Name: Unmasking the Truth with Everlaw and Sandline: Real Time Ediscovery in Action
Uploaded: 2025-05-21T11:16:08.164Z
Duration: 59 min 48 s
Description: Unmasking the Truth with Everlaw and Sandline: Real Time Ediscovery in Action

Transcript for "Unmasking the Truth with Everlaw and Sandline: Real Time Ediscovery in Action": Good morning, everybody. Welcome. Alright. We can get this started. So, let me just make some quick introductions, and I'll give you an overview as well of what our session will entail today. So my name is Jess Wu. I I work at Everlaw. I'm a partnerships manager here in The UK, and I manage our partner relationships in The UK and The EU. With me is Greg Campbell, who's, our subject matter expert today and the principal solutions expert here at Everlaw. And then also with us today is Ralf Kaiser from Sandline. He's the CTO, and, he'll be giving a brief introduction in in a couple of minutes as well about his role and how we work with Sandline. So, today's session, as you might have seen, we're gonna go through an investigation. So it'll be a little bit interactive, and there'll be some role playing elements as well to to get us all in into the mood. So, in a few minutes, I'll go over the scenario that we've got laid out. There'll be some familiar names and familiar, a familiar situation, hopefully. What we'll do is we'll go through some general concepts and workflows around investigations. We'll show you some new and advanced AI tools on Everlaw, and then also how these will work, and fit into your standard workflow as well. And then we'll have some time at the end hopefully for a quick, question and answer session. Just very briefly, a a brief introduction about Everlaw and then I'll pass on to Ralf, to talk about Sandline a little bit. As you are aware, we are a cloud based investigations and eDiscovery software. Now if you've been with us for a long time as well, you know that we've been developing AI tools for a really long time, you know, over ten years. So not only have we updated our tools in terms of capabilities, but we've also created, our AI principles as well, to account for this high stakes industry. So I would encourage everyone to, look look at that and also bear that in mind as well as we look through some of the tools we'll be using today, including some of the more advanced, sneak peeks as well. And with that, I will pass over to Ralf to do a quick introduction of Sandline, and your role as well. Hey, Ralf. Buttons do click at the same time. Yes. Good. Yeah. Welcome, everybody, and, thanks for the for the quick introduction. Introducing myself. Again, I'm I'm the CEO of Sandline Germany. Sandline Germany is the the German European branch of Sandline as well, the global CTO. And I usually spend my day working with our clients across, Europe and Middle East, on regular regulatory and dispute resolution, projects, delivering on the project needs with our PMs and as well as supporting tech needs, coordinating developments for new project, demands that we come across. Sandline is, a a global e discovery provider, and an Everlaw partner. We do some more things besides Everlaw, but we already do a couple of projects in Europe and many projects in The US together with Everlaw where we enrich the, yeah, basically, the Everlaw technology with our experience in projects, and all the parts around it. So what we do is we support our customers throughout all the stages of the DRM life cycle. So, basically, starting from the point where we collect data, if needed and then, through the process of understanding what the project, needs, how we can reduce data, how we can help our customers working efficiently in the platform, like or others, and, get them on the quickest and, probably cheapest way to results, they need to achieve their goals. We started, many many matters really small, and then most of the time, they they grow and become much bigger. And Everlaw always had been, the perfect platform for this because it really helps, growing in the platform. And that's where we where we, across the globe, help law firms and corporations to grow into. We also develop our own products. That's something that comes along with the needs of, ediscovery projects. We're in a continuous change. We have platforms like where we can do a lot of things already. But sometimes it starts in, the places before the data even goes to to. When we start and come across collections, more and more we see modern data developments in our daily life. Our phones change. The applications change on there. We have, different data on and that's one of the, one of the areas where we try to, bring our knowledge into place with, our own development. So there is, for example, a modern ETL where we help our customers identifying modern data on phones after a collection if it was collected by Cellebrite or, other tools, to get them in shape, and bring communications together. Also, we're working on on other demands. If you are if you're familiar with ediscovery projects, which I expect, then you probably came across one of the very modern, issues, which is around modern attachments. How how do we download data? How do we collect data, that we don't even own on the devices that we collect from? So that's parts where we concentrate on. And as well, we're going into AI support summarizations, and we're working on other tools how we can support, the customer needs. We usually live on our customer's feedback, and that's really how we work long times together with our customers. We we listen to the demands in the projects, and that helps us to develop our software and and our services as well. This is a a room of of continuous improvement. So we would love to start conversations with all of you, if you need support, if you wanna work with, Everlaw, and then, yeah, I will will let you go and, let you, get this little, peek through the keyhole into the daily job of finding who, knew what and when, where, Jess and Greg will bring us into. Back to Jess. Alright. Thank you, Ralf. Okay. So, we'll get into our roles now for the scenario. So I think we we built this, the session as, with a tagline of trust no one, question everything, and find the truth. So a little bit dramatic, but I think given our clients, that that's probably, that's probably valid. So, the situation that we have now is a high stakes litigation. We need to investigate all the potential evidence that's out there. Our client is Enron, and they've come to us with this very last minute request. You know? They've given us documents, emails, you know, drawings of, you know, with conversations and and things that, that they've done recently or or actually not so recently related to their practices. And, you know, we need to see what the other side may uncover. So our roles are, I'm the, I'm the lawyer of the law firm that's been instructed by Imran to to look into this. Greg is my colleague and subject matter expert on Everlaw. So he'll be conducting this investigation for us. And then Raoul Kaiser is also our delivery expert and also a consultant as well, on this project because, obviously, project of this size, can't just be done in house, and, you know, with with, with the