And just like that, we are live. I hope everybody enjoyed our super awesome introduction music. Nominate us for a Grammy, if you don't mind. I'm gonna give everybody just a couple of more minutes to filter in before we go ahead and get started, but I just wanna first start off by welcoming everybody. Super excited that you are here. We have a lot of really cool data to dive into today, as you heard in the song, and really looking forward to answering any questions that everybody has and just sort of updating everybody on what the last year of Kodex has looked like in the data. Allow just a little bit more time here.
Shout out to John in the chat with the enthusiasm coming in early. Let's go. I like to see it. Awesome. Well, in the interest of time, I'm gonna go ahead and get us all started here. But first, before we dive into anything, just a brief glimpse into what we're gonna be covering today. Pretty simple, pretty straightforward, breaking it out into three main chapters, calling chapter one the volume. Chapter two will be the clock. Chapter three, the question. We're gonna talk a little bit, on some forward looking stuff, what we think, we are doing to help replace the traditional inbox. More on that to come. And then we are going to answer any and all of your questions.
So please do not be shy with the questions. My name is Fletcher Stubblefield. I am a senior product marketing manager here at Kodex. Very excited to have the opportunity and the privilege to host this really cool section along with, a lot of my colleagues, including Ike, from engineering, Ayush from solutions engineering, and then Jane, from product. If everybody wouldn't mind, I'm gonna allow all of you to give yourselves a brief intro, so everybody knows who our speakers are. I appreciate that, Fletcher. My name is Ike Anude. I am the... One of the engineers on the integrations team here at Kodex. We focus mostly on helping our customers create and build third party integrations.
Awesome. Thanks, Ike. I'll go ahead. Great to see everyone. My name is Ayush. I am on the solutions engineering team here at Kodex. So my day to day is working with many of our clients who are on the team that are actually responsible for doing all the legal responses at their respective companies. Awesome. Thanks, Ayush. Hi, everyone. I'm Jane. I'm on the product team. My team talks to the clients and, looks out at what's happening in the markets and just helps... Works with our engineers to figure out what we build. Great. Super looking forward to it. And, with that, let's go ahead and begin today's session.
Now before I dive straight into the questions with chapter one, I do wanna open with this, because at Kodex, we have a little bit of a theory. And this theory was kind of the basis behind this standard report and having us want to do more to bring more of this data that we're in a kind of a unique position to share with a lot of people in our industry. And so on that theory, we basically believe that there's still a lot of companies out there who are using an inbox that was never really meant to carry what it actually carries. Subpoenas can land in it between password reset complaints and vendor spam.
A warrant may arrive as a scan of a printout of a scan, its signature checked by eye against the last one. And for a long time, we think that that arrangement held decently well, limited by technology, and all sorts of other things. But like anything else, it matures. Right? The problem was is request kept coming and more requests kept coming. And we really noticed when we dove into the data this year that it met four pressures all at once. The first was volume. A lot more requests, 57% more across our network. We're gonna dive much deeper into that here in just a few minutes. The clock, there were a lot more emergency requests this year.
And then more than four in 10 of those requests actually arriving after the office had gone dark. We're gonna touch on that. But then the case, about one in 10 of those reached more than one company. So this stuff can get really, really complex. And underneath every single request, the question of whether the person asking is who they actually says they are is kind of the heart and soul of what we're all concerned with and where Kodex really, really does a lot to help keep everybody safe and and keep this process moving in a... In an orderly fashion. So housekeeping, just very quickly, we're gonna leave the Q&A box open the entire session.
We will have, a chat open for everybody to use. On the data, every single figure used is an aggregate. No client is named. No client is revealed during the research. So just a couple of those things that I want everybody to keep in mind. But please keep your questions coming in, and thank you for joining us. With that, I'm going to go ahead and get us started on chapter one. So the first question I have is for you, Ike. And as I talked about in the introduction, we're getting a lot more volume. We touched on this in our keynote presentation, which for those of you all in the audience that don't know, Ike is one of our superstars here at Kodex who did a phenomenal job in that presentation.
So everybody go watch the keynote if you haven't seen it. But on that 57% figure, is that just Kodex adding companies, or what does that 57% actually represent? And then how do we actually separate network growth from the work on the same desks inside of a company? I appreciate the question, Fletcher. So 57% does encapsulate some of the network growth plus this organic growth that we see for each one of the companies that we have. However, when we take a look at each company that we have on our platform individually, we see on average 20% year over year growth. So this means that every company is getting on average at least 20% increase in the amount of requests.
