Hi everyone and welcome to our webinar on reimagining archiving. We’re just going to give people a few minutes to join and then we’ll get started. Welcome to our webinar on reimagining archiving for AI driven compliance, and thank you for joining us. My name is Yael. I’m the director of product marketing at Shield. Today, we hope to provide you with an insightful and informative session looking into why and how leading institutions are reassessing their archiving away from just a compliance storage and towards being a strategic business asset. But before I dive into what we’ll be covering in this session, let’s start by meeting the panelists. We’re here with Cheryl MacKinnon, Principal Analyst at Forrester. Cheryl has spent over a decade advising technology leaders on enterprise content management, content platforms, information archiving, and records and retention strategies. Prior to Forrester, she led her own consulting firm working across both private and public sector organizations. Joining her is Ophir Shapfay, Shield’s cofounder and CTO. With over twenty five years of experience in product and software development, Ofil’s expertise focuses on the intersection between AI and building large scale real time systems. And last but not least, Anne Gardner, one of Shield’s compliance and eDiscovery experts. Aaron brings more than twenty years of experience in eDiscovery, including serving as a lead advisor at a top three global provider, leading litigation support at a top law firm, and founding his own boutique consultancy. Our webinar today will start with Cheryl taking us through a look at the archiving market as it stands currently and what the latest shifts mean for it. We’ll then move to Aaron who will go through what is needed to support investigations and stand up to regulatory scrutiny. And then we’ll hear from Ophir on what modern archiving from an architectural perspective looks like. And to wrap up, the panelists will answer some of the main questions that you submitted to us. Before we start the session, I’ll take you through some quick housekeeping pointers. The webinar will be available on demand after the session for anyone who wants to rewatch or share. All attendees are hidden and muted during the session itself, but you can ask any questions live in the chat during the webinar, and the panelists will answer directly. Any questions not answered during the session, we will be sure to respond to them by email. And lastly, a note that all discussions here are personal opinions. As you engage in the content, if you feel you’d like to hear more about Shield’s archive offering, you can simply click the link pinned in the chat, and a Shield expert will get back in touch with you. So without further ado, I’ll hand over to Cheryl to get us started. Great. Thank you very much and a big thank you for your invitation to be panellist on the webinar today. So, yeah, as mentioned, I’ve been covering the archiving market here at Forrester Research now for about thirteen years. And if we go to the next slide, we’ll talk a little bit about, you know, what we’ve seen in that time. Now, archiving is a mature market. We know that compliance has been a requirement, storage management has been a requirement for many, many years. You know, we think of an archiving platform as one that helps you migrate digital information from one or many source systems into a repository, you know, to retain it for a specific period of time. Now while it is a mature market, I would say that the last, you know, one to two years have been, among the most innovative that we’ve seen in a very long time. The chart you’re seeing there is from our last landscape report. And, of course, the main trend that we see in archiving in many technology markets is that AI has given fresh life and fresh opportunity to this market in providing organisations with new ways of working with the information that they’re capturing and preserving. Challenge continues to be that there are many data sources. We’re seeing proliferation of communication types and conversation types, so we need to constantly keep pace and innovate on how we capture that content. And the disruptor is that you were seeing, you know, large vendors engaging in things like M and A. So there may be some market disruption on the horizon here. But, again, adoption continues to be strong. Some of our data tells us that eighty percent of organizations have some form of archiving in place. And again, this, while a mature market, is a really intense period of innovation across many vendors. Let’s go to the next slide. Now, the key use cases that we see for archiving haven’t changed tremendously over the last few years, but, we are seeing some emerging use cases. The fundamental use case that we’ve seen is, you know, overall information life cycle management. That does remain the core use case for the archiving market, particularly in regulated industries where we know that we need to capture particular kinds of content and communication, retain it for a period of time and often use that for investigations and legal discovery. If we go to the next slide, we also know that there’s a whole range of other related use cases that some organizations may need, other organizations perhaps may not. We know that in regulated industries, particularly financial services, things like supervision and surveillance of communication is absolutely key to maintaining compliance. And we also know that, you know, because we’re seeing a lot of fragmentation and proliferation of different communication and content types, organizations may have use cases to need to capture and archive perhaps structured data or documents and files in addition to chats and email. So again, the use cases do vary, but that fundamental one of information lifecycle management does remain consistent for most adopters of archiving platforms. Now, we are in a really interesting age of innovation, so if we reimagine what archiving can look like as we enter this age of AI, let’s go to the next slide and talk about some of the innovation that we’re seeing. You know, some of the challenges, and I hear this from, customers, you know, on a regular basis, again, particularly in large regulated industries, is that the conversations that we have, you know, both internally but also with our customers and our clients are now multimodal. And I think this is one of the the factors that’s really spurring vendors to need to continue to innovate and, look at modernizing the way that we think about