AJ Shankar's Everlaw Summit '25 Keynote

Please welcome to the stage Everlaw founder and CEO [music] AJ Shunker.

>> Welcome everybody to Everlaw Summit 2025. Hacked House. This is incredible.

2025. Hacked House. This is incredible.

Thank you so much for taking the time to attend. We really treasure these

attend. We really treasure these interactions. We always take away so

interactions. We always take away so much from our conversations with you that we're probably learning more than you are this week. And if we're doing our jobs well, you're learning a lot.

And I'll say presenting this keynote is one of my favorite activities every year. I get to review everything we've

year. I get to review everything we've learned in the last 12 months and give you a peek into what's coming next. And

I hope I do so in a way that's relevant to you regardless of if you're over in one of our over 1300 customers spanning law firms and corporate in-house teams and public sector organizations or whether you haven't yet started your

journey with Everlaw. Now I've spent the last few years using the stage to talk about AI and guess what? I'm going to talk about it more this year. I can

understand if there's some fatigue.

AI is everywhere, right? But there's a good reason for that. The technology

continues to advance a pace with new capabilities unlocked every year. It's

transforming the practice of law faster than any prior technology. And so, it's important to know what AI enables. But

we know that each of your journeys in AI is different. Some of you are skeptics

is different. Some of you are skeptics and others of you are passionate about AI. And beyond your personal feelings,

AI. And beyond your personal feelings, you have different guidance from your employers or sometimes no guidance at all. Right? You have different risk

all. Right? You have different risk tolerances, different obligations to your clients or internal stakeholders.

Some of them say you have to use AI.

Some of them say you can't use AI. Some

don't care. Others need to be notified.

And still others require explicit authorization. So I don't envy you. It

authorization. So I don't envy you. It

is hard to navigate this level of change. I just wanted to say though that

change. I just wanted to say though that we are with you on this journey. The

reason I talk so much about AI is not to shill it or to coers you to use it. It's

to educate you so that you can make an informed decision about what's best for you and your organization. Our

commitment is to ensure that you always have access to the best technology no matter no matter what it may be, whether it's AI or not or big or small, and that you're fully trained and enabled on this tech. And beyond that, it's up to you

tech. And beyond that, it's up to you how you use it. So, how do we bring you the best technology? So, as a reminder, all of our code is developed by us.

We've never offshored our engineering and we've never acquired another company. Every experience you see here

company. Every experience you see here is built by our R&D team of nearly 200 folks who are just across the bay in Oakland. And many of them are here at

Oakland. And many of them are here at Summit. So, please say hi to them if you

Summit. So, please say hi to them if you see them. These folks spend time

see them. These folks spend time thinking deeply about how new technologies like Genai are going to impact the core workflows and goals of litigation and investigation. We're not

adding AI functionality opportunistically to capture new cycles or do big fundraises. We're not changing our name to Everlaw.AI. We want to solve real world problems and we put each of

our AI features through rigorous testing before release. And so I can confidently

before release. And so I can confidently say that we're adding new AI features because they work. So even today there are so many AI assisted experiences

available on Everlaw, each embedded in existing day-to-day workflows. And we

see that trend only strengthening over time. Eventually ignoring AI is going to

time. Eventually ignoring AI is going to kind of feel like ignoring search. It'll

be an essential technology. So for this talk I want to show you how uh AI is being integrated into the Everlaw platform. I want you to hear the real

platform. I want you to hear the real world experiences of folks who have used these tools and I want you to start rethinking your own workflows to incorporate AI tools where they make sense for your own journey. So before we

dive into all the great functionality, let's do a brief review of how we develop AI responsibly. So first we always protect your data. In addition to our standard rigorous vendor vetting procedures, we ensure that all data that

goes to our model providers is never used to train their models and is never retained after the query has been answered. And second, we minimize the

answered. And second, we minimize the risk of hallucinations in several ways.

Our AI experiences do not use large language models for their built-in knowledge or the particulars of the law, which is a huge source of hallucinations in the legal field. Instead, we focus on the four corners of your documents,

providing the AI with ground truth at the time we ask it a question. We also

integrate our AI experiences into your existing workflows. By defining precise

existing workflows. By defining precise use cases, we can construct prompts for you, which means that using our AI is often as simple as clicking a button. We

can also test more robustly since we're not trying to solve every problem in one interface. And we can provide all the

interface. And we can provide all the necessary context, your documents and your work product so that the large language model has all the information it needs. And in combination, these

it needs. And in combination, these design decisions increase reliability substantially. Furthermore, we

substantially. Furthermore, we understand that any work product is ultimately yours to own. So we make it easy for you to validate the AI's output, whether by providing bait

numbered citations or direct quotations from the source material. And finally,

we take our time with these releases. We

are much more interested in being best than being first. We typically beta test our AI features, gather gathering thousands of items of feedback from our users over many months and iterating on the experience before putting them in

the wild. So in combination, we think

the wild. So in combination, we think this approach balances prudence with pragmatism. We want you to have access

pragmatism. We want you to have access to all the best AI functionality, but we also want you to use these tools with confidence. So with that in mind, let's

confidence. So with that in mind, let's go over the Everlaw AI experience today.