And 72% of our customers say that request volume has grown over the past year. We didn't get a single customer that said it hasn't. And based on the amount of regulations that are coming down that are enforcing a more structured way of taking... Of receiving requests and sending requests, we can only see that number going up in the future. Yeah. And I think that's really interesting. And, Ike, if you don't mind, I'd like to keep you here for just a second and ask a brief follow-up. Thinking back to the keynote, which, again, everybody go check that out if you haven't already, it wasn't just that the request volume itself had increased, but you mentioned something about those requests themselves are getting heavier.
Would you mind just touching on that for us and telling us what does heavier actually look like in the data? So when I say heavier, I mean something like the analyst or whoever's receiving the data has to do more work to comply or complete the actual request. So, these are things like search warrants, things like subpoenas require, oftentimes require more effort than a preservation request. And we see that search warrants and subpoenas have increased quite quickly over the past year. So search warrants have more than doubled, about 130% year over year. Subpoenas as well have increased about 70% year over year, and these are the two fastest growing categories of requests that we have.
Yeah. And those are really great insights. So thanks for touching on that for us, Ike. Now, Ayush, I'm gonna pivot to you for just a moment because we're kind of setting the stage here and talking about how volume itself is increasing. And it's not just the volume that's increasing, but the work itself is getting chunkier and weightier. But one thing that was revealed to me through the data that stood out that I thought was super interesting was that not a lot of these teams are super large. So what does, like, something like a 60 request month, for example, feel like on a team that may only have three or four people?
And then what else happens in the month that may not be typical, or what else may happen in a month that itself is not typical and there's more volume to handle? Yeah. Definitely. I would say from from the data, looking at something like a 60 requests a month average, feels reasonable. It feels definitely manageable. But I think in the real world, it often plays out differently. There's a couple of gotchas there. So for one, that might mean you get... You're getting two requests a day. That could very well mean you're getting on average maybe two requests every Sunday when you maybe don't have coverage or those requests are coming in in the evening, in the early morning.
It also can mean there's all sorts of variation in the real world that plays out. You might be getting 60 requests a month over the course of a year, but that might mean some months you're getting 30 and other months you're getting more like 90. So being able to, you know, accommodate for not just the median, the middle case, but also making sure you have operationally the systems in place to handle higher case volumes as well as the staffing required for those months where there might be fluctuations and you're seeing a lot more requests than you do typically. In the real world, you know, averages don't play out every single month.
It's an aggregation. In reality, every month might be different and pose different challenges. Yeah. And I think that's spot on. And you also touched on a keyword for me when you said systems, which I think is a really natural segue, to bring you into the conversation, Jane, because as our fearless product leader, I know that you have a love and a passion for systems. One of the things that also stood out in the data was that we realized that size actually decides volume far more than sector does. And so if I can just break that down a little simpler, like, companies that are similar in size are more analogous to each other than companies that are similar in terms of sector or the vertical in which they operate.
So from your position and your perspective as a product leader, what should a team take from that when it's planning people and tools and process or any of those sorts of things? Yeah. It means that when you're looking towards other companies to see what you can improve on or what practice you can look at, it's not just the ones in your industry. You should look at ones... Organizations of similar size. But I want to kind of follow-up what Ayush was saying about the spread. That's something that's really important for particularly small teams to plan around because you might have a small team, but you could still get a very large month.
So one of the things that I've seen is that it's important to have a good triage process. Of course, we all want to deal with every request immediately. And when you do get those volume strikes, having that triage process, planning the people and the tools around that, understanding what's critical for your company, that can really help. Yeah. Thank you for that. I agree. And it's been really interesting to watch how you have worked and tried to address this stuff from a product perspective, and we're gonna touch a a lot more on this as we get deeper into the presentation. But thank you for that. And now, Ayush, I'm gonna tag you back in for just a second because to kinda put all these ideas together.
Right? I think one thing we haven't touched on yet is maybe the complexity of some of this stuff because we've talked about volume. Right? Volume is going up. And as Ike has done a phenomenal job at pointing out, it's not just that volume is increasing, but the volume itself is getting heavier. There's all sorts of, you know, unplanned events or externalities that may pop up that may make a a month even heavier or more atypical for a small team. And then we know that size is really what is, the biggest predictor of kind of volume. But the complexity piece is so interesting to me because one of the examples that popped out, and we saw lots of this when actually diving into the data, was this this thing we talk about it internally a lot, these various types of scams.
One of them that I think is super interesting is the romance scam. So for example, we talk about this scam where maybe somebody interacts or meets somebody on a dating app, for example. And then from there, they build up some trust and rapport, and they maybe move that person off of the dating app and over to a crypto exchange. And maybe perhaps they're trying to get some crypto from that person, but maybe they're using an email address that they use to register an account on both the dating app and the the crypto exchange. So when it comes to these kind of really complex, scams and and hostile actions that span multiple accounts across multiple companies, what does that actually look like inside just one of these companies?