archiving. We are not just sending emails or chats anymore. We’re using audio. We’re using video. We have transcripts of these conversations, call summaries that are automatically generated for us. And so the governance can actually become quite complex. We now can see that we have, in many organisations maybe governance gaps across different meeting types. So we may have different approaches for retaining email versus, a video or audio conversation. And we saw, especially, you know, go back five or six years when we were all working remote, we often saw that governance was lagging adoption of a lot of these new technologies. And so we see organizations, in in many cases, almost playing catch up in in some of these areas. So that lack of awareness perhaps about what the requirements are for retention, deletion, and the security of of of protecting some of those conversations, which may contain, you know, confidential information. Legacy archive platforms, those that have been in market for twenty, often thirty years, may struggle with the complexity of some of these new media types, particularly video, which are often very large in size and you know, if for, many traditional approaches where you’re capturing, media formats in perhaps, EML format, we may lose some of the context and texture of some of those rich media conversations. So we’re now seeing, as we shift into alternative, conversation types, there are rising expectations to be able to discover and understand, the communication patterns happening across, you know, an increasingly diverse set of data types. You know, some new risks are emerging from some of this. You know, these meeting records, again, with multiple formats, they are electronically stored information, you know, ESI, which means they could be discoverable. So we do need to think about the diversity of the conversation types that we’re having. As I mentioned a moment ago, videos can be very large, with multiple people with video one, an hour of recording can easily be over a hundred megabytes. So that that scale in terms of what we’re capturing is tremendously affecting how archive platforms need to operate. The rise of things like transcripts. And, you know, I’ve I’ve talked to organizations that now need to think about, you know, is there a different retention for the textual transcript versus the audio or video recording? We need to think through some of these. And it means that we’re now looking at protecting intellectual property, other sensitive information, financial data, road map information that now may be contained within these video conversations. And so again, we are also seeing that there’s a proliferation of these, communication and chat apps, especially in the mobile world, where organizations want to make sure that they’re capturing that information regardless of how customers want to be communicated with. And that does include the rise of, you know, quick video apps for, you know, more collaboration and even microlearning. So again, proliferation of data types and the need to really think about how we capture and preserve and manage that consistently. So that app proliferation has absolutely been a compliance concern for the last several years and, you know, if we go back the last two to three years, there’s been about two billion dollars US in fines that have been incurred by financial services organizations that haven’t been able to accurately capture or consistently capture the communications that are happening on some of these apps, or they’re inappropriately deleting some of those conversations. So the the the the the fines are real. The impact is real. We do need to pay attention. We know that, organizations want their externally facing, regulated employees to meet customers where they are, communicate the way that customers prefer, but we need to make sure that we’ve got the right governance and compliance wrappers around that. So it means that archiving platforms really need to keep pace. Again, the the large array of connectors and integrations that need to help, regulated employees in particular safely engage with customers and meet the the customers in their preferred styles and channels. When in conversations with financial services organizations, they certainly do express the completeness of the capture, and they do worry if there’s gaps in conversation. So if conversation moves from email into a phone call or a chat app, we need to be able to kind of trace that conversation and make sure we have a complete picture, for later supervision and and, surveillance. So AI has had a tremendous impact, you know, on so many enterprise applications. Natural language and being able to kind of to interrogate our documents and our conversations is becoming mainstream behaviour. We know that many, many users are are using things like ChatGPT in their day to day consumer life, and now those habits are coming into the work environment. But it it means that there’s, you know, some opening opportunities for those who are conducting things like investigations and discovery. It means we are bringing the opportunity for natural language to query and identify relationships and and how those conversations perhaps, spread over a period of time. So traditional search, you know, keyword, metadata, that’s not disappearing, but we know that it’s had limitations over the years. So we now can AI enable some of the search to have that best combination of natural language, understand sentiment, intention, relationship, alternative terminology in terms of the vocabulary people use, but we’re not losing the value of things like metadata and dates and so on. It also means that we’re we’re able to access, engage with things like rich media or very long complex documents such as contracts or mortgages. Those are no longer barriers to understanding because we have tools to help us summarize and get to the key points that we really need to understand. And it also means that there’s rising interest in using natural language to query even structured data so I can ask a question about usage patterns and not have to try to interpret a chart or a graphic. This data point is actually coming from our recent software survey about digital workplace technologies And it’s interesting that we asked the question about how people are using things like AI today. And the top two responses I thought were really interesting. It’s kind of fairly consistent across regions. The things that people are doing with AI today are researching and finding information