software. So we have the law firm, so myself and Greg, and also the, project consultants, Ralf. Our client is Enron, and, obviously, know, there's key people, right, who will be affected by this, who may be pulled into this. And, you know, anyone, at Enron, right, who might be discussing, what the opposition are trying to find out. And, obviously, the opposition, we have we have government entities and we have the other side as well, who's who's also, trying to, you know, find wrongdoing that's being investigated here. So what is at stake? Well, we actually don't know what the investigation will find. This information came really late stage, late in the day, and now the expectations of this very demanding client is that we we give them something tomorrow. Right? That we produce something. And then, also, we need to advise them on the next steps. Right? As you can imagine, they're they're pretty flustered at this point. So, you know, what what can we, provide them to reassure them and and, you know, what can we advise on? So this is gonna be an interactive session. So as you're going through this, I mean, we are all pretty aware of, you know, how how this took out in the end. But, you know, as as we're going through this, yeah, maybe have a think about how you would approach or a matter like this. Right? Obviously, these land in your laps, from time to time. We'll be using some polls as we go through to kind of gauge your, how how you would approach this and maybe kind of your thought process as we go and get some feedback as well, so that we have some interaction here. And then, Greg, will present, to us in terms of, you know, each step of investigation, and what we'll do. Alright. So with that being said, we will we will start kind of our roles here in terms of launching into this investigation. Now before we start even on to Everlaw, and, you know, in terms of scoping out this project, now that's something that, you know, we would expect our clients, to to engage with Everlaw and Sandline, you know, quite early on. So, you know, imagine me as a very flustered lawyer. This just landed on my lap. Now we've got tons of data. So much so much data that we don't know what to do with. You know, we're in a panic. We're we're flustered. So, you know, we've engaged with Sandline early on. So I'm now gonna turn to Ralf to be like, look. Here's what we've got. You know? How would you advise us on how to even begin, to begin this journey before even anything gets loaded into the platform? Right. So and and this is, quite the usual scenario that we come across. The flustered, lawyer on the one side, precious project, not knowing anything about the project, about the scope, what we're gonna find, and and jumping in. But it's it's urgent. That's the first place where it starts. So our job in general, is that we at Sandline will will support, to walk our customer, maybe the lawyer, maybe as well, the the the client itself, completely through the whole project from collection to production and help wherever we can. We already heard a lot of the the constraints and things, that Jess, mentioned, and that's what we usually hear in our first phone calls when when someone calls in and says we we have this urgent topic. And and now it really comes to the first round of of asking questions. And the first questions are are very simple. It's it's around timelines, very important, number of custodians, data sizes because that kind of drives how we at Sandline, need to plan this as well. If it's if it's urgent and most of the things seem to be urgent in the first place, sometimes they turn out to not be as urgent. But what we wanna do is we wanna be in the timelines of the project and and support in the timelines of the project, make our resources available, get them in place, and and use our teams there. So we are international. We're following the sun. We have teams in in, North America. We have teams in in Asia. Depending on the the next questions, we can, pull them in and and help. But even if we can't pull them in, we have our people working if needed twenty four seven, to to get things up. First, it starts with the collection. So we need, to to figure out if there is is anything that needs to be collected or the data is already in place. And when it comes to collections from phones and it's it's about, communication data, then then it comes to the places which I mentioned before where we need to bring in our resources to to to get the data in shape that we can can get it into to Everlaw. So we need to figure out what do we need to collect. And with this question on what do we need to collect, we come across the questions around data privacy and jurisdictions. Where are the people, how do we need to collect? Is there any constraints on collections? How do we, make sure that we collect the evidence, with the full proof of of, having done that in the right way and and being defensible, later on. So this is part where we can support or we can help the customer to to really get defensible data with the whole track of of evidence into into. And then we start, helping communications between the different councils as well, supporting on communications between opposing councils, clients, and so on because sometimes we just have experience what to ask or how to understand the requirement and how to streamline that into later things that we will see in the platform, how we support defining filters, searches, how we can help on reducing data, not really reducing data, but reducing the, amount of of relevant, documents that need to be reviewed in the first place. How can we how can we support there getting the right search terms together, with the councils? And then it comes to really use our vast experience on projects that we've, collected over the the really, last decade, not only in Everlaw. Everlaw is is more one of the the younger, platforms, and and that's the the modern technology. But our our experience is is a decade of of, working on eDiscovery projects. And then it comes to including AI technologies and where they come into game and where we can use them and where can, AI be helpful and where potentially not. So, basically, we are the litigation support and disclosure support and full service managed, service firm that's that's, helping here. We can always provide some some level of assistance when we go and, see all the the functionalities in our law that, Greg and Jess will show. Amazing. Thank you, Ralf. And, and, yeah, you know, as the trusted lawyer, I think having someone, who is taking care of all the key elements in terms of collections, right, in terms of the setup of the process, that's something that's really key while the lawyer is, at this point, as you can imagine, quite focused on uncovering the evidence, right, and trying to figure out how to respond and and how to how to, you know, take this matter going forward. So that is very reassuring. Now okay. So we know that Sandline can kinda handle the collections and then also advise on, you know, how to how to manage this product. And and, you know, now that we know kind of the key key timelines, my focus is now back onto the data itself and now that the fact that we have