Yeah. It's a good question. I think from the perspective of any single company, you're only seeing the requests that you receive. You might see references to a broader case number, but that doesn't mean necessarily you have a full picture or a clear view into the breadth of the case, how many more requests might be associated with it. And us being able to kinda aggregate all of our different data, we noticed some interesting trends there. So something like 10% of requests hit more than one company or excuse me, 10% of cases hit more than one company, meaning investigators had to send multiple requests to multiple different parties, whether those are multiple different dating sites, multiple different crypto exchanges.
And that those cases, the ones that are hitting multiple companies, were representative of nearly a quarter of all requests total. So there is certainly this scenario in which a small subset of cases is generating the most request volume across the board. These kinds of romance scams, Fletcher, like you had pointed out, in which an individual might be going to multiple crypto exchanges, multiple dating sites. But, essentially, it's a single person who's running essentially the same crime over and over across multiple different platforms can generate a lot of these requests and make it tricky for the investigators too to track all the different requests that they have.
And having a system in place broadly for the investigators too where they can track that casework and see all the different requests across the different companies, I think makes it easier, faster for them to identify those patterns and finally nail these people. And that's really interesting and consequential to me because a couple of thoughts immediately is one, like, 10% considering that I believe in the report and the data we use that as sort of a a minimum of baseline, but 10% is not a small number. That's a relatively large percentage of these cases. And the example I gave, I feel like, is a a relatively simple example in and of itself with just dating app to crypto exchange, and I'm sure that some of them are much more complex.
So it's just really interesting when you actually dive into the data, how we can start to see these patterns emerge and, just something to keep in mind that I think is absolutely fascinating. So thank you everybody for this discussion on part one, the volume. But from here, let's move into chapter two, the clock, because we've talked a lot about volume and complexity and what these kinds of things mean. But when it comes to the timing of these requests and responding to these requests, I think there's a lot more interesting stuff that we can dive into. Ike, I'm gonna kick it back over to you with the first question, on this topic.
We have an exhibit in the report, and it talks about when these emergencies actually arrive. So let me just ask you bluntly. When do these emergencies actually arrive? So emergency requests are unique in the sense that they tend to arrive about 40% of them, 40%, 45% of them tend to arrive outside of business hours. So we define business hours by 8AM through 6PM in the center's local time. So this can come on a weekend, late at night, like 12AM. It could come at any time, and these are oftentimes the most sensitive the most sensitive requests. Like, there's an immediate imminent threat to life. So when these things come in, we...
These these, our analysts, our customers will help field these requests, and we have seen the turnover go from about fifty five hours to about twenty nine hours over the past year because of all of the changes that we have made through our Kodex platform and some of the changes that a lot of our customers have made internally to keep up with increasing demand. There's... Yeah. Before I continue, before we continue, I did wanna touch on something that Ayush and Fletcher, think you guys said earlier, which is that, like, the... When... For example, in a romance scam, we oftentimes see that it's not just one person.
It's like a network. So, like, you know, it might be affecting... They might set... Like, part of the investigation exists in Match Group or some Match Group brand, and then another part of investigation has to happen on Coinbase or Binance or some other crypto exchange. What... One interesting part of this is that, like, usually that one person that is the target of this investigation, say they're a suspect, is not acting alone. They're acting in groups. And one of the interesting things about Kodex providing such structured data is that it allows for us to understand and see patterns, not just the analysts, but also the agents to see these patterns across different, companies.
Yeah. And thanks for adding that because I think it's just super, super interesting how it all works, and it's really important to keep all of this in mind. And the fact that it's not just one sort of attempt to get data from one person or one company, but it's multiple people and groups of people doing this across multiple companies at once all the time. And really, really fascinating to be able from our perspective to see how that actually starts to manifest itself in the raw data across the network. And we're gonna touch more on some related things to this in a minute. But to pivot just a...
For just a moment back to this idea of emergency requests, because I'm sure a lot of emergency requests spawn from some of these same investigations, one of the other things that stood out, and Ayush, I'm gonna kick it over to you here, was that 28% of teams that we surveyed said that they aim to answer an emergency request within an hour. Now where would you start setting a target if you're one of those companies that may not have any defined metrics or process yet or sort of standardized target across your team on how fast and how timely your responses to these types of requests should be?
Yeah. I know in our survey, we had found that one in five teams really didn't have any target internally for these emergency requests. So my answer here is gonna sound maybe a little too obvious, but the first step would certainly be to sit down with your team and come up with a target, agree upon this is what we think we are capable of doing in terms of turnaround time and all the various considerations you would have internally for, you know, what is a realistic number. And then it's important to also make sure you're operating and working inside a system that is able to track your own targets.