to help them do their job, and they’re synthesizing information to help them do their job. So these these two top use cases, I think, really lend themselves well to a lot of the investigations and discovery that organisations need to do with their archived information. The next thing on the agenda is the rise of AI agents and again, this is entering pretty much every software market that we see today. And I think, you know, I’ve been I’ve been kind of an advocate for the last ten years or so that, you know, an enterprise archive is is kind of sitting on a lot of enterprise knowledge that is an untapped resource for so many organizations. And I’ve often seen that, you know, there’s a lot of hidden insights and, nuances that we can gather here from our archive information. But perhaps without the scale of AI, it’s been a goal that’s been out of reach for many years. Now perhaps we’re on the brink. The rise of agents means that we can start, developing these assistants to help us perform tasks, you know, whether we ask them to or perhaps even out of view. And I think there’s so many opportunities to bring that into the archiving environment. You know, I’ve seen examples such as governance agents to help optimize our storage and retrieval for things like inactive information. So I don’t need to run these reports. The agent is looking at this data for me. Things like supervision agents, they can look at the broad scope of a conversation, so we don’t need to rely on just single keywords or phrases. These surveillance agents can actually look at a whole scope of conversation and help me understand what the intent of the communication might mean. Things like legal hold agents in terms of helping us, hold, emails or other items that may match the intent, if it’s relevant to our discovery. And even to enrich the data that we’re collecting with categorization, classification agents, helping us identify and secure sensitive data and extract more meaning. So I think this is kind of the next brink of innovation that we’re seeing in this archiving market is the rise of agents and the new opportunities it brings. Another data point is from one of our recent AI surveys at Forrester Research. We see sixty percent of organizations are today exploring or piloting agents in one at least one of their enterprise applications, with about twenty four percent being in actual production use, many fifty percent there are in in more pilot mode or experimentation mode, you know, planning or exploring. But there is that twenty four percent with some form of production agent today. But I think that number is going to continue to grow over the course of the year. In terms of what the archive buyers, know, when I’m in customer conversations, what are they looking for today? So again, thinking beyond the archiving assumptions of the last twenty plus years, what are we looking for in this age of AI? First, again, the broad set of connectors and integrations for that increasingly fragmented set of digital workplace applications, whether it’s mobile, web, and so on, that, you know, house the structured and unstructured data that we need to be able to capture and preserve for our our legal compliance and and enterprise knowledge purposes. And it does mean, in many cases, extending that governance to the inputs and outputs from AI. So the prompts that people might be using as well as any content that’s been generated by AI. I’m also seeing a trend towards, cost models and pricing models that support more self-service and use of the data in the archive. So, you know, opening the door to more users, being able to access that archive data and ask those questions to understand what might be going on. And organizations that are, again, looking at more of those knowledge discovery use cases in addition to a lot of the investigative requirements which have been traditional. Customers are also asking about, you know, modern approaches to things like governance and security, making sure that there is appropriate ways to detect and flag any kind of sensitive data, understanding that it can support any kind of data protection privacy regulations that we see in many jurisdictions now, and again, the value of some of the more modern cloud platforms. Organizations can now, you know, rely on their vendors for more iterative feature delivery, continuous improvement. So the vendor takes responsibility for feature delivery, innovation, security patches, and so on. So, a lot more of reducing the burden of work for, in house IT to need to manage these archiving infrastructures. And just as we wrap things up, you know, I think finally, while we know that we’re in the age of innovation and AI, a lot of those fundamentals are really key, to regulated industries. And so having that strong foundation of information lifecycle management, being able to tailor the retention pool rules and policies according to the value and the meaning of those communication or data types is really essential there. So making sure you’ve got the flexibility to identify that information that might have short term value and keep it for perhaps a shorter period of time, being able to identify information with medium and long term value, keep it perhaps for years rather than months or weeks, and even things that have a very long lifespan, even the lifetime of the institution. So being able to have that layered approach to retention, absolutely essential as especially as we enter this modern age of archiving in the age of AI. And with that, I’ll I’ll turn it over to Aaron. Yeah. Thank you, Cheryl. That was a really great market overview. If we can move forward here, I’ll share that one of the experiences I’ve had the benefit of of doing in my career is setting up managed services for some very, very large organizations. Tens of thousands, hundreds of thousands of folks having their communications archived, and the managed services that that I’ve set up are really around the human element in interacting with the archive and actually delivering on discover requests as those come in. And I think that here’s just, you know, kind of a a a view into some of the things that, you know, the the managed services at a day of life that an analyst would be doing on an ongoing basis. And I think Sheryl alluded to this in a couple different ways, but, you know, archives as such, particularly at larger institutions, they have a very, very long life cycle. It’s not unlike, you know, road infrastructure or electricity infrastructure. And over time, many of