When reviewing individual documents in our review window, you can translate foreign languages into English.

You can generate document summaries. You

can create a topic based outline or do sentiment analysis. You can ask a

sentiment analysis. You can ask a question of the document. You can

extract relevant information such as medical record numbers or license plates or PII or anything else relevant to your case. You can have the AI suggest which

case. You can have the AI suggest which codes to apply, providing a reason and supporting that reasoning with citations back to the document itself. And you can

all do all of this across thousands of documents if you prefer. You can run batch translations to convert foreign language documents to searchable English for a fraction of the cost of using an outside firm. You can use batch

outside firm. You can use batch extractions to pull critical information from your entire case. You can use coding suggestions to do everything from jump starting a predictive coding training set to doing an entire first

pass review. And then you can take all

pass review. And then you can take all the hot documents you found and put them to work in story builder. Everlaw AI

will automatically generate summaries of your most important documents. The

writing assistant can generate outlines or even full memos on a wide variety of topics, all grounded by your data. And

the Depo summarizer can distill a critical findings from an hourslong deposition within minutes of receiving the transcript. As you can see, this

the transcript. As you can see, this isn't a single chatbot slapped onto our existing product. It's a deeply

existing product. It's a deeply thoughtout approach that allows you to leverage AI uh as and when you need it.

By the way, we'll soon be adding more functionality as well. Our upcoming

depositions QA tool will allow you to deep ask deep questions of all of your depositions and get synthesized comprehensive answers that site back to source testimony. And our upcoming

source testimony. And our upcoming privilege descriptions tool will generate grounded explanations of why each docu document is marked privileged.

Now, before I start digging into how these tools are changing legal work, let's talk about value and pricing, something that's on everybody's mind.

If you're a longtime Everlaw user, you know that we try extremely hard to pack huge value into our core platform per gigabyte charge. When we first started

gigabyte charge. When we first started the company, we introduced a single per gigabyte fee that changed how the industry charged for eiscocovery.

There's no more ingestion fees. There

were no more user fees or more production fees or OCR fees. We know how much you all appreciate simplicity and we're gratified to know that the industry has moved in this direction.

And we stuck to our guns on this even as we introduced cutting edge technologies over time such as predictive coding, clustering, data visualization, story builder, native redactions, depositions, cloud connectors, audio video

transcription. We didn't charge for any

transcription. We didn't charge for any of those. We found a way to include them

of those. We found a way to include them in the single per gigabyte fee. Over the

years, our users have derived more value from Everlaw for the same price every single release, 12 times a year.

Now, as you also may know, we charge for our Genai experiences through a credit system. Now given our philosophy and

system. Now given our philosophy and history, this is not something we wanted to do. So you can bet that we had a lot

to do. So you can bet that we had a lot of discussions internally before releasing Everlaw AI about whether we could possibly include it in our per gigabyte cost without raising the price.

But there was just one inescapable fact that the state-of-the-art foundation models we use are provided by third party companies who charge us for every single usage. Now when AI usage is bound

single usage. Now when AI usage is bound by individual user behavior, which is really common for most software products, you can advertise those costs in, right? Maybe a novice user uses AI

in, right? Maybe a novice user uses AI five times a month and a prouser uses it a 100 times a month, maybe 20 times as much. You can average those out and find

much. You can average those out and find an all-inclusive cost model that works.

Many SAS companies do this, but our industry is different, right? You can

run AI on five documents or 5 million documents, a million times as much.

There's just no way to make that math work with a single cost. And so at the same time, we know how hard it is for you to operationalize the use of these really powerful tools with a system

where every usage is metered. So we've

been spending a lot of time in the last year on how we can make the experience better for you and how we can give your teams more of the value we've built with Everlaw AI without charging you extra.

So I'm happy to announce that we're including writing assistant and single document AI assistant tasks with your core per gigabyte rate. So we're not forcing you. Yes, this is really

forcing you. Yes, this is really exciting.

>> [applause] >> really excited to announce this. We're

not forcing you to pay more with a new standard gigabyte rate. Our rates are staying the same and all of our customers will have access to these included AI features, even if you aren't currently paying for Everlaw AI. Of

course, you can still choose whether to use them or not. Now, the price change will actually take place next week. Yes,

next week in our October release, this is all going to go live. Really excited

about that. Um, yes.