a mountain of of of this data that we need to sift through. Now as you can imagine, you know, the the facts of this case, you know, they might have happened a couple of decades ago, but now we have the benefit of, you know, newer technology, brand new technology, in fact, that can kinda help us with this. So, Gregory, as the technical expert here, you know, where would you start in terms of, looking through all this data? Yeah. Thanks, Jess. And I'll just get my screen shared in the background, to make sure it's there when we need it. Yeah. I mean, I guess if you look at Everlaw, it's kind of a single platform with a collection of feature functionalities that all sort of work together. I think your your firm has made a good choice in terms of where to start. So, we've been around since 2010. We've, kind of, you know, moved moved an awful lot during that time, and we've made sure that we've kinda kept up with, technological developments, of course, generative AI being the one on the tip of everybody's tongue. In terms of what you're looking to do, we've certainly got got some things under the hood that can that can sort of help you out with this query with kinda getting to the truth, quickly and finding, you know, some of those answers to help you with that meeting you have with your client, which I think is tomorrow. I think it's worthwhile noting, you know, they're they're a big client. They're obviously gonna be security conscious. And by using Sandline alongside Everlaw, having the Everlaw platform there, you're you're kinda getting the benefit of not only, AWS's security and security certifications. That's that's where we're hosting, in the EU. That's in Frankfurt, Germany. Also in The UK, that's in in London. But we've also got our own certifications, and we take security very, very seriously. So we've got ISO 2,701, 20 seven hundred and 17, and 2,718 certifications. We're SOC two type two, certified. In The UK, we've got cyber essentials, cyber essentials Plus. And further afield, we're also StateRAMP and FedRAMP authorized in The US, which shows we we meet the high standards that, the US government requires, kind of on that side of the pond. And just to emphasize, those security certifications are our own, and we're annually audited on them. So you can rest assured that your client's data, is safe with Everlaw. And the same thing, we were talking a little bit about Everlaw's AI principles earlier. We've been mindful in how we've integrated this into the platform, and it's a true integration. So if you're used to using Everlaw, if you've seen Everlaw before but haven't seen Everlaw AI, what you'll note today is that everything is seamlessly integrated into kind of your existing experience so that you can jump between tools and you can use them as necessary, as as sort of part of your workflow. In terms of what I'm gonna show you today, the place where we're gonna start is a a new Everlaw tool. It's currently in closed beta. Jess, your firm, is fortunate that you are kind of part of that closed beta. You've sort of taken part in that process, so you have access to it. So, Everlaw project query will be, going into open beta, will be going to general release, at a time to be, sort of confirmed. You can certainly get in touch with us to have that discussion, after this webinar if you'd like. But, ProjectQuery uses retrieval augmented generation or RAG, as it's commonly known, which allows kind of legal and investigatory teams like like yours just to sort of ask questions across terabytes of data. So let's have a little bit of a look at how that looks. Now my understanding is let me just jump into this project query tab. My understanding is that you you sort of don't know too much here. So I think what we're gonna do, we're gonna ask, a high level question. Let me just get my mouse out of the way, across the sort of corpus of documents that we have here. And this is, you know, genuinely a high level question. So it's about a new matter. It's coming urgently, and we're asking ProjectQuery to set out the key topics and issues that we need to start to investigate from them without delay. So there's a lot of talk about sort of lawyers, investigators becoming coders. We don't believe you need to be a coder, and this sort of technology very much helps with that. Because as you can see, I'm asking this question in a very natural manner. What I can do once I've entered the question, I can submit it to the system. And what we'll do, we'll have a look at a question that I I previously submitted. It's the same one, that that that sort of I put in yesterday just while that's generating in the background. But what this is doing, it's getting citation backed answers in seconds. So you can see we've got a breakdown here, about the matter. And, Jess, for your for your notes, it's, you know, it's pulling out various information. So for example, about the LJM raptor and condor arrangements that we we might want to have a little look. Lots more besides that, but you can see all of these have and and this is just finished in the background. I'll stay on this one while I'm here. All of this has been searched across, the document set. It's found a 63 potentially relevant documents. We could dive out and have a look at those if we wanted to. What it's done is to automatically generate facts from those documents, and that's so that your team can kind of look through those. If I want to look at any of those, I can pop that document open, and you can kind of see it in Everlaw review experience that we'll see a little bit more of later on. So it's very easy to kind of eyeball that. That verification step's really important. Obviously, everybody is concerned about the issue of hallucinations with large language models. And, yeah, Project Query is using that, RAG technology, as I said, which is utilizing both vector databases and also, embeddings together with large language models, behind the scenes here to make this work, to make the magic happen if you like. But that's why it's important that we do deliver these citations, that you do check them. Because whilst the answers are pinned within your document set, that is the whole point of that kind of RAG technology to pin your answers to the set to give you that increased confidence, in in in sort of what is, being output by the system. It is always worthwhile, to check that what is being given to you, is is is kind of correct. So we're able to check these. The documents are actually being, assigned a relevance here as well, and it's doing that automatically. So I'm gonna be showing you some coding suggestions later where you're prompting. This has all been generated from that natural language question that I've asked. So even for novice users, it's incredibly easy to use this. And then what it's down done out of all of these generated facts, it's kind of boiled this down to 55 facts used in the response. And you can see each of these has kinda got a reference here. And when I hover over that, you can kind of see that snippet there. So this is designed to be very accessible to get your team into a position so that they can, talk to your client tomorrow with confidence. I'm not sure if that