these archives were originally installed without a clear, to Cheryl’s point, last point that she made, a clear disposition strategy or a lot of thought put into what what we’re gonna need to do ten years from now. And as a result, some of these archives at bigger institutions have petabytes of data in them. And so even as folks have, you know, the legal profession and and, you know, records management have evolved and, gotten their arms around what are more reasonable retention, policies and and strategies, there have still been kinda gaps in in operationalizing some of that. And and the result of all these things, it’s it can be difficult to switch off of an established archive that has ten, some cases going on twenty years of data sitting in it. And so what that means is that in the in the day to day life of of someone in the trenches responding to discovery requests, they’re often working with technology that is at least ten years old, some cases twenty twenty years old. And so this has led, and and then you combine that, as Cheryl was pointing out, you combine that with the evolution in data sources. There is increasing pressure and expectation as she alluded to in the discovery process for, organizations to produce things like collaboration, teams, chats, SMS, these other, formats that didn’t exist, right, you know, twenty years ago in more, quote, unquote, native formats, right, so that the context is not lost, so that the receiving party can, understand better, some of the nuances, in the communications. And so this has led to a situation where there’s a ton of work to do, and there are many different kind of manual processes that have had to be put in place around the technology in order to satisfy these various requirements. And as just one simple example, you know, some some something I see pretty often is that, okay. Well, you know, these ten data sources all go into our main archive. Right? And we do all of our discovery on those ten data sources in that archive, but we’ve got these three data sources over here that we just can’t do that with. And we have to do some manual process and go to the original source and pull those out and put those through some completely separate process. And, of course, in a in a, you know, regulatory response or or a discovery request, there’s a a huge emphasis on chain of custody. And so having having to prove that you got the data you were supposed to get, you ran the search you were supposed to run, the search returned the result results you were supposed to get, and then you ultimately delivered those results itself is often a manual process that gets wrapped around with with ad hoc, you know, pieces. People have built specific reporting tools, to to help with with those things. And the the standard in in discovery, at least in US style discovery, is close to perfection. You know, there’s not a lot of leeway to be able to to have many errors. And and so the the upshot of all of this is that it it tends to take a very long time, sometimes just back to the technology. Sometimes it’s just because there’s such a huge volume of data. The technology wasn’t originally designed to be able to search such huge volumes. Searches take a really long time. Searches come back, and they’re incomplete. And so there’s, you know, there’s just challenges with meeting, meeting deadlines. It doesn’t scale particularly well. Some very large organizations, you know, again, I I set up managed services have, you know, dozens of people, you know, supporting, an active, you know, litigation or regulatory portfolio. So you need a lot of people. It doesn’t scale very well. And those processes are pro are brittle. You know, if when there’s so much manual effort involved, things can get missed. You know, there can be inaccuracies and so on. And if we can move on to the next slide for, again, someone in the trenches, a lot of this comes back to basic blocking and tackling. You’ve gotta be able to be sure that you got the data you’re supposed to get in the first place. Right? If your strategy is these ten data sources, I’m gonna do my discovery in the archive, but you’ve got a six month gap in in a in a particular feed, that’s a that’s a big problem both from a regulatory perspective as well as from a downstream discovery perspective. Again, completeness and accuracy of search results, not to pick out any particular, you know, platforms out there, but there are platforms where I have seen searches just not return results from particular datasets. Long term identity management as as people leave the firm. They change their names. You know, things happen over time. I I’m I know of organizations that have built entire processes around just being able to manage the identities from twenty years ago so that when they do a search, the analysts have to go and look up all the different IDs so that they can run those searches. Getting the data out quickly is critical, particularly in, you know, in regulatory responses. And and sometimes, you know, in even in civil discovery, you can get run up against the court deadline, and then you need to get that data out quickly, which continues to be a challenge. And then, again, having the ability to report and evidence all of the activity that’s taking place is critical. And so I think Sheryl did a really good job of of painting what the what the opportunity is now. Right? So I think the question becomes with twenty five plus years of experience with these types of requirements and, and sort of the benefit of learning from, you know, maybe not having a a coherent disposition strategy at the outset. How can we bring the latest and greatest technology? And and Bofira will go into this, but, you know, the shift to the cloud is a game changer because, as Cheryl mentioned, you can continue to evolve on that platform. And how can we use a a fully modernized stack to help the the analysts and the trenches and and the folks that are relying on them, the in house counsel, outside counsel, for example, to be able to do this in a much more efficient way with much more reliability and much more scale scalability. And with that, I’ll, turn it to Afir to give his thoughts. Thank you, Arun, for the insights on the daily life. I’m Ophir. I’m the CTO and cofounder of Shield. I’m going to focus on the modern archive and, of course, AI. And I would like to cover why moving to a modern archive is not only a technological transformation. It opens up opportunities for redefining and reimagining business workflows and much more. So