We have spent a lot of time planning and we're excited to execute, right? So,

with all these tools available, we think it's time you start reimagining how your team works. With every non-batch AI

team works. With every non-batch AI feature now included in your Everlaw subscription, right? Your reviewers can

subscription, right? Your reviewers can summarize or ask a question of a long complex document to ensure they're not missing anything or make a better coding determination or do topic analysis and

extractions. Your case team can add

extractions. Your case team can add documents to Storybuilder and have them automatically summarized. Also included

automatically summarized. Also included is every AI writing assistant feature in Storybuilder, including memo writing, outline creation, building lists with relevant citations, or creating tables to represent key facts or entities or

whatever you'd like to specify. You can

also create new or rewrite existing content. You can use our predefined

content. You can use our predefined templates for factual, misconduct, thematic, or character and entity analysis, or make your own templates.

writing assistant can argue your side in the matter with key evidence that you have or opposing council's side if you want to prepare. You can analyze depositions extracting summaries or finding key information. You'll soon be able to ask questions across all of your

depositions in a matter. The [snorts]

bottom line is we think there is remarkable functionality in a review assistant and in story builder with writing assistant. We want you to try it

writing assistant. We want you to try it and if it delivers, start using it on every every case at no extra charge. So

taken all together, I know this can feel like a lot, right? But I want to emphasize that these tools are designed to be used right in your existing workflows. You don't have to learn a new

workflows. You don't have to learn a new product. You don't have to memorize

product. You don't have to memorize everything because these tools are now included in your Everlause subscription.

You just have to be willing to look for that AI icon and click on it to try these tools out and see whether they're helping you get your work done better and faster. Now, this isn't a onetime

and faster. Now, this isn't a onetime promotion until the end of the year. It

is real and it is long term, right? This

is consistent with our belief that as your finest technology partner, our job is to enable you to use the best cutting edge technology out there. Now, for

large batch actions like batch coding suggestions, batch translations and extractions and summaries and our upcoming deep dive feature. Anything

operating on many documents at once, we are still going to charge for those using credits. Trust me, we did try on

using credits. Trust me, we did try on this one. We evaluated raising

this one. We evaluated raising everyone's per gigabyte fee to include them, but that inclusion would have still required caps limitations. So, we

optimized for giving you control and value. You should be getting value out

value. You should be getting value out of every dollar you pay us. We will not ask you to subsidize AI usage. So, we

give you the choice to apply these tools when you want them on the cases where they matter the most and only pay when you use them. So, we think we've hit a solid middle ground here and we're going to continue to work with all of you as we always have as the technology and

costs evolve. This allows your team to

costs evolve. This allows your team to make the most of Everlaw AI during their daily work at no extra charge and also still allows you to leverage powerful paid batch tools to accelerate your

entire review as needed. Now, we know these batch tools are really valuable, so we're still focused on making them accessible to you. So, I'm excited to announce that for our most popular batch action coding suggestions, we're

reducing our price by over 40%.

Yes, we want to make this accessible.

So, we're going to continue to work hard to ensure that we're making our batch AI tools affordable even as they evolve. U

we also heard that purchasing AI separately from the core product was difficult. Right. Today we're announcing

difficult. Right. Today we're announcing that uh you can purchase through unified contracts where with one contract you get access to staging drive to ECA active and suspend and AI credits for

use and batch actions and deep dive. You

set the commitment and you control the spend all through one unified contract.

So that's a headache for someone in your organization. We want to make that less

organization. We want to make that less of a headache for them. All right. So I

know that was a lot. So let me summarize. Right? So review assistant on

summarize. Right? So review assistant on single documents. So that's

single documents. So that's translations, coding suggestions, summaries extraction sentiment analysis, and Q&A. And all writing assistant features, lists, memos, outlines, rewrites, depo analyses are

now included in the core gigabyte rate.

These charges, these changes will start with our release later this month with no per gigabyte price increase. Batch

actions will still be paid, but batch coding suggestions will be 40% less than before. Also starting with the release

before. Also starting with the release at the end of the month. So I just wanted to get all that out of the way up front. We want you to get excited about

front. We want you to get excited about these features. We think they're going

these features. We think they're going to elevate your work and help you do your job better. And of course, we'll keep shipping other innovative functionality every month. So, let's

have some of our customers who are using these tools up to speak about their experience. I'd like to invite up Steve

experience. I'd like to invite up Steve Delaney, director of litigation support at Benish.

[music] [music] Much Steve, thank you for being here and spending time with us today.

Um, so why don't we start by having you tell us a little bit about Benish and your role?

>> Well, uh, Benish is an AM law 200 law firm. Uh, we actually just hit the 500

firm. Uh, we actually just hit the 500 attorney mark.

>> Congrats.

>> Uh, founded in Cleveland, Ohio.

[applause] So, I oversee the firm's eiscocovery, analytics, and document review capabilities, and I manage the litigation support group. So I advise

clients and litigation teams on uh evidence management, ESI protocols, uh best practices, trial technology, and eiscocovery workflow.