So you have data as you work through these emergency requests. You're able to also pull data and see how you're tracking against that internal target, that internal turnaround time that you may have established. I think another important thing is also to account for, you know, specifically for emergency requests to account for the fact that nearly half of them arrive outside standard business hours, which I believe in our survey we we had defined as between 8AM and 6PM in your respective time zone. So it's important to make sure you have coverage for, you know, that big bulk of emergency requests that are landing outside. And amongst our clients, we see all kinds of different tools, processes in place to kind of accommodate for that.
So whether that's an integration with a system like PagerDuty, having an on call rotation, even integrations with, like, Slack, somehow when an emergency request lands, making sure somebody at the company is notified and aware. And there's even other processes, next steps we could take to even try to automate some of the steps in an emergency. So when they do come in early morning on a Sunday, they're not just sitting in the queue for hours and hours. And I'm super glad that you pointed out the connection there because you prevented me from having to ask you my next follow-up question was... Which was going to be, so if you take these two sort of numbers together, how do they interact and how are they related?
But I think you did a great job of really explaining sort of how these things kind of start to stack on each other and start start to compile a little bit over time. And to add to that complexity, Jane, I'm gonna tag you back in here for a second because something we haven't touched on yet is holds. And one thing we know is that a hold that, excuse me. And one thing we know is that a hold nobody extends ends at ninety days, and only 6.8% of investigators ask to actually extend that hold. How does the product itself handle that that clock, that internal clock, or is that something that's entirely sort of on the teams themselves to manage?
Yeah. We've been seeing a real increase in preservation requests in our platform. It's now one request in 10 now. That's up year over year. I would say, you know, we talked about how important emergency requests. It's so obvious. Obviously, it's time sensitive, but preservation is also very important because it means that for those more complex cases that take longer to solve, you're preserving that data, making sure that's not being deleted or not disappearing. It really does surprise me that only 6.8 of law enforcement agents ask for extension because we do make it pretty simple in our platform where the company can configure that whole period, the default is ninety days as you've mentioned.
And then an extension isn't just another email that you have to find or like look through your system. With this more structured request, it's just a comment on the request itself and we notify the law enforcement agents at least ten days before it expires and then on the day itself so that they can ask... They know when to ask for that extension. So this is definitely one of those improvements that you can make if you're thinking about moving from an inbox to a more structured request. So everything, the preservation and the subsequent requests, if it does come, is all kept inside inside the same case.
Yeah. Thank you. I think that's a great call out, and I think it's just a testament to sort of how we have over time been able to take a lot of this data and compile it into such a way where we can activate it inside some of the product stuff. But I don't wanna go too deep into that today, but just something that I think is, really cool and and that we should always remember. Now I kinda wanna start to combine all of these ideas because in chapter one, we talked about volume, and you all did a great job of explaining what the numbers behind that volume looked like.
And then we've spent some time talking about the time. Right? The clock itself. But combining those two things, Ike, I'm gonna bring you back into the conversation for just a second because as we talked about, I think the very first question I asked you, keep me honest if I'm wrong here, was about volume rising 57% across the network overall. One thing that jumped out to us in the data was that the typical request actually still resolved about 40% faster, and I know you dove into that in the keynote as well. When you first saw that in the data, were you surprised? Did you doubt it?
Were you skeptical? Kinda... What was your reaction to finding out even though that volume was up so high, requests are still getting resolved much, much quicker? Yeah. At first, it was a little surprising. And I think on face value, there's more requests and they're heavier. So you think it'd take a longer time to resolve them, but I think our customers and our platform has adapted to this, and we have... There are about three levers that I kind of... Well, three levers that I have or really that customers have to be able to comply with these requests more quickly. So this is one better structured intake from law enforcement, increased notifications.
So... And as soon as a request lands in your queue, you get notified. And then internal policies and expectations on the customer side around when, well, how quickly we need to resolve all of these cases. So those are the three big things that I think have contributed to an increase in increase in volume and the ability for our customers to be able to resolve them about 40% faster. And we see that the industry and really government, also governments are starting to realize the importance of things like structured intake. A good example is that e-Evidence is... One of the big things about e-Evidence that allows for structured intake across countries in the EU, and that is a huge far cry from the MLAT agreements that we used to have, and really even the precursor to, the e-Evidence Regulation, which was European Investigative Order Regulation.