when I’m looking at the market, I’m meeting all kinds of organizations. I’m seeing organizations that are still running on an infrastructure and an archive that was built fifteen to twenty years ago. And the processes around the archives are bound by the capabilities and, of course, the limitations of the technology stack fifteen to twenty years ago. If we’re looking at legacy system, one of the critical aspects is they all started as an email archive. But over the years, there were more and more data sources coming in, chat platforms like Slack, Teams, SMS, WhatsApp, WeChat, endless amount of new communication channels. But all those communication channels were converted back to email because this is what the legacy archive know knows how to work with. This created a fragmented data. Some archives are dealing only with electronic communications and other archive system for the files, additional one with a different practice for voice. It’s just not combined together. And it’s also great inconsistent formats and metadata. There’s a lot of metadata in the in each electronic communication that is relevant for use case for compliance use cases, but not only compliance use cases that we’re just losing in in this conversion to Yemen. Additional aspects of the data completeness. Twenty years ago, the the requirements were were completely different than what they are today. We’re seeing a lot of pressure around data completeness and data integrity in the past several years, and there is a limited visibility. And the last part is the user experience. The end user are experiencing manual and time consuming investigations, and the user experience of the EMS is quite painful for the end users to go over external communication record by record. So I obviously, you know, when you are moving from a legacy archive to a modern archive, you are going through three simultaneous shifts that happens in the course of the last fifteen years. The big data, the the ability to basically handle the ever growing amount of communication channels and volumes, the cloud which provides us the scalability and elasticity to handle those loads. And the last part, is the most exciting one these days, obviously, is the AI, which brings us extreme efficiency. And we will deep dive into it in the next slide. So let’s understand what is a legacy archive compared to a modern archive. So the modern archive, the first and most fundamental aspect is start looking at at the archive as a trusted data foundation for the whole organization compared to just a long term storage with retrieval capabilities. This is the core principle of the modern archive. And when we’re looking at the infrastructure, we’re moving from a passive static infrastructure, usually on prem and siloed into a unified place in the cloud, which is able to support multi regional data residency requirements. So you can have one archive that is able to store the US data in the US, the EU data in the EU, APAC Australia in in one archive and not have multiple deployments in each region. The data itself is we are moving away from flattened EML formats to unified, rich, and most importantly, future proof that we’ll able to sell more and more use cases in the future data model. And from there, the actionability is completely different. It now supports investigations, surveillance, AI pipelines, and advanced data analysis requirements, and again, future proof. And with that, the stakeholders change. It’s no longer only an IT and records management teams. We’re talking about compliance, legal surveillance, supervision in the US, and other use of external stakeholders. So now I would like to focus on the AI revolution. And what does it mean going from simple storage system to an agentic archive? When I’m talking about agentic archive, I mean extreme efficiency, not a marginal one. And this extreme efficiency is an opportunity for you to reimagine your business workflows for every operation that you do internally within your team, but also externally with the external stakeholders. The the the endless of opportunities these days with AI to reimagine and redefine those items. These these are only few examples in the core of the archiving work that AI agent can really provide this extreme efficiency. The first one is data integrity agent that continuously monitor data completeness, identify gaps, recommend on remediation. This is a very tedious and time consuming work these days by IT teams. The second one is the discovery agent. The whole record collection and deep dive and and and reviewing of all the electronic communication for discovery request, it can all be automated with AI agent, saving a lot a lot of days of work. The last one is an export agent. The export, obviously, can provide export packages with with advanced use cases and specific use cases for each export, which would you normally won’t have in a standard export capabilities. Those are only few items, but now I would like to drill down into one use case, which is the current search capabilities or search method. The current search method is keyword based. It’s syntax. It’s not semantics or context. Let’s understand the flow that we have today. It starts with a discovery request to find conversations of a specific topic. In this example, there’s an employee named John who is bullying another employee. Let’s assume this is the case. Then you would need to anticipate how this topic might appear in conversation. It it can appear differently in email compared to the language in the in chats, compared to spoken, in voice recording transcriptions. Maybe John is also Spanish speaking and is is bullying in another language, in Spanish or another language. So you would need to anticipate it in different languages now. And and the the language itself is different. English in the US is completely different than English in the Nordics and how it is used in communications, especially chat, which are informal. From there, you will need to identify all the relevant keywords and permutations, and you will apply wildcards and fuzzy matching, and the results will be too many irrelevant results and tedious work of of reviewing all those records that came up in the result set. And the next step would be to search within the search, refine the query, apply complex operators, reduce the noise. This is a very tedious work and, to be honest, a question about quality of results. And I would like to show you what we’ve done in Shield. We have introduced Sheila, which is AI assistant for discovery