigger that skill I should say and it will so it's it's shareable but it's not as easy as sort of sharing your URL it's not like oh here's a URL to a tool I built try it it's more of a download of file upload a file now you can use it in your instance of code work the the v i coded stuff we we yet to solve that that little last mile problem so folks have v i coded really incredible tools so good like really complex domain specific legal research tool for example that that just follows lawyers the way that they would conduct research it's just step by step the way that they would do it the sources that they would tap into including sources that are not just websites then more kind of they don't they kind of sit behind logins and that sort of thing and they grab grab data from I don't know if it's city or high login actually but they're not just a a basic sort of search legal research regulation website there's a lot more layers to dig through some government websites to get where you want that's probably how what I'm trying to say and vibrating a solution that is able to take that and really dig through the full architecture of like a government website to grab that information from a website that has horrific UI for example and then pull that into a full legal research workflow and have that have that sort of sitting there it's it's built in called code then the next thing is all okay well how do I then how do I then share that with a colleague and at the moment you can kind of download the whole package of files that called code will create the next person can then upload it into theirs and try and run out but we've we've run into some issues with that so that I can't say that we've nailed that last mile the next piece would be okay well if we've got the code and I've built a few things in the last week or so that you know run locally on my machine that's great or is sitting on some sort of URL within the my code platform but it's only live while I'm while I get like a an active chat going so I can share it with a colleague that sometimes it's failing because there's no active chat on my hand so we really need to then export that code and self-host it in some way within our company environment so that's that's the next piece I think probably certainly by the middle of this year I think we'll have that part sort we'll find a scale of way to take five code solutions and self-hosts and within the organization structure somehow we've got a like a AI discovery week coming up in May and then as part of that they're the hack of long so perfect time to divide codes and solutions and then work with some debts to actually get it up on the net needed behind some basic user signing and all something like that and then we can have completely customized tools to our workflow the way we work that's that's really exciting and I think just around it or if you're enjoying our conversation with David please do rate the show please do [BLANK_AUDIO]
but probably more importantly to Alex and I, please do share it with someone who you feel would also benefit from listening to this conversation. Now back to it. - Yeah. - On the legal research case study you mentioned, I mean it's joyous. I mean you get involved in things like accreditation or two type things and you go through the motions and you think, oh this is a lot of effort like for this individual thing. And now if you go back and look at everything you've submitted for those things, it's just absolute gold. I think the other week I found something we submitted as a team which was all the regulation, policy and guidance that we feel we are subject to or monitor. And then in the world of like, Claude and creating skills files. I mean it's just absolute gold. What you can do with that now is remarkable. You can build newsletters which are rising scanning for your people team about labor or updates around the world. You know and it's like, it's a really interested moment, isn't it? - Yeah. The regulatory monitoring thing like that's been hitting on multiple funds but internally for us but also just being involved on various legal ops and in-house forums like the regulatory monitoring. It keeps coming out because there is an absolute explosion on all things digital and regulatory requirements. So it's fun and mind for so many people and I'm finding this power of being able to tap into proprietary APIs where possible. So that's what co-work can unlock that. You need the tools can now unlock the average user with no understanding really about the technical side of APIs can tap into these APIs and create custom solutions using proprietary data coming in. But then on the UI side, custom, I'm like, "Well, we get the data coming through. "We used to seeing that type of data on a SaaS platform "with search and filters and whatnot, "but we can't quite customize it. "We want the impacts for our business, for example. "We want commentary from XYZ to feed in here as well." So you can combine the APIs, design the UI of your dreams, then run searches on the net about the different laws and regulations that come through. So get that commentary from large firms or where it may be and then run more AI analysis on your particular company's priorities, essentially. - And even as simple as your organizational tone of voice, how you're like to communicate with one another. I mean, historically, when you used to pull from these sites, let's say, I think you described it as that awful with Angle Government website. You pull in a wall of text that is not canned for communication friendly. And so now the ability to sort of like pull in but then manipulate according to your communication style and different things like that. You're kind of taking over the UI because you know how you can control how it will land best in a way that these big organizations hadn't thought through. It's super fascinating. - Yeah, I mean, we work with a number of SaaS providers in the legal space and sometimes you've got your tool and you want it to do X. You want it to notify me when this happens. You want this little