These... The creation of a structured intake process makes it way easier to comply with the request and makes sure that anything that law enforcement requests from our customers or really anyone is appropriate and legally scoped. Yeah. And I think that's a great response. And I just... I can't help but thinking back to... As I'm reading through all this reporting myself, that seems to me that particular point right there, Ike, like, a real testament to sort of the network power itself. Right? Like, the fact that we can handle so much more volume across the network as a whole, but we're still making huge, huge strides in in cutting down that response time and just making it so much more efficient to respond.
And, Jane, you touched on this a lot, from your perspective on the product side, but something that I think is really, really cool. Now, Jane, since I've already got your attention and keeping it here for a second, one thing we haven't touched on yet are actually rejected requests. Something that we saw in the data was that 44% of rejections came down to the legal process. A quarter came down to basically not having an account. 11%, if I'm remembering correctly, to incomplete or duplicate requests. Which of these, you know, sort of rejection types can a better request process itself prevent? Yeah. I I think that, obviously, there's there's a lot of education that we can be doing for law enforcement agents to help them understand what's required in order to get a complete or prompt response.
That 44% is high. So being really clear, creating good guidelines. Maybe if you're using a structured system, you're enforcing some of those requirements in the intake itself. And then making sure you follow-up on a request. So simple things like just saying, Hey, we got it, so that the requester doesn't submit a new one thinking maybe it didn't go through. That's going to get rid of some of your duplicates. And looking ahead, and I'm sorry, this is the product nerd in me, I think there's really an opportunity for the industry as a whole to start using more automation to check requests before a person has to.
So there's a lot that we can do here to get that 44% down. Yeah. And I think that's awesome. Exactly what, I was looking for. So thank you for your insight there. And I think we're at that point now where it's another good natural segue. Let's go ahead and dive into chapter three because we've talked about the request volume itself. We've talked about what I'd like to call it the clock or sort of the timing behind all of this stuff. But going to the question, which is my favorite chapter in bringing in a lot of sort of the data that we got from our anonymous user survey that we sent out and all of these other pieces.
Jane, I'm gonna keep you for just one more second. So sorry for me hitting you with double triple back to back follow-up questions here. But on this idea about who's really asking, one of the things that we learned is that more than one in 10 sign ups of Kodex itself ended up being suspended. What actually happens... Just something I think I would be really interested for us to be able to share with everybody today is, like, what actually happens between applying for access to the Kodex network and actually sending a request? Yeah. This is the core of what the Kodex platform does. And we all saw this year what can happen if these checks aren't done outside our platform.
So just to kind of go through, every law enforcement agent that signs up is verified before they can send anything to any of our clients. So even a suspended account, so if anything looks off, we will actively monitor all these signals. And anything that's off or gets flagged, our team is on it and looking closely at it, and we will suspend the accounts so they can't send a request. So I personally, before Kodex, I spent a decade in cybersecurity, so I know that the threat landscape is constantly evolving and how important it is to never get complacent. So we're constantly evolving our detections as well, looking for new signals, new techniques, new ways to detect this anomalous activity.
And this is really the core of it. It's defense in-depth. If you're thinking about doing verifications, you have to remember no single signal is going to catch everything every time. It's all about the combination those signals lay together and that's what gives you that protection. Yeah. And I think that's a great call out, and I love your background and all the experience that you have in this type of security. Myself being a little bit newer to the industry, I definitely don't have that background that you do. And I... I'm actually gonna bring you in here to help me out with sort of my follow-up to that because something that stood out to me as somebody who has spent a lot of time recently sort of learning the industry and learning how all of this works, it really jumped out at me when I learned that most of these suspensions, most of these attempted sign ups that end up getting suspended aren't actually attacks.
So if they're not attacks and if they're not a threat, what are they, and what actually surprised you, or if anything? Yeah. I would say that surprised me too, that when we look at the suspensions, they're not attacks, they're not from people we can confirm as threat actors. So out of the 727 accounts that were suspended on security grounds, only 56 of them were actually assessed as being compromised or being actual threat actors. Most of them... So the 676 of the... 671 of them, were for anomalous activity. So this could be something along the lines of maybe they log in from a location or through an IP that is very different than what they usually do or what their agency does.
And this goes to show that, a, we are hypersensitive, which is a good thing in this case because we're trying to make sure that everyone who accesses the platform is verified and that our front door is working. And it is a lot better to reject and then reconfirm someone than to allow someone in. So we'd rather fail closed than fail open because it only takes one for an attacker to compromise the network. Yeah. And I think that's really important to call out, and I'm gonna daisy chain my follow ups here by Ayush. I'm gonna bring you in now because one of the immediate questions that popped up in my mind when I'm kind of going through our preparation for this stuff and I'm going through the report, I'm looking a lot of this data, really was interested in what some of our our clients had to say and looking at some of these survey results.