is enough to get you started or if you think we should maybe kind of bring this to the next stage. But how how did that feel to you, Jess, coming out of that in terms of getting you set up for tomorrow? Okay. So yeah. I mean, I can see that very quickly. Right? It's uncovered not only relevant topics. It's it's uncovered, the documents relating to the specific terms that we're looking for. Right? So we know that raptures is a special, I think treating vehicle that they've been using. So we know this is potentially relevant, right, to to what we need to discuss tomorrow. But I do want us to maybe dig in a little bit more. So then at this stage, right, we've shown this is me kind of coming out of my role a little bit, so I'm speaking to the audience. So at this stage, right, we've seen that Project Career can really quickly uncover, you know, relevant information, relevant people, and, you know, relevant documents as well. So at this stage, you know, for the audience, typically, you know, here, where once you kinda know this kinda headline information, how would you approach the problem now, and where would you start investigating? So if my backstage manager can, display the first poll. And then also, you'll see a pop up in your in your screen at the bottom left hand corner. If you click on the green go to the polls tab button, your poll will fly through over to your chat and your poll and your q and a section where you can now, answer answer the poll. So let's give it a few seconds. So, yeah, at this stage, how would you approach the problem, and where would you wanna start investigating? With that, would you wanna start with the email investigations? Would you wanna go towards the known custodians at this point? Would you look towards the org charts or the team structure to see, you know, who would be most relevant? Or at this stage, you know, do you not know yet, and do we need to uncover more information? Alright. Should we close it? Great. Can we see the results? We can't see the results. Damn. Did you see the results? Oh, let's see. Okay. Looks like most people would start with known custodians. Alright. Let's see where our investigation takes us. And I think, yeah, that's we'll see if, Gregory agrees with with the next step. Alright. So we've seen a little bit about the issues and, you know, we suspect, right, there'll be some key key players in this. So so, Gregory, what's the next step here? How can we drill into this a little bit more? Yeah. And certainly filtering down by sort of metadata including custodian information can be a really good way of of kinda cutting through your data. And by using a platform like Everlaw where everything's kind of integrated into one place, you can certainly use those tools together. So I'm gonna show how you how you are able to do that, but I'm gonna maybe go through a slightly different path, because we're a little bit AI focused today. So we're gonna have a look at some of the, I guess, some of the more traditional forms of AI that are kind of available in, in in in a tool like Everlaw. So I'm I'm gonna kind of move to our clustering. Just wanna show you while this kind of while we were looking through this one, same questions generated in the background, but let's dump, sort of jump ourselves into Everlaw clustering. So Evaluor's clustering tool, there's no additional charge to use this. It's, automatically available, on any projects that you have, including partial projects. And what this does, I guess, kind of in common with the the more modern generative AI technologies, this is machine making decisions. So what the machine has done, using, sort of, an algorithm in the background is to identify which documents, are kind of, more similar to each other. And what I'm actually able to do is to kind of zoom in here. You know, this is a kind of a very fluid experience that that we have. When I look at this large cluster, this kind of risk management portfolio cluster, these are the cluster defining terms. So these are the three most common terms in here. They're not necessarily named as such within the documents, but these are concepts that link the documents together. And they're kind of defined within this this sort of cluster bubble that we have here. And each of these documents let me just zoom in a little bit more. So each of these dots here actually represents a document. And if I wanted to in a similar way to how I did in project query, I could actually have a look and and kind of see what's in there. What I have noticed, so, Jess so we we noticed when we were looking at our documents, that we, had some potential interest, around this kind of LJM, partnership and also around this kind of Raptor vehicle here. So I'm gonna select let me just center this a little bit more. I'm gonna select the, Raptor one here and what you can see. It's pulled me in some documents, but importantly, it's given me my top three defining terms. It's also given me the other seven. So it's giving me a total of 10 here. If I want to add in additional clusters, I can do that. So I've selected that LJM cluster, and what that's done is to slightly revise what we have here. Now, obviously, if we were kind of working on this is on on this matter in anger, and it might well be in anger as the meeting's tomorrow, so we're gonna have to move quick on this. I would probably want to dive down a little bit more, do a bit of further exploration. But I think in the interest of time, what we'll do is to maybe show how some of these technologies work together. So these cluster defining terms, can be very useful to help you to generate your your sort of search terms when you don't know too much about matter. So, obviously, we have found out information from project query. So this is another layer of information we're gathering, and we're kind of using the sources together to to help us drill down on that. But let's maybe create a search from these terms. So what I'm gonna do, I'm just gonna create a basic search, and I'm gonna use these top 10 clusters finding terms. I'm gonna hit on create. It's gonna pop me out into another window. And what that's done is to sort of, open me out into this this kind of, list of, documents here that are responsive to the search that we've kind of created, out of, clustering there. But we don't want to troll through all of these. I'm gonna show you various ways that we can kind of avoid having to do that in at least in the first instance. But, what I'm actually gonna do is have a look at some visualizations here. So for these, you know, 10,000 documents that we pulled out here, I can jump out and kind of visualize these. And I note that the audience said about kind of drilling down by custodian. Well, actually, using the tools that you have here, you could, of course, do that. I've actually got my custodian, filter pinned to my dashboard here. I can actually jump out. And if I wanted to, look at various custodians, I could do that. And I've got various different sort of operators that I can apply there. But instead of drilling down on custodians here, maybe let's have a little look at who's talking to whom, because maybe it's a little bit early in the investigation that