>> Good person to have up on the stage. So

uh now I know you and Vanish have really embraced coding suggestions. Could you

tell me a bit about how you're using them?

>> Well, we've been using coding suggestions in a variety of ways. We we

started using it for first pass review for you know client documents uh for production but then we moved on to reviewing opposing productions for

issues and trying to find key documents and uh finally we've just started trying it out for privilege review.

>> So I understand you built a really rigorous process for applying coding suggestions at the firm. Do you mind telling us a little more about how that works?

>> Yeah, sure. Sure. So the key to success with coding suggestions is building a robust process for drafting and iterating your prompts. So you're you're coming up with with revisions and

improvements to your prompts as you go along. So first what we do is we we

along. So first what we do is we we identify key issues in the matter and build prompts for each issue. Then we

build small targeted samples for each issue. uh we're choosing these samples

issue. uh we're choosing these samples in order to uh uh be good exemplar documents for each issue that it should respond positively for. Initially maybe

five to 10 documents per issue very small small sample at first. Then we run coding suggestions against that sample and review it. So the attorneys are looking for all the areas where they need to revise these prompts to get the

results they're expecting on these samples. And this is where th those

samples. And this is where th those gradients uh of yes, soft yes, no, soft no gets really important because not all issues are created equal. Uh some issues

are uh more sensitive where a hard soft yes is even enough to consider it responsive and and others are are a little more more uh less sensitive. So

then you revise your prompts accordingly. You run suggestions again.

accordingly. You run suggestions again.

You re-review that sample until you are comfortable that that sample is 100%.

And then you move on to a slightly larger sample. And you keep reiterating

larger sample. And you keep reiterating this over and over until you have the the confidence that uh that when you run it across the full set, you're going to have a great result.

>> That's great. And so I'm so excited about the rigor you put through uh this process here. How do you keep track of

process here. How do you keep track of the iteration and the prompt evolution to make your work defensible? We do all of our tracking within Everlaw itself in story builder in the drafts section. So

we use it to to track all of our iterations in the and all of our validation process. We keep tables in

validation process. We keep tables in there of every step and justifying all the reasons why we chose that step. And

we also use it to um as our collaborative prompt revision editor. We

have all of our prompts in in one story builder draft. And because the attorneys

builder draft. And because the attorneys can all collaborate in there at the same time, you can have different attorneys working on different uh uh prompts and revising them in real time. Uh we also

take advantage of the history capability in there that you could you can go back to uh two, three, four versions ago of your uh prompt revisions because you might find as you're testing them that

your latest change made things worse and not better and you want to go back to a previous version. So we find drafts is

previous version. So we find drafts is very helpful and it's helpful to have all of that in the one platform.

>> And so that's fantastic news. What about

an area that hasn't worked so well?

>> Well, as with any review process, right, the key to success is making sure that the attorneys working on the review are fully engaged in the process and and committed to doing all the steps you put

in front of them. they take shortcuts in reviewing samples or if they, you know, really aren't looking at each document in depth and just kind of skimming through, uh, you're not going to get as good results. So, you're you're kind of

good results. So, you're you're kind of wasting everyone's time if you don't get that commitment to engagement right off the bat. Um, so I try to equate the work

the bat. Um, so I try to equate the work that and when I'm trying to promote this to the attorneys to get them to to to buy into that process, it's just like for linear manual review, right? You've

got to get the contract, the key attorneys have to get the contract attorneys up to speed. They've got to draft a protocol for them. They've got

to review the work of those contract attorneys, especially early in the process and uh and re-educate as necessary. You know, are they misreading

necessary. You know, are they misreading or misinterpreting the protocol? Uh so

we need to catch those mistakes and and correct them. So it's the same with

correct them. So it's the same with prompt iteration. Uh but the two major

prompt iteration. Uh but the two major advantages of the coding suggestions over the contract attorneys is number one, you have two attorneys. You're

going to have two different opinions on things. And imagine 50 contract

things. And imagine 50 contract attorneys and they're going to interpret everything differently about your protocol. You're not going to get that

protocol. You're not going to get that with with the AI uh coding suggestions is going to be consistent across your entire document set. Uh and then the

biggest difference of course is speed.

um you can get so much faster to a to a completed first level review with coding suggestions. It's no comparison.

suggestions. It's no comparison.

>> So given that you've actually built this process yourself, what are any takeaways you have for folks?