And one thing that stood out to me was that 100% of those teams that we asked said that they were confident in those requests that they were getting across the Kodex network, which awesome. But I also saw that more than one in 10 of those sign ups were suspended across the network. And then on the the survey itself, a bunch of people who responded, 28%, in fact, said that they still found verification kind of a hard process. How do all of those three things hold? How do those three sort of different ideas work together? Yeah. It's a great question. I think on the response, seeing that 100% of teams said they were confident in the requests coming in via Kodex.
I think this is because in the Kodex network, the verification process, the ensuring that the purse... The submitter is a valid legal authority, all occurs before the request lands on any company's table. It's not that we have to be reactive when a request comes in and now a team needs to scramble, find a phone number, find email addresses, figure out who the authority is, do all the reach outs while simultaneously trying to figure out how to address the legal request itself. All of that is happening prior to that request even landing. The 28% number is interesting too. I believe 20% of the respondents still did say that verifying requesters is one of the hardest part of their job.
And I think what's driving this is that the world of legal requests and legal response is huge. You get requests... You might get requests through all kinds of mediums for certain emergencies. You might get phone calls from a 911 operator. You might get envelopes mailed to your office room. And at an international scale, right, there's so much variation in how you might receive these requests, what the different standards are for different jurisdictions. There's a lot of variance there as well. So certainly very... We're very sympathetic to that concern around how can you validate every single legal request regardless of the medium through which you might receive it.
Yeah. And I know that you've spent a lot of time looking at this survey data and sort of the customer and client response stuff. If you don't mind, I wanna ask you just one more follow-up question because another thing that we asked on that survey was, what do these teams... What do you guys wish that investigators knew when opening a request? Or what do you wish they knew about the sort of process internally inside these teams in general? What did they say, and which of those would you underline or emphasize here? Yeah. Definitely. I think this was one of the most interesting parts of the surveys, of the survey that we had sent out.
The top response on there was that companies want requesters to give them complete accurate requests, things that are legally valid that have all the information inside it, all the metadata, all the various identifiers that the company needs to go and produce the data, find the data internally in their own systems. After that, a a common response to this question was also a desire for the requesters to, one, read the guidelines that the company provides and publicly publishes, as well as making... You know, wanting the requesters to choose the right track, making sure that they're submitting through the right process under... Labeled clearly that emergencies that are being submitted are genuine emergencies.
I would highlight that one, making sure the the requests go on the right track because it allows for efficiency on both sides. It allows for the requesters to get the responses more reliably, more quickly, and it was... It makes it possible for companies themselves once they have these different rails set up to build and think about how they can automate and be more efficient in how they respond to these various requests. I also thought that was super interesting. And, at the risk of embarrassing myself, I will stick up for some of those requesters for just a second as somebody who in my own personal life, probably way too much gets asked, before anything else if I actually read the instructions, which I will save you guys from what that answer usually is.
But from there, Jane, I'm gonna bring you back in again because when looking back at sort of these three different topics that we've tried to focus on from a few different angles, We talked about volume. You know, Ike showed us how much that volume is increasing, why the volume is getting much heavier. We started to look at the complexity of... It's not just volume itself, but it's what's actually making up that volume that can add complications and team construction and all these other three different pieces, bringing in our customer and our client feedback, bringing in survey data, bringing in sort of our our newer better understanding of the timeline of a lot of these processes.
One thing that I feel like it kinda leads us to when talking about or thinking about this report in general is I think what it's really saying is that a process now has to be shown working. It has to be shown working on a clock, and it has to include a record. Let me ask you, what does that record actually look like inside of Kodex? First of all, Fletcher, I pegged you as someone who would read the instructions first, pay... Cover to cover. Sometimes. Oh yeah, a record is a timeline of every action on request. So who asked, the legal process that was sent, who on the team handled it, what they decided and why, what was sent back and when.
So basically think of everything you need as the audit trail. One thing I found interesting in the survey was that many teams expect new regulation to be one of the biggest change for them in the new year. I think that was a third of the teams. So as you move from an inbox to a more structured system and you're thinking about these records, think about how that's processed, how that data is captured so you can report on it easily, not just record by record but in aggregate. So we know usual saying how to improve these processes, having that data so you can really reflect and look at what went well, what's working, what's not working, how long you're taking on average to respond and whether that's improving or not, that's really important.
But then also publicly, it builds trust. It builds trust with your users, your regulators, the authorities, really builds trust that you're doing what you need to do, your part in the response. So that's something that I think something like Kodex, whether it's Kodex or another more structured system, would really be advantageous for. Yeah. I agree completely, and thank you for that insight. And you've actually done a really good job of kind of teeing me up for my little bit of a a quick close here because I think you've touched on a lot of the highlights of where I kinda wanted to take this. And in my opening statement, I talked a lot about this idea of pressures.