feature, that little feature. It's been quite amazing to be able to, I mean, I'm now getting into more like the core extent, like the Chrome extension because I think that that turns any website but particularly with your SaaS tools. Your SaaS tool that's open on its hub, you've got core Chrome extension happening on the sidebar as well. And that you can now get that SaaS tool to do things it's not designed to do because you can get the core extension to take actions on that SaaS tool for you all run analysis on the thing that's up, you know, that's being shown. You know, like, I wish there was some sort of analysis on this particular, in this particular, I don't know, it needs to analyze something. The SaaS tool does it in a certain way. The content is there, but I want it to be done in a slightly different way. So then when something like the Chrome extension you can actually take those actions, analyze it exactly the way you want in your product. It just, it's mind blowing. It's also the thing that when you watch that inaction, probably scares you most about the future, I find. When you see the whole Chrome Cloud Architecture roll together including web actions, it's like, oh my word. - Yeah, but where are we going? - I mean, it's definitely been this past, the way I explained it was, you had the ahah moment when you first used an LLM, well, that was two years ago, 12 months ago, six months ago. You have that moment and then this past three weeks has been that ahah moment all over again with these, I think true agentic solutions that actually do things for you, where you have something broken over here, you're getting it to perform a task, then you move to something else and you get it to perform another task. Some of these tasks would actually be a lot quicker if you did them yourself, but the thing is you just delegate it. So it's like giving it to an assistant or a colleague or someone that is painfully slow, but it works in the background and then it gives you a little pin when it's ready 15 minutes later or something like that and then it'll say, here's the page and that's that I draft directly in-comfortable and for you, formatting and everything. And then you say, I quit looking, it's maybe good to go or maybe need a couple of edits and then not only that if you've prompted it to then draft, but not send an email or draft, but not send a Slack message that's kind of where I'm at, draft that don't send. And there's, and you open up the Slack message and there's your draft, kind of just really to go before you. So the way of working has been really, it's changed for me personally in a massive way. It really made me rethink and it's just interesting this whole arm, the multitasking versus deep work debate and I've always thought there's a ton of value in deep work, but this is like multitasking on another level. You've just got a bunch of different browser that had running different jobs that are getting done for you and you just check back when you need to check back. So some of those smaller things can run in the background or maybe you do the deep work and yeah, getting when things are ready, but it is a very multitask centric is where I'm finally at the moment. Yeah, and maybe because we hammered the lawyers a little bit, which we can all do because we're lawyers, but when we flipping it to the legal ops and legal tech side, I guess this is opening conversations to you to say, when new demands come in in terms of like, what can we do process and tech wise that might have previously been, okay, we need to open up an RFP or an RFI and start figuring out what's out there is the sort of legal ops equivalent of saying, have you thought about the AI is that kind of already into your thinking to say, when we get a demanded now, the first question is probably not looking at the market. It's probably more a question of, could we do this ourselves? Could we build something here with a frontier model? That's sort of interjecting its place and then I suppose if that is the thinking, how are you seeing sort of, I'd say you guys have got a pretty mature tech stack, a diverse tech stack as well. How do we navigate this world now knowing that we've got this really powerful tool that's sort of in our grasp, we can deploy it where we want to and where do we stick with tools that we have invested in? Is it time to just get rid of everything and go with this stuff or is it time to push on some of your existing managers and say, look, we're now changing. You were delivering tech towards in this guys and now we're here, we're over here now, like how are you going to come and play with us over here? Are you thinking about that at all, David, yeah? Yeah, all the time, we've got some big systems that we've had for quite a few years and just on what they do and what they can do, could you replicate that? Could you via Coda? I mean, broadly speaking, yes, that they've been in better than the organisation for a long time. They integrate with salesports, they integrate with Zip, like there is the whole sales team that uses the platform, there's procurement, they use it the platform and it works. It broadly works and could you replicate the basics yet, yeah, you could, but I definitely do respect having worked as a product manager in startups for four or five years in previous lives and work directly in tech teams and I've seen the things that we build constantly break and I've never quite figured out why when no one starts to go, things break and there was constant updates and maintenance and all of that. So I have lived through that and very much understand there's a lot of to it. So we're not about to kind of vibe code our way into getting rid of some of our core solutions, but it's the small periphery ones where we're probably making a bit more progress where we think, oh, there's this net little thing, like yeah, and maybe the regulatory minotrames based something like that. There is available data, there is available APIs. It can be created to all ourselves or can we combine it with something that we have internally? Like you.
