The legal tech market is expanding quickly, but is it a bubble—or the beginning of a fundamentally different era for legal services? In this episode, Jen Leonard and Bridget McCormack are joined by Nikki Shaver, co-founder and CEO of Legaltech Hub, to discuss how law firms, corporate legal departments, and legal tech companies are navigating the next phase of AI adoption.
The conversation explores how the legal tech market has shifted since the launch of ChatGPT, from experimentation and piloting to licensing, ROI scrutiny, pricing pressure, and strategic investment. Nikki explains why firms are moving beyond efficiency as the primary measure of AI value, how AI-native firms and legal tech startups may disrupt incumbent firms, and why differentiation will depend on combining firm expertise, data, client relationships, and carefully chosen technology investments. The episode also examines agentic AI, governance risk, MCP, and what may define the legal tech market in the year ahead.
Key Takeaways
- AI value is moving beyond efficiency: Law firms are beginning to measure AI not only by speed or time savings, but by quality, client outcomes, effectiveness of advice, and the ability to deliver more value.
- Pricing conversations are becoming unavoidable: As AI changes how long legal work takes, firms are increasingly focused on revaluing legal work and rethinking how pricing fits into their business models.
- AI-native firms may challenge incumbents: New firms built around AI from the start—and legal tech companies launching service models—could create real pressure for traditional law firms, especially as the innovator’s dilemma takes hold.
- Technology selection must start with the problem: With legal tech startups proliferating, firms need to resist chasing shiny tools and instead focus on specific use cases, system requirements, integrations, jurisdictions, languages, and client needs.
- Differentiation will come from expertise and data: AI enablement is becoming table stakes. The firms that stand out will be those that build around their strongest practice areas, embed AI into client service, and deliver differentiated value at strong margins.
- Agentic AI raises new governance risks: As AI agents enter legal workflows, firms need to revisit security reviews, governance frameworks, acceptable use policies, agent identifiers, and how autonomous systems access data and tools.
Final Thoughts
Legal tech is no longer a support function at the edge of law firm strategy. AI has made technology central to how firms compete, price, deliver work, and serve clients. As the market continues to expand—and consolidate—law firms that combine strategic clarity, disciplined experimentation, strong governance, and deep client alignment will be best positioned to navigate the next phase of legal innovation.
Transcript
Intro + AI Aha!
Jen Leonard: Hello everyone, and welcome to this episode of AI and the Future of Law. I’m Jen Leonard, founder of Creative Lawyers, here as always with my favorite collaborator, Bridget McCormack, president and CEO of the American Arbitration Association.
How are you, Bridget?
Bridget McCormack: I’m well. Good to see you.
Jen Leonard: Great to see you.
We are really excited today to be joined by one of the preeminent experts in legal tech, Nikki Shaver, who is co-founder and CEO of Legaltech Hub, the go-to resource for law firms and corporate legal departments evaluating their legal tech solutions—very relevant these days, and front of mind for all law firm and corporate leaders.
Nikki co-launched her company in 2019 because she saw an absence of a centralized, searchable resource for legal tech. She has done a lot of other things in her career spanning literature scholarship, and she was a defamation litigator early on in her legal career.
She has also held knowledge management and innovation roles at major firms around the world. She teaches a course on AI and legal tech at Cardozo Law School and has been recognized by Fastcase 50 as one of the legal industry’s most courageous innovators.
Thank you so much for joining us, Nikki. We’re thrilled to have you.
Nikki Shaver: Well thank you for having me and the very kind introduction. It’s lovely to be here.
Jen Leonard: I don’t know if you remember this, but you visited our class one time, and I asked ChatGPT to put together a bio of you.
When we read it, you said it was correct except for one fact about where you had practiced law. And that was the one fact I had added on my own because I thought, from memory, that you had worked there.
So as Bridget would frequently say, we humans hallucinate.
Nikki Shaver: That’s exactly right.
Jen Leonard: Nikki, every episode of our show, we have an in-depth interview with our guest about their expertise. But we also like to kick off with what we call an AI Aha!, which is something you’re using AI for in your life—professionally, personally, whatever—that you think is particularly interesting. So we’d love to hear your AI Aha!
Nikki Shaver: I love the fact that you ask this question. I think it’s so clever, and I’ll have to go back and listen to everyone else’s AI Aha!s. I have a couple for you.
My first experience with Claude Cowork was when I hooked it up to my Gmail account. I think a lot of other moms did the same thing. I had it gather all of the emails from my kids’ schools—they go to different schools—over the past month, and then provide me with a short list of the things I actually had to know and any action items.