And one thing we really wanted to look at when accumulating all of this data and trying to put it together into a report that our our clients would find interesting and useful that may be able to learn and take these these things from where we were talking about pressures and then seeing if we could go identify the source of a lot of that pressure through the data. Well, in going through that exercise and putting this together, I think that this report has really shown us a few things. One, taken one at a time, there is an answer to the origin of all of the pressures in this report.
And one thing we talk about the evolution of the inbox or where we are really, really trying to push the future of this industry is that it doesn't always have to be the case that more requests can only be met by more people. It doesn't always have to be the case that more emergencies can only be met by putting a bunch of more people on call after hours. A case that spans five companies is, to any one of them, maybe just another request, but that's not the reality of what this stuff is. And by building all of this infrastructure across a network, we can really start to observe and track and learn and defend against in real time.
And I think sort of we're we're kind of in a unique position to be able to look at that data, not just in the isolate across one company or another company or another, but across a really broad network, where we can see a lot of this stuff from multiple, multiple different angles. So really, really interesting stuff. Let me just pause here for just a second and thank all of you, our amazing panelists, and my amazing colleagues for, answering all of these questions and indulging, some of my ridiculous rabbit holes that I like to dive off into. But, with that being said, I wanna go ahead and open it up for questions if we have any questions that have come in.
And let's see what we got. And we do have a question. I'm gonna go ahead and make this one public for whoever wants to grab it. But how will AI help with document processing, for example? Let me start with this one because I think I alluded to one of my responses. I'm sure Ayush and Ike have thoughts as well. This is where, you know, I think the stat was that 44% of requests get rejected because of the wrong legal process or it's incomplete. So this is where potentially we've been working on a feature that's in beta now where we can use AI to help interpret some of this information and get it into the right format, make sure you have all the information you need to respond.
The first step is actually pulling out the targets, but we have further enhancements further down the track that will help with other parts like validating the legal process type and other opportunities that we see here as well. So that's really by using AI to help check on the completeness of the queries, the information that's there, and maybe presenting that information more at your fingertips, you'll be able to respond faster. Awesome. Thank you. Ayush or Ike, do you have anything that y'all would like to add to that? I think it could massively improve response times, especially in sort of exceptional cases that come in. I'm sure everyone on the call here has gotten those kinds of requests that were just very difficult to comply with for a multitude of reasons.
One story from a client is around, you know, getting a PDF of a legal process and on it is maybe 250 different, let's say, crypto wallet addresses. They're not structured at all. It's just a PDF. The reality of how you handle that is you open up the PDF, you physically copy and paste each wallet address into something like a spreadsheet. Now if we have AI document processing that's able to pull all of those addresses out, get them to you in a structured way that's not just structured as a spreadsheet, but structured in a way that it can be accessible via an API so you can get it in a nice clean JSON format.
Really opens a lot of doors for one, not... One, just getting rid of the manual effort into transforming this into structured data. But then once you have structured data, even without human in the loop, being able to do some kind of automation, looking up those wallet addresses and identifying, oh, turns out these... This subset of wallet addresses are not even on our blockchain network, but these ones are. And now your team is able to do focused work rather than really repetitive tasks like copy and pasting from one tab to another. Yeah. I think Ayush covered pretty much everything I was going to say. The structured data is really the fundamental kind of lever that we're pulling here.
Being able to turn these documents into structured data makes it so much easier on the analyst side, and this is realized by a lot of governments, for example, in the EU, who are using this as an excuse to reduce the amount of time that you have to comply with an order. So a good example is that MLAT used to be twelve to eighteen months sometimes. Six months is usually the minimum. The EIO, which was kind of the... What came afterwards, was a hundred and twenty days. And now with e-Evidence, it's ten days you have to comply. And all this is enabled by more structured data, and this is another thing that document processing processing will help our clients with.
Awesome. Thanks for diving into that, guys. I'm going to go ahead and make another question we had come in public. Give me just a second while it populates here up on the screen. Here we go. Okay. So we have a question from an attendee. You mentioned significant increase in volume the past year. What types of subpoenas are you seeing the volume increases in? We have not seen that. Whoever would like to take a stab at that. Ike? Yeah. I think when it comes to the types of subpoenas, that is a question that is, like, a little bit difficult to answer because it really requires us to go through and actually look through each of the type of subpoenas that we, that we're receiving on our platform.