cases where the whole process is now completely different. You just need to ask where your query in natural language. In this case, show me all the records that John is bullying another person or show me any signs of profit from a specific stock. Whatever use case, you can just ask it ask it as is. And behind the scenes, AI will gather the information, review their communication, filter out what is not relevant, and show you what is relevant only based on relevancy rate. And we’re already seeing, as you can see, massive gain in efficiency from this process. Final thoughts. Think about the archive as your data foundation for the organization combined with the extreme efficiency of AI agents, it really opens up the opportunity to redefine your business workflows. My recommendation, find a vendor that can become your partner in this journey because it is a journey, and you will find unknowns, and you will face surprises in this process, and you will need a partner that is able to support you in this journey. For example, the migration, you will probably find some data and formats that you you didn’t know exist from twenty years ago. And, of course, the AI approval process is still evolving, and you need a vendor and a partner that can help you evolve and meet those requirements. Thank you. Thank you so much, Ophir, Aaron, and Cheryl, for your insights on the market today, what it looks like in practice, and and what it means from an infrastructure perspective. We’re going to move on now to the interactive part of the webinar. So we ask you, the audience, to share your burning questions with us, and we’ve cherry picked a few of them to ask our experts to address today. So let’s jump in. We got this question from head of market surveillance at a global bank. And the question is, many vendors mentioned AI during their presentations to us. How do you see AI playing a factor in archiving? So, Cheryl, if you could get us started off with your thoughts on this. Sure. Sure. Yeah. No. As as I mentioned, I think, AI has the opportunity to transform, the way that we engage with information, you know, whether it’s email content, other kind of collaborative artifacts, really across that whole life cycle. So even from that moment of ingestion, the ability for AI to help enrich that metadata, extract insights perhaps, you know, do do some validation checking, make sure that we’ve got everything from the, from the source application. So, you know, helping us, you know, validate the the information coming in is is is accurate and able to be searched and engaged with. From a user point of view, anybody who’s engaging with the archive, certainly the ability to use natural language to query and interrogate it. And then as mentioned, I think also that opportunity for agents to start automating a lot of the tasks and looking more broadly than, you know, what Humanize alone could look for, especially with things like investigations and supervision. Okay. Wonderful. Ophir, Aaron, do you have any thoughts you wanted to add about how how vendors can be making it clearer the role of AI archiving for their firms? Yeah. I mean, I think I think the agentic development, right, is potentially really huge. I think I think we’ll probably see as as archives as modern archives enable these types of capabilities, I think that we’ll probably just we’ll see movement around QA agents in the short term, sooner than we’ll see, you know, agents helping us just respond to discovery requests, let’s say. Now that’s not to say that there’s not a really great place for, like, a Sheila in kind of, an internal investigation, let’s say, context where you’re just trying to get your arms around what’s happening. What what are you know, what what is this complaint about John. Right? And and so I think in the short term, there’s a huge opportunity there, because often, depending depending on what the issue is, right, there’s a there’s a lot of time pressure in an internal investigation scenario, maybe not the John one, but, you know, maybe there’s an allegation that, you know, the CEO is having an affair. You know? So, you know, some these things that’s gonna go to a board, like, a board level. Right? And then there’s, like, there’s a lot of time pressure to understand, okay. What’s in the communications? What is the risk? And what are we dealing with? And I think a a Sheila type agent, it provides tremendous value in that type of context. And the opportunity in of the agentic is also a business opportunity for you to reimagine your roles where you have agents working for you. There is a difference between I’m doing my job and I get some help with AI for some of the task and then compared to let’s reimagine the role if someone if there was an agent that was doing my tedious work and leaving me free to deal with the real stuff or real challenges. Yeah. And I think that also links back to what Cheryl showed us with the graph as to how people are using AI right now that for the most part, they’re kind of using it as this helper to, you know, help help find things in that the future is going to be those agents who are You know, working working as a team for the person. My my my point is don’t wait for the vendors to pitch AI for you. Reimagine what you would like to get from AI and then have it in a requirement for the vendor because you know your life and business workflows better than than us. Yeah. I love that. Pitch to the vendor. Don’t let them pitch to you. Okay. Great. The next question that we received, this one is from a senior compliance officer at a tier one bank. And, also, Sharon Sheryl talked about this a bit earlier. They’re asking, as employees use tools like ChatGPT or Copilot, how are firms approaching the archive and supervision of those interactions today? So, Aaron, get us started with your thoughts on this. Sure. I I think this is very definitely a hot topic right now. You know, it’s I started first hearing about this in in kind of customer conversations probably about a year ago as as actual, you know, business adoption, you know, really ramped up. Right? Of course, we’re, you know, two plus years into the into the LLM revolution, but, you know, it’s it’s taken some time for for firms to actually start to adopt these things for internal use. But as they’ve done that, they’ve started to think about exactly this topic. And and one of the first use cases I came across was simply, you know, a a data a data access and and policy compliance, you know, approach, a use case, which is are people trying