we don't know too much about to make assumptions about who's doing what because there may be some obvious names in there. We maybe have some individuals who have more, sort of documents than others in here, but, actually, maybe there's some other players here, that that that we should be having a a little look at. So this is where we can potentially, use Everlaw's communications visualizer. I'm just gonna pop this out on the screen now, and I'm gonna point out I'm gonna explain what all of this means in a second. So I'm just gonna hand back to Jess first. But just to sort of set out what's here, we've got a conversation going on here between, the these sort of five individuals. We've got a smaller conversation going on between two individuals down at the left. And then we've kinda got these three individuals, kind of floating here. They're not having conversations with anyone, But if you remember, this is all on topic. These are all email communications about these terms that we pulled out of our clustering. So I'm gonna pause there for a second for sort of just to sort of digest that, for everybody to sort of digest that. Digest that. But yeah. Jess Wu, back to you for the second. Yeah. Alright. So looks like we've already uncovered, the key topics. So it's to me, it's a little bit concerning that, you know, it's it clustering especially, right, showed lots of mentions of Raptor and LJMs, which we know is the, vehicle that, you know, our clients been using quite regularly. So, yes, this this is now starting to be a bit be a bit more serious. And now, yeah, I think it is important that we drill into who who the key people are and what they might have been, communicating with each other. Now, have we looked at, so these are the key people. What about the domains and, you know, where where they might be communicating? Is that something that we can we can see as well through through this? And is there anything else that we can we can get from this, comms visualizer that we have now? Yeah. There there there there certainly there certainly is. I mean, this might be a good time to kind of run a poll here to see what the audience thinks we we should do. But maybe shall I explain a little bit about what these different things mean before they, before they vote on that? Does that does that seem sensible? Yeah. Yeah. Yeah. Can you explain a little bit more? Yeah. Absolutely. So, if we look at these these sort of circles that we have here, these are nodes. The larger the node, the more the email communications that are kinda present. This, again, a bit like clustering, is a very fluid experience. You can see Jeff has got more emails than Anne, does here, for example. There you go. And if I look at the lines joining them together, these are, we we call these the edges. So if we have a look at these different edges, here's a good example, Jeff talking to James. You can see that this line is actually thicker than that of Anne and Jeff talking together, so there are more communications between these two individuals. The arrow that that we kind of see here a little bit over the middle line shows that Jeff is sending more emails to James. So that could be useful if you're interviewing people. If, Jeff says, hey. I, you know, I never I never communicated with James at all. You you can kind of see here that there's maybe, you know, a chain of inquiry to follow here as to what's actually going on because it appears that Jeff is indeed communicating with James. When we have a look at the others, I think I mentioned before, this is a much smaller conversation. You've got Sarah and Mary talking together, and it's actually worthwhile noting here there are conversations going on between Sarah and somebody else, because it's kind of going back to to kind of our central point here. So there are conversations that Sarah's having about these topics that are not with Mary. But then, of course, I'm just gonna also sort of pause on the fact, we've got Vince, we've got Douglas, and we've got Kaye. They're not having conversations with anyone here, but they're communicating with somebody else. So I think I'll pause there, kind of, ready for the next poll, and then maybe we can have a little bit of a look at maybe who we should investigate, in in terms of what's going on. Yeah. Thank you, Gregory. Very interesting. Now, yeah, let's turn to the audience. And so at this point in the investigation, right, you've seen a couple of tools. We paused on, communication visualizer. So how would you investigate this next? You know, would you look at the counts of emails if there's a sudden increase in in terms of messages between custodians, would you start to investigate, custodian communicating with different domains, like different email addresses? So something that Gregory pointed out here as well. Would you also would you wanna start looking at a spike in communication outside of business hours? Or based on what you've seen so far, nothing is really standing out to me. Give it a few seconds. Alright. Let's see if I'm not able to share right now, but I can obviously share the results. Let's see if we have a winning streak here. So it looks like most people are saying that they would want to start looking at a custodian communicating with different domains or email addresses. Yeah. It looks like that is our winner. Alright, Greg. How would we, yeah, how would we keep looking at this? And, you know, how would we where where would you drill into next, I guess? What's most interesting, do you think, to investigate? Yeah. I mean, I think as the audience has identified, obviously, we we will want to have a look at what's going on with with with these conversations because there may be things of interest to us. But maybe these unknown conversations, particularly in the kind of scenario we've found found ourselves in here, are going to be useful. And one of these is actually standing out to me, and that that's kind of Vince because Vince was a, a a senior executive at Enron. So maybe let's see what Vince was up to. So if I select Vince's node, it allows me to add that in as a filter. What I'm then able to do, is to work out, you know, who who Vince is actually talking to there. So maybe I'll go to a high level, so that I'm I'm not having to look at multiple email addresses and look at the domains. So this is gonna look at the email, domains that that Jeff is kind of talking to. So there there's quite a few here, but there's one particularly large one in the middle here. And if I was an investigator kind of working in the early two thousands, this AOL email address would kind of stand out to me. Sorry. I should say, domain would stand out to me, as likely to be a personal, sort of domain. So maybe this is something we should look at. So what I'm gonna do, I'm gonna add this into my filters. And you can see we've kinda gone down from volume to relevance. We started, with 8,800 odd documents, and we've gone right down at the bottom right hand corner of the screen to 38 documents. So let's have a little look at what Vince is up to. So this brings me out to my results table again that you saw earlier. What I'm gonna maybe do here rather than breaking out for a full