We can only say that as an aggregate, we're seeing an increase in the amount of subpoenas that are being requested on our platform. Now, this could also be because there are different regulations in different countries. It's not evenly distributed amongst everybody. We do see an increase on average, but it doesn't necessarily mean that everyone is experiencing the same significant increase. So, some industries and some sectors. So, going back to e-Evidence again, e-Evidence does not apply to financial institutions, so we would not see an increase of e-Evidence related, requests in that way, but there could be other types of regulations that do apply only to financial institutions, but not to other types of institutions.
So it's not evenly distributed. Great. Does anybody else have anything else they'd like to add on to that? I think there's also geographical. So we've seen US increase, but it may be different with different sectors within the geographical locations. Looking at the data, subpoenas increased by 70%. What we also saw was that it wasn't necessarily more subpoenas per investigator or per law enforcement agent, more agents signing on to our network, sending more as an aggregate rather than per investigator as well. So those are some of the kind of variables that I would look at when you're thinking about projections. And I do believe we provide a pretty decent breakdown of kind of what data...
Like, what the data we pulled actually looks like and how far we broke it down. And I know that we didn't go as as far in some cases because we wanted to keep everything sort of anonymous and as, aggregated as possible, but we do have some charts and graphs and some, some additional metrics that we haven't actually touched on in today's presentation in the report. So I would also strongly recommend everybody if you haven't downloaded that and looked at it yet, I would do that because you might find some of that stuff in there, interesting as well that may touch on a lot of this, just a little bit further.
Let me check to see, if any new ones have come in. And it looks like we have mostly gotten to everybody's questions. I'm gonna throw y'all one more curveball because I have a question of my own. What stood out to each of you most in the report? What was one thing that you saw and you were like, oh, okay. That's really interesting, and and it just really caught your attention. I think one thing that, really stood out to me was the decrease in the amount of time required to comply with an emergency emergency request. So it went from fifty five hours last year to twenty nine hours.
I think it just goes to show that there's a lot of space to improve. Right? Like, there's just so much, so to speak, alpha to be gained here. We could get it as small as we want, right? And we see from regulators that they're putting more pressure on service providers to comply within a certain time period, and, I think that's, like, Internal regulation expectations and internal expectations from the service providers can allow us to comply, maybe in the future on some automated fashion with some of these subpoenas and some of these requests. So, I mean, that's kind of the holy grail, the goal, but I think seeing that 55 to 29 decrease was...
Is very, very encouraging. And then something else that I remember, to piggyback on what you're saying, Ike, was, like, way more than half of them are getting responded to in, like, forty five minutes, which I just thought is absolutely incredible, because that just seems so fast. When you think about the nature of what an emergency response is, they have so many more than half of them getting responded to in less than, you know, way under an hour is just... It's it's awesome. It's it's real world good. Jane, what do you have? Yeah, so my number comes back to the verification side where one in ten, that was something interesting, one in ten signups were suspended for anomalous activity.
Now that's different from, sorry, for any reason. And then within that one in ten, one in one hundred were suspended for anomalous activity. So that's real bad or wonkiness that we're seeing and we're stopping. So that always makes me feel good that we're at something is working and we're really checking under the covers. Couldn't have said it better myself. Ayush, we'll end it on a strong note. Yeah, I think the number that surprised me, and it should really not have surprised me because if you just considered a random distribution, it should not be surprising that nearly half of emergency requests land outside business hours. But it still surprised me because I realized I think it is, in some ways underappreciated that there needs to be business processes in place to address the fact that so many of these are landing outside those normal hours.
It's almost obvious that they would be landing outside those hours. Emergencies by no means are a nine to five. So I think it's important to highlight that that as an industry, you know, it... It's it's our responsibility to make sure we have systems in place to be able to handle these kinds of requests since there could be lives at stake. Yeah. Really, really well said, Ayush. Couldn't agree more. With that, thank you everybody for attending our presentation today. We wanted to keep this a little bit more casual, have this feel like a little bit more of just a conversation instead of us trying to just throw a bunch of numbers at you via webinar.
So as always, we would love your feedback. We always love your engagement, your participation, and reach out to us. Reach out to me. Reach out to anybody. Reach out to our team. We love talking about this stuff with everybody. We're happy to answer any questions. I would love for everybody. If you haven't already seen our keynote, go check out our keynote. It's phenomenal. You'll see some friendly faces that you saw today. Like I said, Ike is an absolute superstar. You wanna make sure you check him out. Go check it out over there, as well as the standard report itself. So download the standard report, read it, see what you find interesting.
Maybe there's some tidbits that you can take with you back to your teams, and just maybe it offers a little bit more perspective that, you may not always have the opportunity to see of of what this stuff looks like, across the industry as a whole. So with that being said, thank you again, everybody. Look forward to the next one.