While advances (and mishaps) involving generative AI create nonstop legal industry headlines, the infrastructure powering how legal professionals find, use, and govern their knowledge is having a moment.
That’s because a knowledge work platform is the foundation for even the flashiest new tools. Maintaining a robust system at the core of your business provides the data access and governance that enables success with emerging technologies.
This is particularly good news for law firms and legal departments, according to iManage CEO Neil Araujo, because most of these organizations have been investing in how they manage their knowledge for decades.
As a result, they are now uniquely positioned to lead the next wave of innovation.
“The one thing that we have always said is that good information management is a cornerstone for whatever might come next,” says Araujo, who co-founded the company during the 1990s. “And that’s certainly held true, and it’s actually been magnified in the AI world.”
Araujo notes that the demands on the knowledge work platform are set to grow exponentially as AI changes how people work. This means the platform will need to perform under significantly higher load, and the underlying architecture to support that scale, reliability, and performance will become critical.
The iManage platform has been reimagined for the AI era — for both humans and agents — boosting its capabilities in these specific areas when compared with legacy systems, the company says.
Here, Araujo sits down with Above the Law to discuss how the document management system (DMS) has evolved into a knowledge work platform — and why that platform is now the foundation for AI in legal.
This interview has been edited for length and clarity.
ATL: To echo the start of many legal tech conversations these days – I’ll begin by noting that we’re in the middle of one of the biggest stretches of disruption. What changes have you seen for your customers over the past five years?
NA: Well, I think it’s become increasingly clear that this technology can have a profound impact on the quality, effectiveness, and efficiency of legal work, and that organizations are going to have to rewire how they operate to take advantage of it. They’ll also have to reconfigure their business model to align with how they’re providing value — rather than time being the only proxy.
I also think there’s a lot of experimentation that’s been happening over the last couple of years, and we are beginning to see more and more of that experimentation mature into: ‘OK this is what we can use it for on a sustainable basis.’ We’ve certainly not seen the next stage, which is: ‘OK, what is the economic impact, at the end of the day, of doing all of this?’
I don’t see a lot of organizations at that stage yet. But what has become very evident — to us and to our customers — is that the effectiveness of LLMs is significantly higher when you combine them with good data and good signals, rather than using the LLM on its own. Some of this is obvious, and some of it is subtle. LLMs are very good at picking up signals when they’re available in the data and catching the nuance that makes a difference in the response.
The good news for most law firms and legal departments is they’ve had discipline around managing their data going back decades. They’ve had to do it for governance reasons, for improving collaboration. Particularly when COVID hit and remote working meant you had to find a sort of electronic place where people could share information. I think all of that has come back to be a positive tailwind for the era of AI.
What uses do you see emerging here?
We are seeing firms think more about not just capturing the data, but data quality. You know, ‘How do I identify my best work?’ ‘How do I capture the signals that go beyond the document itself so that I’m providing these hints that a good LLM can pick up to create better outputs?’
We’ve not necessarily had a lot of demand from customers for what I’d call ‘generation use cases’ — drafting and content creation — because of the risk of hallucinations. It’s certainly not been where our focus has been. We’ve seen a lot more on the research side, you know, ‘How can I find the golden nugget that was impossible to get to?’ ‘How can I find what’s trending in my own data?’ Because law firms have a lot of data that helps lawyers negotiate better on their clients’ behalf.
So, a big focus for us has been: How do you enrich data so you’re capturing those signals that can make AI more effective?
An example could be jurisdiction. It’s really important to know which state’s law a document is referring to, because a clause valid in one state may not be valid in another. Typically, jurisdiction wasn’t captured in any structured way — just inferred from language buried in the document. Now we’ve built custom models at iManage that identify the jurisdiction, trained specifically for that one task.
And so now you can say, you know, ‘Find me all shareholders agreements where the governing law is the state of Texas,’ and you can ask a subsequent question: ‘OK, what was the termination clause for agreements that were generated in that particular jurisdiction?’
So we’ve gone from capturing data to capturing data plus signals that inform. We’ve seen a lot of interest now, from both legal departments as well as law firms, to invest in this foundation because they understand that your data foundation, plus an LLM, plus how you change your process is what gives you AI outcomes.
This sounds like a good segue into the next topic I was going to ask you about — the role of the knowledge work platform in a firm that’s taking on this technology. You touched on it just now: better research, stronger information governance, a single home for a distributed workforce. Could you talk a little more about its strategic role as it relates to new technology?
At its core, the knowledge work platform helps organize your information and secure it, right? So, only the right people can see it, and it can be found easily by the folks that need to use that information. As knowledge work platforms have matured, it has moved beyond organizing and securing content to activating it.
Increasingly, the ‘folks that need to use that information’ means ‘AI agent.’ It’s an entity acting on behalf of a human and operating at enormous speed. So you’ve got to have stuff in the right place, and you have to have it described properly. Otherwise, you’re going to have chaos.
In that world, it has gone from being something you had to use for governance and compliance reasons to something you’ve got to use just to function in an AI-enabled world.
I’ve not seen a report on this yet, but we are investigating that the effectiveness of an LLM with good data is significantly higher. I don’t know by how much, and I don’t know if we are quantifying it right now, but we certainly have ample anecdotal evidence that that’s the case. Our view is: The knowledge work platform plus LLM is two plus two equals 10, not two plus two equals four.
You mention the rise of AI agents and that law firms are going to be increasingly bringing these on. Could you share a little bit of guidance you’d have for firms starting to implement this tech, based on what you’ve seen?