>> Well, the biggest takeaway is that if you haven't started using coding suggestions yet, like do it, start find a way to to

to get yourself using it. Um I think the the lowest risk lowest bar to entry is for incoming production. You you you get an incoming production from the other side. You want to understand what the

side. You want to understand what the issues are. You want to get to those key

issues are. You want to get to those key documents. Um that's where you can you

documents. Um that's where you can you can kind of get your feet wet with it and you don't have to worry about being challenged by the other side. So, um,

and one great benefit is is once you've gone through that work and you've got your prompts in great shape, a new production comes in, you can, uh, you can apply it right away and get instant

results. Um, we've had cases where we

results. Um, we've had cases where we had a, uh, deposition the next day and we get a production just before midnight and then of course they want to waste all the attorney's time instead of

preparing for the deposition to be slogging through all these documents and we kind of shortcircuit their strategy on that. So by using coding suggestions

on that. So by using coding suggestions has worked out really well. So I I think that the my last takeaway though is that this is the key time to be taking

advantage of this. Not everyone is using it right now. So if you want to get competitive advantage on something, you don't get competitive advantage by doing what everyone else is doing. The best

you can get then is par. So right now you can actually get ahead of your opposing side by by using this at this time.

>> That's incredible. Thank you so much, Steve. Appreciate it. Thanks for your

Steve. Appreciate it. Thanks for your time.

>> All right, great stories there. Next, I

want to bring out Ed Valio from Geico to talk about a really interesting use case that moves between our new generative AI features and our more traditional AI features. Ed, you there?

features. Ed, you there?

Here we go. Come on out. You're the next contestant. Good to see you.

contestant. Good to see you.

>> Thanks for being with us.

>> Tell me a little bit about Geico and your role.

>> Well, as you may know, Geico is one of the largest autoinsurers in the US and I lead the eiscocovery and records management team in the legal department.

>> So, I understand you had a really tricky situation where you used Everlaw in a pretty unique way. Can you tell us a little bit about that?

>> Yeah, we we had a situation where we were presented with an unusual question and in order to answer it, we had to evaluate tens of thousands of contracts in order to answer it. Um, and we had a

very short time to do it. We needed to understand some very specific information about those contracts. Um,

we're a rapidly modernizing legal department and our leadership is making significant investments in technology, but our contracts team didn't have an appropriate solution in the moment and the universe of contracts and legacy

systems they had to work with had not been well organized to provide an answer to this type of question.

>> So, well, what did you do? How did you tackle this situation? Uh we first collected all the contracts from their respective repositories and uploaded them into an Everlaw project. Uh we used custom extractions to uh get the key

parties for each of the contracts and put them into a metadata field. Um we

then developed what would be the high value search terms that could help us determine where we might have relevant information in these contracts. Um next

we used the review assistant to evaluate the documents that hit on those search terms and provide coding suggestions for relevance. Um, using the output of the

relevance. Um, using the output of the review assistant, we created a predictive coding model to then score the full 50,000 contract set to help find any contracts that might also fit the profile that we were looking for,

but didn't contain any of the highv value search terms. We provided the highly scored documents as well as a sample of the entire set as the output to our colleagues.

>> And so what happened with that that process? What did you find? Um, the

process? What did you find? Um, the

great news is we were able to turn around this entire project in 48 hours and provide a definitive answer to the business. Just one document out of the

business. Just one document out of the 50,000 contracts had the exact characteristics that we were looking for. That allowed us to understand the

for. That allowed us to understand the situation very quickly and clearly across that whole set.

>> Great. That is first of all say that's incredible. 40 hours to find one

incredible. 40 hours to find one document. It's really amazing and such

document. It's really amazing and such creative work when using the tool so effectively. So having done that, what

effectively. So having done that, what did you do next? Well, we wanted to get more context around that specific contract and we already had the predictive coding model. So, we

identified the key individuals involved with that contract and pulled in email traffic from the relevant time period to gather the context around that specific contract. So, having the full endto-end

contract. So, having the full endto-end workflow in Everlaw with AI in the right places at the right time uh is really what made this process so simple to execute and allowed us to move so

quickly. So what advice do you have for

quickly. So what advice do you have for folks about applying Everlaw in interesting cases like this? Um from

this one key takeaway, you know, don't keep yourself in a silo. The tools and the skills that we as eiscocovery people use every day aren't just valuable for litigation and investigations. They can

solve high value business problems across an organization. When you have colleagues who are facing a problem that's document heavy, it's an opportunity to help. You know, 50,000 documents might feel overwhelming to

some people, but it's routine for us. We

we've built workflows that handle, you know, hundred times that many. Um, so we can help accelerate their projects, reduce risk, and improve decision-m. So

look for those moments and offer your expertise because the more we apply these capabilities outside eiscocovery, the more strategic and indispensable we can become.

>> Thank you so much, Ed. Appreciate your

time.

>> I love hearing these stories. These are

really smart people applying technology in really creative ways. And I think as just as Ed said, the more use cases you can find for the tooling and your expertise you've built, the more valuable you can be to your

organization. So look, I will say these

organization. So look, I will say these are really really powerful tools and and these folks give great examples of that and they can solve major problems, but it does take some work and it does take

a little creativity to put them to use.