in air table to sort of do like a quasi-noco, but then using AI to tap those APIs and get something in. There's a lot of smaller point solutions that I think are absolutely viable to build yourself and customize to your needs. And then I will also say that there are, you know, those legal technology companies that have been around for a decade, like some have been able to rasp at the AI wave and kind of catch hold and then hang on quite well. And others, I think, have really just missed it. And so there's this certain amount of tolerance, I think, to sort of stick with a provider that hasn't quite caught it. Where you say, "Well, hold on, there's these things that your platforms still can't quite do as well as just doing it in Chatchee PT or Quarital or Gemini, there's just going to be a pressure I think there for those providers. There's a limited amount of time that we can wait, but you would have to go with the best to breed a performative tech, I think. But then maybe the final point, I know this, then a bit of time answering this question. The final point is when you're shopping, I get to hard one, eating looking for point solutions, like the thing that you demoed a couple weeks ago, maybe more, a couple months ago, it can be fundamentally changed in a range of, say, call it six months, to be realistic, call it six months, the thing that you might have demoed at month one could be fundamentally different and improved by month six, but it just takes a long time to go through the motions of doing the demos and going through your proof of concept. And so someone that you might have excluded at month one may have come lips and bounds by month six and it makes the buying decision very difficult, especially if folks who used to running months long six months or 12 months long proof of concepts, I mean, I don't know how you could run a 12 month proof of concept, but the way things are running now because the tech that the landscape changes within every vertical so quickly. So I think everything is being expedited and it does make the buying decision. Tough, tight to commit to a tool even for a more to year contract. I mean, it says, well, hold on, that's three year contract, hold on, what the tech landscape going to be looking like in three years time. I mean, one years, I've enough to stomach with the way things are moving. So they're actually tricky. I've been so dynamic. For sure. It's a super interesting watch the space moment. So we will include in the show notes a bit of a call to the action to our audience to get in touch with us if they've seen any law firm productivity boost or dividend as you describe it, David. I think for their benefit, though, David, where can they find you? Like, I'm a big fan of your in council weekly newsletter. How did they find that and how did they subscribe? Yeah, there's just a very simple website in council.com.au. That's just a newsletter sign up and yep, look for it. Hit that. I'm on LinkedIn, so feel free to hit me up. But I've got one of those email gates there. So you do need to put in my email address there. So probably not their most frictionless way to get in touch. But yeah, I'm sure I'll find a way. It's generally they're linked in all the news letter directly. We're going to get all your email now. David, they're going to come to us and say, you need to tell David I've got this in for it. But that's a good way, right? Yeah, no, that's great. It's been really interesting to speak to you because I think you're a very real person in the sense of you want to talk about this stuff. You're excited by it. And then you're also willing to talk about the challenges and the downsides. And also you guys, a Canva really, I mean, I think you're a super impressive team on the legal side and on the legal upside as well as a big fan of the company and the product as well. We super appreciate you coming on and sharing and I'm no doubt we'll have requests for you to come back if we can if we can ever do that in the future. Yeah, thanks Alex. Thanks Tom glad to any time. I'm sure things will have changed in the next time we speak. (upbeat music)
Podcast Summary
Key Points:
The speaker describes a recent "aha moment" with AI agents that perform background tasks and deliver completed work, significantly changing their workflow.
In the legal context, there is a shift in thinking from outsourcing to considering whether to build solutions in-house using frontier AI models.
A discussion highlights that law firms are not passing on cost savings from AI to clients, despite increased adoption and public promotion of AI tools.
Legal operations teams are exploring ways to enforce AI-driven efficiency, such as through billing guidelines, RFPs, and invoice line-item analysis.
There is a trend of rising law firm rates, coupled with no evident AI dividend, pushing in-house teams to insource more work using advanced AI tools.
Summary:
The speaker begins by reflecting on a personal "aha moment" with AI agentic solutions that autonomously perform tasks, revolutionizing their work process. The conversation then shifts to legal technology, where the initial question for new demands is whether to build in-house solutions using frontier AI models rather than outsourcing. A significant portion of the discussion focuses on the lack of cost savings passed from law firms to clients despite widespread AI adoption.
Legal operations professionals note no reduction in invoices or evidence of AI efficiency gains, prompting considerations to enforce such benefits through billing guidelines, quoting processes, and detailed invoice reviews. Concurrently, law firm rates are rising sharply, further incentivizing in-house teams to leverage AI tools to insource work. The speaker emphasizes the growing capability of foundational models and legal-specific AI, advocating for in-house counsel to take more responsibility for preliminary work, using AI to complete tasks before involving law firms primarily for validation.
FAQs
AI enables true agentic solutions that perform tasks in the background, allowing professionals to multitask more effectively. It acts like an assistant, completing work and notifying when ready, which has transformed personal workflows.
Currently, there is no evidence of law firms reducing fees or providing discounts due to AI productivity gains. Clients are not seeing lower costs or line-item savings on invoices despite firms adopting AI tools.
Teams can enforce savings through RFPs, billing guidelines, and invoice reviews by requiring firms to commit to passing on AI benefits. They may flag line items where AI should have been used and request corresponding reductions.
Law firm billing rates are increasing significantly, often by double-digit percentages, despite AI adoption. This creates budget pressures for clients without corresponding productivity gains being passed on.
In-house teams are incentivized to insource work due to rising external costs and the availability of AI tools. This allows them to control expenses and leverage technology tailored to their specific needs and risk profiles.
Teams have access to legacy legal tech, newer AI-native legal tools, and foundational models like Claude, ChatGPT, Gemini, and others. These tools expand use cases and enable more work to be handled internally.
Chat with AI
Loading...
Pro features
Go deeper with this episode
Unlock creator-grade tools that turn any transcript into show notes and subtitle files.