review, I'm gonna use quick review. So this will enable me to kind of look through my documents, kind of in a in a in a quick user friendly manner. But, actually, what I've noticed, on this first document that came up, we've got Vince emailing himself emails about, coping with ethical issues. So this may well be something that you want to get on top of Jess Wu before that meeting with your with your sort of client tomorrow, work out what's going on. Sure. There's lots more going on, but this does look a little bit like a potential smoking gun. Alright. Thank you, Gregory. Yeah. The we're getting a lot of information in very quickly, and, you know, it's it's it's mounting up in terms of the the stakes here. Okay. So we've seen you've shown me a couple of tools, right, that that gives us an overview that can kinda pick out key documents or key relevant things. Now how do we kind of put this together? As you know, right, I have, I do have a meeting tomorrow. You know, we we we probably wanna start looking and reviewing at these documents a little bit as well before we do that so we can, you know, assess the relevancy here or, like, you know, the the, yeah, what we have here, right, in terms of what we can use. Is there anything else that you can show me that can get me really prepared very quickly for tomorrow, as well? Yeah. Absolutely. So, you know, just to emphasize, what I'm showing you here today except for project query is in general availability. So the, you know, the following, features I'm gonna show you are in general availability. So you could use them in Everlaw today if you're an Everlaw user. But, we we're gonna make some assumptions here. We're gonna make the assumption that you've used some additional tools in Everlaw, be it search, be it filtering, be it dish be it additional use of the communications visualizer or clustering. You've drilled down a little bit more on your documents. But you're you're kinda getting to midafternoon the day before the meeting, and you've still got 5,000 documents to look at. So, I guess back in the old days, you may have urgently drafted in a team of paralegals to work overnight, and to to kind of work through that. But this is just a preliminary meeting, but it is an important one with this important customer of yours. So what we can potentially do here is to maybe start to drill down on these documents using some of the generative AI tools that are available in Everlaw. So to do that, the first thing I'm gonna do, which is always a sensible thing to do when you're sort of working, with, you know, sort of, I wouldn't say large numbers of documents, but more voluminous numbers of documents, is to have a look at doing some sampling. So I'm gonna take a fairly small sample. I'm gonna look at 31 documents here, and that's taken a random sample, of my documents that I kind of have here. What you'll note in my view, I've brought in some additional fields that don't have anything in them. So these, which have these kind of three stars next to them, that's indicative of the fact that they are generative AI features. They are Everlaw AI features. So we wanna get down into the the details of these documents quickly, so let's start using them. First thing I'm gonna do, is to maybe help your team so they don't have to read everything. And we're gonna have a look at generating some descriptions. And when we get to the face of the documents, we can look at some summaries there as well. This pops up a screen, kind of a cooling off screen where it tells me how many credits I'm gonna use to to kind of generate summaries of these documents. I can confirm that. That will then start generating, those in the background. And what you'll see, those will kind of start to drop in, in in kind of real time there. If I want to have a little bit more of a look at my description here, I've got a bit more detail here. It's worthwhile noting for longer documents as well as description. You also get a summary on the face of the document. We obviously don't want to clutter this for you too much. But when you get into your documents, that's where you can, kinda see that that additional level of detail. What I might also want to do, particularly if I've got longer documents or kind of report type documents, is to run some topic summaries. Again, I just select this. And, actually, I just want to emphasize at this point, to get the descriptions and to get these topics, these are just clicks. Anyone can do this. And it you know, as long as you can, sort of click a box on the screen, you can generate these. You don't even you certainly don't have to know how to code, but you also don't even have to know how to prompt because Everlaw is kinda doing the heavy lifting for you behind the scenes. You know, we're helping to reduce hallucinations. We're setting the length of the answer appropriately. We're getting to it to format it appropriately for you too. So when we have a little look down here, we've got some high level topics. It includes sentiment analysis. We've got a little bit of positive sentiment there, which is always nice. And then on this document, we've got a bit more detail here. This is obviously a longer document, but you can see we've got a mix of sentiment here. We've got a little bit of positive sentiment. We've also got some negative sentiment. We will also identify sort of a harmful or abusive sentiment there, which is quite useful when you're dealing with email communications, particularly in investigations. Again, you get lots more detail on the face of the documents. We just keep it brief here. Something else that you can you can do here that that may help you, I'm just gonna give you one example of it for for simplicity that you can run multiple of these at a time, is custom extractions. So I'm gonna configure a new extraction. I'm gonna call this names. You can do various kinds of extractions. You can do a a text extraction, so it pulls you text out of the document based on your prompt. You can call numbers, so that can include currency, for example. Days, times, useful to get off contracts, useful to get off of letters, just as examples, but also entities. And I'm gonna focus on these entities here. I have just copied let me just tidy that up. A basic prompt here. So this is where you can start prompting. So before you saw the button clicks, this is where you're prompting. I just want to emphasize this isn't a technical prompt. You don't have have to know how to code. It does help if you, for example, read out guidance or you speak to Sandline about kind of optimized methodologies for doing this, but, actually, this is very natural language. You know, any any kind of sort of data protection, privacy type work that you've undertaken, you'll recognize the terminology here about natural living people. I'm also asking it to exclude out names of organizations. I can configure multiple of these if I want. I'm just gonna run this one. I'm gonna start generating that once I've hit my confirm button. And that's gonna look through these documents and kinda start pulling out names. So particularly in the context of an investigation, this might help to give you some additional insights as to, you