And so our point of view is that your team is going to benefit from your guidance, right? your authorization to

guidance, right? your authorization to use these tools, your policies and training on how to use the tools, and the redefinition of your workflows to show your team when to use these tools,

just as Ed and Steve have done. Um,

ultimately, I think these changes are going to require your organization's commitment. And because this is such a

commitment. And because this is such a significant undertaking, we want to help you behind beyond just providing the technology and we have many resources and programs to facilitate navigating this change. So, first, if you're not

this change. So, first, if you're not interested in using Everlaw AI, no problem. But if you are interested and

problem. But if you are interested and are struggling with any obstacles right whether economics functionality firm policy or guidance client approvals team training and enablement or anything else

we want to hear from you and we're committed to helping in any way possible. Now our value AI team is a

possible. Now our value AI team is a group of highly experienced legal professionals that have worked at some of the largest law firms. They have deep discovery and AI experience and have helped many of you use Everlaw AI and

they're here at Summit to hear about your challenges. So, in the event app,

your challenges. So, in the event app, you can sign up for a private 30-minute consult with our value AI team about using Everlaw AI in your organization.

We'll learn about what you're facing and help if we can. And if not, we'll make a note of your challenges, brainstorm about how we can help with them, and then get back to you in a few weeks.

Now, if you don't have time to meet with the value AI team at Summit, please reach out to your customer success manager to arrange a time to talk with them. We really want to contribute to

them. We really want to contribute to your journey here. All right, let's talk about another big update. Last year on this stage, I demoed a very early version of Deep Dive. This is our tool that enables you to ask questions of

your entire corpus, your project of millions of documents, asking them a question of all those documents at one time and that can provide value at every step of a litigation. So shortly after that, we launched our beta program where

we had many of you using deep dive day in and day out, meeting with our product team and providing insightful feedback.

I'd like to thank all of our beta testers. I will clap for them. Thank you

testers. I will clap for them. Thank you

so much folks that spend their time helping us develop these projects. You

ask thousands of questions. The average

database size in the beta was 166,000 documents and our largest had over 10 million documents. We got overwhelmingly

million documents. We got overwhelmingly positive feedback on the beta and you can see a few quotes here. Um my core takeaways from the beta about deep dive are twofold. One, it's really easy to

are twofold. One, it's really easy to use. Even senior partners will want to

use. Even senior partners will want to use it and two it quickly finds needles and hay stacks with link to the links to the key documents which makes it useful in many many parts of a litigation or

investigation process. So, [snorts] I'm

investigation process. So, [snorts] I'm excited to announce that we're planning the general availability release of Deep Dive in December. We think you'll find it's been worth the wait. Now, yes, I'm

really very excited about this one. Um,

and what I really want to what I want to do now is invite up one of our beta testers to come and share some of her experiences with Deep Dive so you can understand how it's used. So, please

welcome Julie Brown, director of PLA practice management at Vores.

Hey, so good to have you up here, Julie.

Thank you for being here and sharing your expertise with us. Could you please tell us a little bit about Vores? Sure.

And your role.

>> Thanks, AJ. Um, Vores is an AmLaw 200 firm with 375 attorneys with offices across the United States as well as one in the UK and one in Germany. Um,

myself, I've been in the legal technology business for over 30 years, um, in both law firms and an in-house litigation department. Um, I currently

litigation department. Um, I currently lead a team of highly highly skilled legal technologists, which I'm very fortunate to do. Some of which are in the audience today. So,

>> that's so great to hear. Uh, I heard this is a surprise for them. So, it's

great. Um, so I really like to thank you for being one of our deep dive beta test organizations. It's never easy to get

organizations. It's never easy to get legal professionals to use technology.

So, tell tell us like how how did things go with the deep dive beta?

>> Sure. Um, so remarkably easy to be honest. um because it's so intuitive and

honest. um because it's so intuitive and user friendly um it really wasn't a heavy lift. Um it was really getting it

heavy lift. Um it was really getting it in front of the people um in front of the attorneys. And so we started out by

the attorneys. And so we started out by just asking the attorney to create a description of the case um and then draft 10 or 20 questions that you'd like to ask your data set. Once we got that

information, we were able to enroll the case in the deep dive beta and we would do the demonstration for them, the onboarding using the questions they had.

So, um the attorneys were just in awe when they saw the results and um a good example is a case we our first case we

did um was about 300,000 documents. So,

pretty sizable case. The good thing was the attorneys were pretty far along in the case. So they they had a lot of

the case. So they they had a lot of knowledge about what the documents contained. They were actually in depo

contained. They were actually in depo prep. Um so we tested that was our our

prep. Um so we tested that was our our very first experience with deep dive and the answers they were getting back were like spoton. Um so they were really

like spoton. Um so they were really really impressed that they were able to get those type of responses. Um they

also identified some new documents that maybe they weren't aware of. So it did provide some additional nuggets of information that they could use. um

based on just that one test, the words spread. I mean, so if attorneys like

spread. I mean, so if attorneys like something, if they don't like something, they're going to tell everybody, right?