Anyone who has kids at school knows that you get a gazillion emails and texts, and it can be very difficult to keep up. So that was super helpful. Then I had it create an alert so that now I can get a regular list of things I need to know at the end of a week, rather than having to keep track on a daily basis. So that’s one AI Aha! that I have for you.
Another interesting one is that I recently found, through vibe coding quite easily, that you can create a fairly complex widget with filters and all kinds of things across an Excel spreadsheet, which can actually be hosted on a web page and create a really interactive way for users and other people to see and find data on a relevant topic.
So anyone who has been keeping a long spreadsheet of something they think would be interesting to other people—where it makes sense to have a subsection list with filters across it—can use Claude Cowork for that as well.
How the Legal Tech Market Has Changed
Bridget McCormack: Well, Nikki, we’re so excited to talk to you today. I feel like there’s nobody more knowledgeable about the legal tech market, so it’s really incredible that you joined us. You’ve been tracking legal tech for Legaltech Hub for seven years. I’m interested in the most recent few years—that post-ChatGPT era.
When you compare conversations you were having in early 2024 to the ones you’re having now, in May of 2026, what feels different? What are the big differences that you’re seeing?
Nikki Shaver: There are a few real differences. It’s been very interesting. Since ChatGPT launched, we’ve seen waves of adoption and behavior in firms in particular. In 2023, obviously, it was all about experimentation—figuring out what the technology could actually do, what the use cases were, and whether there were use cases at all. That was an initial question.
We then moved into a phase of heavy piloting of technology platforms, with firms often piloting multiple platforms simultaneously. And that’s still happening. But then we moved into more of a licensing mode—short-term licensing or renting of technology—because the tech is still evolving at such a fast pace that people don’t want to make a significant long-term commitment. That also continues.
But the real difference now, in 2026, is that a lot of firms have been at it for a number of years and are looking at ROI, and the shape of the ROI argument, in a different way. In 2024 and 2025, the focus around ROI was really on efficiency gains.
But I think there’s a shift now—a recognition that while efficiency is great, and there is value in being able to act faster, there is more value in quality: the quality of output, how you can increase the outcome for your clients, how you can increase the effectiveness of the advice for your clients, and how you can give them more. And how do you measure that? What’s the metric around that? So that’s one very significant difference.
The other, I would say, is more seriousness from firms around pricing and how to handle that. Ironically, at the time when everyone was super focused on efficiency, that was also the time when a lot of firms backed away from having serious conversations about whether legal work needed to be repriced—even while they were deploying something they knew would decrease the time it would take.
But in 2025, around March, we started noticing that managing partners were really focused on having those challenging conversations: thinking about pricing, thinking about revaluing legal work. And that continues even more seriously in 2026. So people are very focused on the value problem, but in two different ways.
Jen Leonard: Nikki, I wanted to follow up on something you talked about, which is the investment that firms are being asked to make today in technology that they’re not always confident in how they’ll use it.
Do you think the major platforms that are more general-purpose in nature, but legal in their domain specificity, the cost will come down in the years ahead? And if so, why? And if not, why not?
Nikki Shaver: First of all, I think there has been some bad information out there in the market. A couple of years ago, someone said the cost of Harvey was something like $1,200 a month. That was never the cost of a seat. Generally speaking, for these platforms, it’s more like $200 to $300 per month, per user. And that’s obviously negotiated.
When you look at the actual token cost from using something like Anthropic in the same way you would use something like a Harvey or Legora or CoCounsel, it is outrageous. Our legal workflows are super document-heavy and super language-intensive. And that means when you put together an incoming and outgoing token cost, it is extremely significant. Costs escalate significantly.
So you’re actually getting away with something by using the legal applications, because at the moment they take on the negotiation with the frontier model labs. Your token use, therefore, is kind of unlimited within those platforms, within your license.
I know of firms at the moment, for example, that are also using Claude for some build. But in order to save money, they’re taking some of that and running it through Harvey, because it’s cheaper to do that than to use Claude. So that’s one thing.
The other thing is that firms are making a huge investment, but they want to make that investment. AI is a massive strategic focus for law firms.
Prior to 2023, if you had asked managing partners what their top three strategic concerns were, IT or technology would not have been one of them. IT was typically seen as foundational support. Now it’s seen as hugely strategic, which means managing partners are willing and eager to invest because they see it as part of their delivery model for clients.
So I think the key question is more: how will the investment over time make sense within the firm’s business model, and how will firms be able to accommodate the ongoing investment? Especially because the investment is not just in the technology. It’s also in the resources and the people you need around it, which are very, very significant.
Given the risks we are going to see from AI—and already see—not just in the AI that we deploy in our own environments, but also in what we get from other sides, and the need to review it and deal with it, that is such a significant threat that you want to have a platform and provider you can really trust.