Agents can be a huge productivity driver. I think there’s significant investment you’re going to see — from iManage as well as the infrastructure providers like Microsoft — in ‘How do you manage agents, and how do you secure them? How do you monitor their usage and what they’re doing, what data they’re accessing? Where are they succeeding? Where are they failing?’
It’s going to be increasingly easy for organizations to create these agents, not just for the tech people, but for the end users themselves, to create an agent to do something. A tech-savvy lawyer may just go into Cowork, for example, and whip up an agent for a specific task.
I think having the right guardrails around the creation, use, monitoring, and data access of these agents will be critical for this technology to scale within an organization. And that’s a space you’re going to see significant investment in — from iManage as well as the broader infrastructure ecosystem.
I hesitate to predict the future in a world that’s changing so fast, because typically when folks like me have predicted the future so far, people eventually have forgotten what we said. But in today’s world, you know, you turn around and the prediction is already becoming a reality.
I’m curious to hear if there’s anything else you’d have to say about how a knowledge work platform could specifically be used to roll out agentic AI within an organization.
Sure, I’ll tell you how we are doing it. There’s sort of two parts to this, one which is already available to customers, and it’s already having a significant impact. I can tell you it’s impacting my own usage of the platform.
The first example of what we’ve done with agentic AI is for search. Think about what you would do if you had an assistant who you wanted to find a particular file or a set of files.
You’d give them some broad instructions, they’d search through everything available, and say, ‘OK, I found 10 of these, but these seem overwhelming. Let me try to refine it further and get to just the one or two documents that are most responsive to what you’re looking for.’
So, we’ve built an agentic search layer into Ask iManage, which is our legal AI assistant. What it does is it takes natural language queries, and it will use an LLM to create the search. It will also look at how the search is performing — is it giving back results that the LLM is expecting? If not, it will tweak the search query to either get more or get less.
Then it takes the results set, looks at the results set, and gives you a summary of an answer, in addition to the list of documents. So instead of seeing a list of 5,000 that you have to scroll through, you get a much smaller set back, and you get back an answer to the question that you were asking. For me it’s transformational in my usage of the platform.
I’ll give you a very trivial example that happened just last week. I was doing an interview and I needed a bio. If I go and search ‘bio,’ I get, I don’t know, 200 results. And I went to ask and said, ‘I’m looking for Neil’s recent bio.’
The search prioritized what [an iManage communications professional] had done last year. It also saw that it was the most recent, placed it at the top, and then gave me a bio that I just had to cut and paste. I didn’t even have to open the documents.
Think about someone looking for a clause, or looking for an exception to a particular clause that is typically in every single agreement. You could ask those types of questions and get responses literally in seconds.
That’s one example of how we are leveraging agents for search, and that’s embedded under the platform itself. The second half of it is that we’ve developed what’s called an MCP protocol.
You know how APIs are what let applications talk to each other. MCP is sort of the equivalent for agents. It is the standard that gives them a consistent way to connect to a system without custom connectors. So, you can describe what you want in plain language, the AI figures out what to do, and through MCP it can go off and interact with your knowledge work platform and take action. And you can build very complex workflows that go across applications.
For example, let’s say you’ve written an agent to look at all of the leases you have for a particular landlord? Tabulate who the tenants are, tabulate for each tenant what the termination date is, and generate an email for anyone who’s terminating in the next one month.
You could get an agent to do that. And through MCP, all that automation is possible without a single line of code. You could basically go into Cowork, and we build the plumbing to make all that a reality. I would say those are two examples of how we are enabling agent workflows.
Are there any other areas where law firms and in-house legal departments can ensure they’re getting maximum business value from their knowledge assets?
The knowledge work platform is a supporting function, right? The benefit you’re going get from all of these technologies is: When you fundamentally take a look at your process and figure out how that process gets redone in an AI or in an agentic AI world, then go back to, ‘OK, what do I need from my data to enable that process to work?’
So my advice always is to start with the business process and the economic impact that you’re trying to drive from that process, and then work your way backwards to how the data and the large language model come together to solve the problem. It sounds obvious when I say it that way, but you can get by with experiments without building that type of plumbing. But if you want to scale it out, you need to sort of think about it that way.
You’ve co-founded iManage roughly 30 years ago and been a part of the company ever since. You said earlier you don’t like to make predictions about the future, but looking back, could you have ever pictured the way the legal industry’s operating today?
That’s a great question. Look, there was no way we could have predicted how technology has evolved.
The bit that’s not changed for us is we are an extremely customer-focused organization, and our values have remained consistent. We care a lot about the outcomes that our customers are driving. We are very proud of the impact we have for the end users that we serve.
And I think the value that we provide has grown exponentially in that 30 years. So we feel blessed and fortunate that we are in that position and feel like we can drive even greater impact in the years to come. It’s a mixture of optimism and humility, I would say, that drives us. There’s still a lot to learn, and we were not smart enough to predict this.
The one thing that we have always said is that good information management is a cornerstone for whatever might come next. And that’s certainly held true, and it’s actually been magnified in the AI world.
Any parting thoughts?
The single biggest change that we are feeling, and I’m pretty sure that the economy in general is going to feel, is the accelerating pace at which the world is going to change. You’re going to see new capabilities, new ways to solve problems, new types of problems that folks are solving. It’s exciting and daunting at the same time. I am certain we will see productivity gains, I hope we channel some of the productivity gains to do more and drive greater impact for our clients and society in general.
The way you keep pace is by focusing on the fundamentals — making sure you’ve got the right foundations in place. That’s going to determine the speed at which your organization can move. And I’m a strong believer that without good information governance practices, you’ll be slowed down in a world that’s pushing you to move faster and faster. The real question is: ‘Am I ready for change at the speed at which it’s going to come?’