But if they like something, they're going to go tell all their friends. So,

ever since then, we really haven't had a challenge getting additional cases. In

fact, most of the time, attorneys are coming to us and saying, "Can you add my case to Deep Dive?"

>> That's fantastic. So, where have you and the team used Deep Dive over the case life cycle?

>> Sure. So probably three key areas I'll highlight. One is investigations. The

highlight. One is investigations. The

other one is quality control. And then

the third I would say is depo and trial prep. So for investigations, a lot of

prep. So for investigations, a lot of times you're just trying to figure out who are the key people, what are the events, what are the issues, what's going on. And so you know you might ask

going on. And so you know you might ask deep dive who are the key people who have knowledge about so and so's non-compete agreement. Um, and it'll

non-compete agreement. Um, and it'll come back and give you just a really detailed list of not only the people, but then explain how they relate to it and how they tie to it. Um, so it's

really a great investigative tool. Um,

for QC, we had a case come in with two million documents. I might have some

million documents. I might have some people cringe in the audience that are here. Um, and we had a production

here. Um, and we had a production deadline of just a week literally um, to get the documents in, reviewed, and and um, produced. So initially we thought

um, produced. So initially we thought about deep dive but we were worried about defensibility. So we were like

about defensibility. So we were like okay let's let's start with some search strings then we're going to move to um predictive coding so and really pump

through pump this through quickly so we could get the production out on time. Um

but then what we did is we use deep dive as kind of the QC tool and say what did we miss right because we were able to ask questions and say are there any documents not in the productions that that relate to X Y or Z. So um that was

another use of it. And finally in deposition and trial prep, it's really really helpful to identify key documents. Um you can ask it what a

documents. Um you can ask it what a particular deponent knows about a topic and it's going to go through all the documents and come back and say well here are the things he knows about and here's why he knows and here's how it

worked. Um you can also ask if there's

worked. Um you can also ask if there's discrepancies within the documents. So

you can say can you and you can do this across the entire database or sets of documents. you can actually target

documents. you can actually target certain sets, but you can ask if there's any discrepancies. You can ask if there

any discrepancies. You can ask if there is there anybody else who's who really knows about this topic. Um, so that's another good question. So ultimately I I

really hope someday this changes how we do document review and we don't go page by page anymore. We actually go in and say give me the most relevant documents and that's how we do our job.

>> That'd be amazing. And so one observation you had is just how people interact with the system and refine their questions and almost have a conversation with it. So maybe if you could tell us a little bit more about how folks are using deep dive and and

iterating with their questions.

>> Sure. I I think the biggest thing and what's important is attorneys are curious and they ask a lot of questions which is good, right? So when they go in they can approach the case with from different angles and ask different types

of questions. You can ask a very broad

of questions. You can ask a very broad question like why did company X end their relationship with company Y. Um

you could also ask it very detailed questions. Um we had a case where we

questions. Um we had a case where we were looking to determine if somebody um either gained or or lost weight over time in medical records. So we were able to look at that. Um the first time we

wrote the prompt to do that analysis, we missed the mark. It couldn't find anything. So we went back, we tweaked

anything. So we went back, we tweaked it, um and ultimately we got the answer we were looking for. Um, so I think, you know, another key thing I see is when an

attorney um sees a new person, like who is this person? Why is this person showing up in my database? They can just say who is this person? Um, and it comes back with a complete description of who

the person is, how they're involved, who they talk to. I mean, it's it's truly amazing. So, um, I think it's just been

amazing. So, um, I think it's just been a huge success for us.

>> That is fantastic to hear. So, let me ask the obvious follow-up question, right? What's the best way for teams to

right? What's the best way for teams to get started with deep dive?

>> Sure. Um, first of all, use it. If you

haven't used it, I highly encourage you to try it. Um, you can try it on a small case. Um, and especially where attorneys

case. Um, and especially where attorneys already know the material because that's going to give them confidence that it's actually doing what they want it to do.

Um, just ask them, can you give me a description and 10 key questions that you're interested in? Um, I usually tell them when I do the demo, I'm always like, "Are you ready to see some magic?"

And they kind of look at me and I'm like, "Just wait." Um, and by the time we're done with the demo, you can just see the look on their face. And it's

just, you can just wow them. It's

amazing.

>> That is so fantastic. So, that's all the good stuff. Anything we should still be

good stuff. Anything we should still be building into deep dive as it approaches general availability.

>> Yeah. So, one thing that would be helpful right now, um, when we find key documents, we have to click a button. I

know it's doesn't shouldn't be that big of a deal to um actually go tag the document or send it to story builder.