to use ChatGPT, let’s say, to discover information in the within the organization that they wouldn’t otherwise have access to. Right? Like, in the old world of file shares and intranets and and all these other things, you know, you you couldn’t just go to the intranet and find anything you wanted if you weren’t supposed to find whatever that might be. And, you know, there’s a there’s a risk that now if you’re if you’re having, you know, a GPT crawl all these different data sources, are those bear barriers being maintained? And so the use case that this particular organization was was first looking at was, well, can I just from a surveillance perspective, can I just figure out if people are looking for things, if they’re violating policy and looking for things they’re not supposed to be looking for? Now that’s evolved, and I think a lot of what I hear, more recently in conversations is is yeah. You know, are these communications and thus subject to, you know, SEC seventeen eight four type regulations. Right? And I that’s an ongoing debate. Right? I think maybe the prevailing viewpoint I’m hearing at the moment is, look. As long as it’s just me and the and the GPT, not a communication. Right? But as soon as I share that communication with someone else or I, you know, share the the chat because, you know, at least I have the most experience with ChatGPT OpenAI. So, you know, you can share you can share a chat that you’re having with someone else. And then and then, you know, the viewpoint is, well, now that is a communication, and now it does need to be captured. So I think I think there’s a there’s kind of a and and we’ve seen this play out, you know, many times in in my career when you have a kind of a new data source, a novel data source. You know, there’s a there’s a natural approach to to wanna try to resist having to do more. Right? This is just another source that we have to track. We have to archive. We have to discover, etcetera. And so people are looking for ways to sort of try to avoid that burden. But I do think that there are aspects to this which are gonna be unavoidable. So Yeah. I think that’s a really interesting point that you raise about whether you’re sharing it with a person or just chatting to the GPT itself. Because at the point at which it’s just you and the AI, it’s essentially thoughts to yourself that are being written on paper and and and whether we’re identifying the AI as a as a sentient other person on the other side or whether it’s just you thinking to yourself is a is a key definition. Well, right. And yeah. And sorry to just elaborate one more point. I think then but then you can you can kinda play this out in other ways. Right? So if it’s if it’s actually an agent, right And it’s doing stuff, well, is that still really just me talking to myself? Because now it’s actually doing things, you know, in the business that an employee would do. Right? So it’s it’s it’s very far from a a cut and dried conclusion, I think, and it’ll be really interesting to see how this this plays out over time. I think we’re seeing early adopters because I’m they’re taking a risk based approach. They are considering the Copilot interaction the user interaction with Copilot as a record, and some of them and we’ve seen it more and more are considering them as a record and basically basically archiving it as an additional data source in in the archive. But we’ll definitely see how it evolves. Yeah. I mean, ultimately, you know, if it’s electronically stored information, you know, it’s potentially discoverable. So, you know, even if it is that note to yourself, you know, it’s, if it’s if it’s, you know, in writing, you it’s it’s kind of your rationale behind the decision or why you’re making a particular business decision, then it it potentially, you know, is discoverable. Oh, yeah. As as a requesting party, I would be after this stuff all all the time. Right? If you think about the criminal context, right, I think there’s really well publicized cases. You know, I think, you know, AWS gets, like, subpoenas all the time, for, you know, for the the the the chat device, you know, the Alexa, you know, for because it’s it’s evidence in in a in a crime, you know, for example. Yeah. Yeah. Sure. Okay. We’re moving on to the next question. From what you’re seeing, how comfortable are regulators today with AI playing a role in investigations? And where do they still expect manual validation or supporting evidence? I can take a a first cut at this. So, you know, in my experience at least, regulators tend to be pretty outcome driven in the sense that if you provide the information that they ask for and there are no obvious gaps and over the course of the the matter, you know, gaps don’t come to light, then they’re generally satisfied. You know, it’s not to say that they’re not concerned with the underlying processes, but, you know, again, they tend to be more outcome driven, in my experience at least. Where I think this gets potentially more interesting is in civil litigation because in civil litigation, there is an incentive for the requesting party to try to dig into the discovery and understand and pick at it. And and and so if if if you’re going to be representing as a responding party that you’ve used AI to find the information, I would fully expect to be questioned on that. But that’s that’s kind of my hot take there. The solution would probably to add a human in the loop or AI that provides you additional reports that later on you can audit the behavior of the AI and the decision making behind the scene as needed. It’s kind of a complementary to what you said, Aaron, in order to gain more confidence and also approve it internally. Sometimes I see the challenge is to approve the use of AI internally with the with our internal auditors in in the bank even before it goes to the regulator. Yeah. I mean, just, you know, building on the point a little bit, I think, you know, that, you know, the standard of care, if you will, right, in in US civil litigation is is often that, you know, we’re gonna negotiate search terms, boolean good old boolean search terms may be pretty complex, may be long phrases in any given term. And we’re gonna as the requesting party, we’re gonna send that over to you, and you’re gonna give that back to us. And right or wrong, better or worse, you know, that’s that’s a technology that everyone has a fairly good understanding of and, you know, leaving aside questions of, well, is it properly indexed? Are you know, have is the index