know, who, what, what, where, when, why, etcetera of of what's going on. You can see we've kinda got a longer list here. Not all of the documents will necessarily have names pulled out of them, but they're kind of generating in the background. Final thing I'm gonna show you, and this is really gonna help you to kinda chomp through these documents, quickly is kind of the coding suggestions. So I'm gonna click in on these coding suggestions. Again, these are something that you prompt just to focus in on the issue codes that we have here. So this is where you're tagging your documents. You're, classifying them through your review. What we're doing is to use coding suggestions to, automate that initial classification. We are not coding the documents. We are making suggestions about that. You can adopt those suggestions. You can do that on batch as well if you want to. But we we sort of current strongly believe that with the current state of the technology, having that human in the loop, particularly where you're kind of deciding about classifying your documents like this, is very important. So I've put various prompts in here. I've told us what I've told the system what the issue of Raptor is, obviously, you know, relevant to what we were discussing earlier, Jess. So it's about the complex and intricate financial structure known as Raptor. We found a bit more about this out from our, you know, investigations we did using project query, etcetera. But we've also got some sub issues here, which, are kind of, going to kinda build up our picture. I'm gonna run these. Gonna confirm that. And what that's gonna do is to look at these documents, in very much the same way that a human reviewer would, a human first level reviewer. I'm not conflating human review and LLM review, but a lot of it works very similarly. If you know how to put together a review protocol for a review team, you will be able to put together, your coding suggestions in Everlaw. And by doing this iterative process where you're maybe doing it on smaller numbers of documents, that's very similar to the sort of calibration exercise that you would sort of do with with human reviewers. But you can see these have started to drop in here, and there's there's kind of various different levels here. I'll just pick on this one. It hasn't got all of them. But the the coding suggestions stretch from yes and no, which are basically a hard yes on the prompt you put to a hard no on the prompt that you put in terms of whether it agrees or disagrees that this document is responsive to that. But also, in the middle, you have soft yeses and also soft no's. Here's a soft no as well. That can mean lots of different things. I mean, for example, it could mean that, this is a document that a human would struggle with. So, you know, that obvious human in the loop factor there is important here. Could also mean that maybe you need to refine your prompts. It's maybe not kind of understanding what you've put in there. And very much the same way as you would with human reviewers when you were doing a calibration exercise, you can kind of refine that as you go. So you can kind of run this as an iterative process to to kind of give you those answers. So far so good, Jess. Yeah. Thanks, Gregory. Okay. So you've shown me and my team a lot of time saving, you know, more parts of the workflow that we can we can adopt, right, using this new technology. And it seems like, you know, the quality, is will will stay as we can basically, right, look through the, the reasoning and the logic that it's given us, as it's, you know, pulling things back. So okay. That that's really good. But, you know, as we've mentioned, this meeting is tomorrow. You know, you must have another time saving, tool that you can help me. For example, if I wanted to create a memo or a document that can outlines, you know, the work that we've done so far, right, so we can present it to the client tomorrow. Is there something that can be done on Everlaw to facilitate that? Absolutely. There is. I'm just gonna spend one or two minutes on the face of the document. I think this, this kind of warrants sort of a further in-depth dive. If anybody's interested in looking at, that sort of our off after this event, please do get in contact. But I just want to sort of show in terms of these descriptions, you've got a report format document here, Hundred and two pages. It would take a reviewer a long time to look at this. We've created a description. It's a longer document, so you've also got a summary there. I can jump out to verify that this information that I'm happy with it. The topic summaries, again, this is a longer document. It's generated lots of topics. And sort of what you'll see here, we we've we've kinda got these topics here at the at the top. There's 18 more that we're not seeing, in terms of looking at those documents. But I can actually expand out on this. And as well as the top level topics, you're also kinda getting these subtopics. I can jump through the document to see where it's got this information from. Again, that importance of verification. It's pulled out some sentiments, some entities as well. Finally, I'm just gonna touch on the coding suggestions before we have that that little look at how we can maybe start to put together a memo for your customers. I'm just gonna look at this issue coding here, around Raptor because this is what we looked at. And you can see we've we've kind of got some, sort sort of takes from the LLM on this if you like. Over on the right hand side is my coding panel, and you'll see that somebody's actually tagged this document already for ethical considerations. So, when I look at ethical considerations here, you can see the LLM is actually disagreeing with that reviewer, and it says it's not mentioning the code of ethics or ethical considerations. If I agree with it, I can then remove that. And, likewise, for these particular codes, it's actually given me, the the sort of high level overviews. So you can see where it got its reasoning from. You can actually jump to it in the document. I can apply these if I want. Also notable that, you document. I can apply these if I want. Also notable that you can kind of do this at scale as well. So if you're happy with these, you can batch apply them to your documents with, obviously, the caution that you need to be confident in what you're doing and thinking about the purpose for which you're doing it at the end. For example, if it's for internal use, if it's for investigation, or maybe if it's for a litigation, arbitration, or regulatory matter. Let's jump out and have a little look at generating a memo, And I will do this in in just a few minutes. So I jumped into story builder, which is Everlaw's narrative building toolkit. What we've done is, using the tools that we have to drill down on the data, we've identified a certain number of key documents. You'll see these documents. We've pulled information from their metadata. We've also pulled their descriptions, in where applicable to, StoryBuilder. It's very much like a virtual folder. Events are like tabs. I can look in my 2,000 tab, and then I can look at my stickies, which are kind of these