So, it'd be really cool if we could do that right from the the deep dive interface.

>> I agree. Yes. And we're building that right now. So, yeah. Exciting. Thank you

right now. So, yeah. Exciting. Thank you

so much, Julie. Really appreciate the time and expertise.

All right. So, deep dive is going to be a batch feature and it's going to have an additional cost, but pricing is going to be straightforward. There's gonna be a onetime per gigabyte ingestion fee for each of your documents and that's going

to give you access to deep dive for the lifetime of the case with unlimited questions.

All right, so I do want to add one more thing here. Um,

thing here. Um, you might ask where are we going with all of this, right? We think soon enough using these tools is going to become second nature to you. Your workflows are going to incorporate them. They're going

to improve the quality and speed of your work. You know, I keep using this word

work. You know, I keep using this word workflow. We speak with so many of you

workflow. We speak with so many of you and we know how critical and complex your workflows can be, whether they can benefit from AI or not. Why are

workflows so important? Well, in some industries, it doesn't matter how you got a result. The ends justify the means, but not our industry, right? Our

results need to be defensible. Our

processes need to be repeatable. You

have obligations to your clients, to the opposing party, to the court, and ethical obligations to the profession.

We need to ensure that when you do work, it stands up to inspection. And this is especially relevant in a future where AI agents are running rampant. Right? Sure,

yes, an AI agent can write a blog post for you, but you don't want an agent deciding how to run a discovery process.

It's likely to do something different every time you ask it. And on any given time, it might take the wrong steps, right? So, we think AI has a very

right? So, we think AI has a very important role to play in the future eiscocovery, but it needs to be harnessed and orchestrated by humans.

And we know that you're already running complex workloads today, often in spreadsheets or other ad hoc mechanisms. So we do think your discovery platform should help you manage those things, right? And so what if we could address

right? And so what if we could address both problems with one solution? So we

wanted to help with workflows for a long time since at least 2020, but back then we didn't feel like the time was right.

We're now creating a workflow builder to help you specify and execute your workflows. This is not an AI tool, but

workflows. This is not an AI tool, but we hope it's a very useful tool. Um, I

want to say right away that this is in the early stages of development and it's a long process. I want to shout out uh Manfred Gabriel at Holland Tonight for getting super excited about this idea.

Yeah, when I brought it up last year, um his excitement was one of the reasons that I we really decided to reinvestigate uh a lot of the work we had done around workflow building in the past and realized this is the time to do

it. So with this tool, you'll be able to

it. So with this tool, you'll be able to construct complex workflows and have your documents automatically flow through them with reporting on each step. The steps of the workflow could be

step. The steps of the workflow could be anything in Everlaw. A first pass review, a second pass review, running search term reports, applying batch actions and other tasks, and everything else we can hook into. But you can also

plug in AI where it makes sense. Writing

assistant, coding suggestions, translations, extractions, and more.

You'll be able to trigger workflows automatically when new documents come in or manually. You can add on sampled QC,

or manually. You can add on sampled QC, conditional branching, and gating based on human approval. So imagine this future, right? Instead of getting in the

future, right? Instead of getting in the guts of Everlaw, you're orchestrating outcomes. This is not just an AI

outcomes. This is not just an AI autoimmator. This is about coordinating

autoimmator. This is about coordinating everybody's work. Even when the task

everybody's work. Even when the task doesn't require any AI at all, your colleagues can step in at exactly the right time to add value in a defensible, repeatable way. You can use AI to solve

repeatable way. You can use AI to solve problems when the budget requires it or when the client stakeholder allows it.

You can see exactly where each of your documents is at any time. which ones are moving along, which are stuck, and which will require additional resources to hit your deadlines. So much like I

your deadlines. So much like I demonstrated deep dive on the stage last year, workflow builder is in its early stages. And our success here is built

stages. And our success here is built around partnership between us and you.

So if you'd like to be a design partner on this, please reach out to your customer success manager and we can get you into the program the same way we did with our other betas like with Deep Dive. So why does all this matter?

Dive. So why does all this matter?

Ultimately, our technology itself is only a means to an end for you.

Providing better outcomes for you, right, for your clients if you're at a law firm or internal business teams and stakeholders if you're at a corporation or accomplishing your mission at a public sector agency or nonprofit.

Workflow builder is another critical piece of that solution. We think it'll help you be an even more strategic partner for your stakeholders.

I'll just say first of all, thank you so much for your time with me today. Um,

we're committed to being your best technology partner. This means creating

technology partner. This means creating the best tech, making it accessible to you, and pushing the frontier of what's possible. It also means thinking long

possible. It also means thinking long term, anticipating your needs, being a trusted adviser and a thought partner, and providing stellar support and service. These are lofty aspirations,

service. These are lofty aspirations, and we work towards them every day to ensure that you're getting the better outcomes that you deserve. Thank you.