corrupt? Are are the results actually be like, people are relatively comfortable with that. When that turns into, okay. I’m gonna I’m gonna give you give you these search terms, and then you’re gonna use an LLM to turn those into into a natural language query. You know, that’s where it starts to get, you know, interesting. And I think there will be, you know, a lot of a lot of, you know, internal discussion, right, around that. And that kinda goes back to one of the things I said before and and the AI approval that you were just mentioning. Right? So if I’m if I may announce, you know, eDiscovery Council, you know, it’s my view. You know, I would probably be relatively comfortable saying, okay. I just need to figure out what’s in this dataset, and I can use, you know, an agent for that. I can use an LLM for that. But I’m going to if if I’m responding to a Discover request, I’m look. Out of the gate, I would probably prefer to still use the the keywords because, again, that’s that’s known. I don’t I’m not gonna have to get an argument about it most likely. But, you know, to the point, as the data volumes grow and so forth, maybe I I look at adopting that, but I’m gonna need some validation. Right? I’m gonna need some validation to say that this is the true and correct data set. Okay. Great. We’ve got one last question here that we’ve received from a head of compliance. It’s an international bank. How do you and I’ll open this one up to the forum. So how do you balance out strict retention policies with the need to keep data usable for investigations and AI models? I think this is gonna be an interesting area of kind of reflection and debate, you know, as we really embrace the age of AI. You know, I’ve often seen that there’s a bit of a tension, between, you know, the school of thought is, you know, keep everything, you know, for a long period of time, you know, but you know, kind of or have a very compressed, life cycle management to, you know, dispose of things so you don’t have to go through large scale, you know, multiyear discovery and so on. So I I’ve seen kind of those two kind of polar extremes in in in several organizations. But AI does bring a a new opportunity to engage with and and use that that corporate memory, you wanna think of your archive that way, in a lot of different ways. And so, you know, for organizations that have had a very, aggressive, you know, deletion cycle, it it means that they essentially have holes in their corporate memory. So I I think this is gonna be a really interesting debate, and we’re gonna need the point of view from both the, the records management, information management point of view, as well as, you know, knowledge managers and AI experts in terms of what that rate balance is gonna be. And it may it may vary, you know, from from sector to sector and and organization to organization. Yeah. I would add what’s up, please? Continue. Thanks. Yeah. I think, you know, I think I think the basics, right, at least in a kind of a US regulatory and and and, discovery context, you know, haven’t changed. Right? I think, you know, you have an obligation to retain what you are obligated to retain under various regulations. You have an obligation to, you know, preserve information for for potential litigation. And beyond that, you’re gonna retain information that’s useful to you as a business. Now where the huge opportunity is now, and we’ve touched on, you know, classification agents, for example. Right? Like, I think the the problem in operationalizing that and why many large and even small organizations as well have essentially kind of, you know, thrown up their hands over the past several years and said, well, we’re gonna we’re gonna keep everything forever because we can’t we can’t actually operationalize the classification to be able to make those decisions. Right? And so it’s it’s better for us to risk, you know, maybe overproducing and and and having stuff that we could discover that may have reputational risk and other things like that. It’s it’s better for us to live with that risk than to live with the risk of a discovery sanction because we deleted something that we were we didn’t know we were supposed to have or what have it. And so I think and and, you know, you know, I think, Cheryl, you’ve you’ve lived a bunch of this. Right? It’s like, yeah. You know, you you try to set up a a program where people classify all the you know, this is this type of business record. This is that type of business record. This is a broker dealer communication. This is right? People have never done that. That type of that type of approach has never worked, but this is a great job for agents. Right? Agents can do that job. You know, I think the challenge will, of course, be just be comfortable that they’re doing it correctly, but I think that’s a massive opportunity. I I would add that there’s another limitation maybe today, which is scale and cost. Sometimes this is kind of a showstopper for having it in such a large scale deployment in terms of volumes of data. And although we don’t hear a lot about it lately because everything is AI, they’re in parallel. There are continuous there’s continuous innovation in the in big data, in querying big data and applying AI on big data in the cloud. So we have if the if we are adopting new technologies, we have the ability to basically search with AI, apply AI on massive amount of data in a reasonable at a reasonable cost. So it’s something that happens in parallel. Okay. Wonderful. And we’re actually coming to time now on the webinar. And so I want to take this opportunity to thanks to thank our wonderful panelists. This has been a great discussion. It’s been incredibly informative, and I think it just goes to prove that archiving can be a lively and innovative topic. So if I’m looking to sum up one main takeaway for our viewers, I think it’s that archiving is no longer just about retaining data. It’s about being able to trust it, to be able to use it, and to be able to act on it. And the foundation for all of that is the quality and completeness of your data. Because without it, firms can’t move fast enough. They can’t investigate effectively, and it just won’t stand up to regulatory scrutiny. So that’s why this shift that we’ve discussed discussed today matters. It’s about turning the archive from passive storage into something that actively supports investigations, decisions, and real business workflows. And so on that note, thank you once again to our speakers and to everyone who joined today.