labels. So I'm gonna drill down to maybe look at some core documents. You know, this document from Madea, this code of ethics, that's potentially an interesting one for us. But in all in terms of sort of saving you your time, this is where, maybe we can start to, generate, a memo for you. So I put my manual text, which I can put in there. Let's ask it a question, though. I'm gonna jump into my assistant and start to draft a new memo. I'm gonna ask it to analyze the role of the Raptor special per special purpose vehicle in fraudulent accounting practices. I can hit generate on that. That's gonna start generating in real time in the background, and you'll see that start coming in in just a minute. And this isn't designed to be the final piece of work, Jess. This is kind of a first draft. This is to get you away from that blank page, but it's a good place to start if you are, kind of involved in that discussion. So as soon as that generates, it will allow me to insert it. I don't have to pay for it until I hit insert. Give that a few more seconds. I can hit insert, and that's added that to my document. Just wanna show one final thing, for maybe for our sort of international audience here as well. We could actually also ask it to, draft this in a different language. So I'm gonna ask it to draft it in German. I'm gonna hit generate. You'll see this coming in in the background. I'm gonna pause there, and and sort of hand over to you and Ralf, Jess. Alright. Thank you, Greg. Okay. Seems like we have all the things that we need, in preparation for this meeting tomorrow. And now one last poll for the audience. You've seen a couple of tools, at this point. There's no wrong answers here. We're we're just curious. Yeah. So based on what you've seen, you know, which Gen AI feature that you've seen do you think would help, you the most the night before a high stakes meeting like the situation that we set up? Would you think project query is the most the writing assistant that we've just seen is the most useful, the draft summaries and as as in, you know, it goes through your documents and and provides, summaries of, of all your documents, and then the coding suggestions as well that Greg showed earlier where you can batch code, documents, and then it will give you whether it's relevant or not. Alright. Let's see the votes coming in. And it looks like for most people, they anticipate draft summaries, to be to be critical, in terms of, in advance of the high stakes meeting. That makes sense to me. Alright. Okay. So, Ralf, you know, we've gone through all the things, all the steps that we need to do on this platform. You've been here to to make sure that, you know, that the law firm doesn't, doesn't drop any, drop any balls at this point, which is very, very common, very possible during these, kind of these high stakes, investigations and matters. Now that, you know, we've uncovered some information and we're kinda ready for the next steps, is there any is there anything that you would advise a law firm on at this point and anything that, you know, a law firm typically tends to to miss at this point as well? Yes. And and thanks for showing all this and and the ease of use and time saving tools in in Avalon. Just wanna remark here. So you as the flustered lawyer would probably not be willing to find all those easy clicking buttons. It's very easy to become a specialist in Avalon because it's ease of use. It's everything there. But you don't wanna do this, shortly before your high stakes meeting. That's where we come in. We are the specialist on demand. And what we then, usually suggest, law firms, when they started using Everlaw in a case like this, is continue using Everlaw and getting prepared for the next ones, but we always dare to support. What comes usually next after that? Yeah. We we are in this point. We we have the the submission done. Now it comes to to additional reviews that we see. It comes to and this is usually where we support, production assistance, loading of additional data. Sometimes we get, data of of opposing councils. Sometimes that's just to to drown the the parties with unnecessary documents and data. It's a lot. Again, that's something where we can again push it into Everlaw and use the capabilities in Everlaw to to do the other side of the game. So once we've done the submission, that's basically before the next one. And then, finally, there is is considerations on, how do we how do we handle the time after, Everlaw, hopefully. And not from Everlaw perspective, I guess, but from from clients and customers' perspective, there will be a time after that case. So case will be settled, closed, data needs to be gone. So how do we handle this? That's something where we as well can help and and support on discussions, how we do disposition. And then it's it's basically us being continuously there on on communications, on the case, and helping on all of these. Alright. Thank you, Ralf. Absolutely. All really important considerations, in a lifetime of a matter. So just to wrap up, what did we figure out today? Well, we identified key terms that were central to the practices of our client, Imran. From from the first step, we uncovered mentions of Raptor and the LJM partnerships, that were really crucial in this case. We also looked at key communications between key players who would again be central to the matter that we've identified. And also, we were able to see, you know, which conversations were irregular, and that need to be investigated. We're also able to quickly identify, surface, and produce documents that needed to be discussed essentially the next day. And we can be confident that they're relevant because the tool provides us with citations and reasonings as well that we can review. I will say that we already seen clients, you know, use these tools on LiveMatters. On our website, we actually have a success story that's based in The UK with a partner gender and in The US as well as also a a partner based, case study that's anonymized that shows you how this these tools, can be used in practice. We're a bit out of time for q and a. But, obviously, you can reach out to Ralf and myself and Greg for any questions questions around Everlaw and Sandline and how we can help you in your next matter. I do want to plug that we do have a joint event, in June on the nineteenth. It is titled the on prem era is ending, what comes next. And, so, yeah, we'd like to invite you to join us for a discussion about the tech transformation that's been happening in industry. You know, you hear from people that are are in it, that have experienced it or are, you know, planning to go into that journey. So you can see the ticker below. Click on that link, and you'll be able to register yourself for the event. And with that being said, thank you everyone for your attendance and your attention today. Hopefully, that was an interesting look or keyhole look, as Ralf said, into into a kind of day in the life of, the work that we do. And, we'll see you at the next event, and, get in touch if you have any questions or if you have any requests. Alright? Thank you, everybody, and have a great rest of your day. Thanks, Paul. Thank you. Bye.