You want to know that you’ll be able to use it in a way that provides the kind of robust security you need and your clients need. So will token costs come down over time? I’m sure they will. Will the investment in those platforms overall come down? I’m not sure. Will we accommodate the way we price and value legal work to allow for the cost of our investment on the tech side and the people side around it? Almost certainly.
The Innovator’s Dilemma and AI-Native Firms
Jen Leonard: You recently gave the keynote at a LawDroid conference, Nikki. And shout out to our prior guests, Tom Martin and Sateesh Nori, who are doing great work at LawDroid.
You talked in your keynote about how the legal profession is entering the “innovator’s dilemma” zone, where incumbent firms are thriving. Many of them are coming off the best financial years they’ve ever had.
At the same time, AI-native firms are starting to scale their businesses around the law firm.
So what are you seeing on the ground? You have such a great bird’s-eye view of what’s actually happening across all these different segments. Who are the disruptors you’re watching, and what are you seeing?
Nikki Shaver: I’m definitely watching the AI-native firms. It’s really hard at the moment to tell who the disruptors are going to be, and whether they are, in fact, going to be the AI-native firms.
There are different types of these firms. Some of them are more AI-first. They’ve just launched with a business model that puts AI and tech at the center of it, which is different from incumbent firms that are trying to move toward that but haven’t had it from inception. So it’s much easier to scale in an environment where you’ve done that from the very start.
Some of these AI-native firms that really run on AI come from quite interesting places. An example is AI-native firms that have been launched by technology providers. This is a company, for example, that starts out as a product company and then launches a law firm that is services-based, but runs on the underlying AI platform that they initially created.
That is such a shift in the way legal services are delivered and the players in that market that you could see it as being really quite disruptive. The key, of course, for the innovator’s dilemma is that as law firms see their bottom line continue to grow, there’s no urgency for them to shift the way they’re doing things. But it’s exactly at that moment where disruptors can really come up and take hold.
There are a couple of examples. Eudia Law was obviously launched by Eudia, or Eudia Counsel. I think the model of General Legal—which was founded by some very experienced founders from legal tech—is quite interesting, where they’re automating about 80% of the law and then sending the top-cream legal work out to top-tier law firms.
So they’re a client of those firms, but also leveraging them to disrupt them at the same time. That, to me, is really fascinating.
Bridget McCormack: It seems like it would be a very hard time to be a very successful incumbent firm. For the reasons you just said, it’s really hard for incumbents that are doing well to make any kind of significant shift to face down an innovator’s dilemma. But it’s also such a hectic marketplace.
I feel like I meet a new founder in legal tech every day. There’s somebody who wants advice, and they’re building a thing, and I think, “Oh, I talked to somebody last week who is also building that thing.” And they’re all really impressive. The number of founders in legal tech right now is really stunning.
How are you advising, or how would you advise, an incumbent firm to make sense of all of the startups out there fighting for tech spend? What’s signal? What’s noise? How do you sort it out? Or are we not going to know for a couple of years?
Nikki Shaver: It is so hard at the moment. I really do feel for the people making those decisions.
On the other hand, I guess it’s not that different from before. The volume is higher, but the way you approach it is the same. The problem is people get distracted because there is all of this shiny new stuff, and there is so much of it. It’s easy to think, “We need the latest and the greatest.”
For me, the advice I always have is that it still comes back to the problem you’re trying to solve or the use case you’re trying to address. Sometimes that use case can look a little different than it might have before.
Your use case before might have been: we need to reliably extract data from contracts at scale in order to populate a deal database and allow our lawyers to understand what’s market. That is still an existing use case today.
But your use case might also be: we need broad AI enablement for our lawyers. That is still a problem to be solved. We need them to be fluent. We need them to have access to something that allows them to have support across the core functions of AI. So it still comes back to the problem to be solved.
Once you’ve looked at that and your own systems environment—the integrations you would need, the preferences your people have, the jurisdictions in which you operate, the languages in which you need this to operate—it reduces significantly the number of products you’re looking at.
Of course, we support that as well. We run technology selection projects where we can recommend based on requirements in our platform.
We do the tracking for you so that people don’t necessarily have to have a dedicated person internally, because it is way too difficult now. One person alone at a law firm—it doesn’t make sense for them to be tracking this market.
But the other thing I would say is that businesses outside of law firms, for years and years, have had R&D departments. And that makes a ton of sense.
Part of what you are validating is: does this technology actually work the way we think it will, and will it be useful here?
As the tech continues to expand and evolve so quickly, in order to really retain that AI fluency and AI literacy internally, you need to have your fingers in this. You need to actually be experimenting on an ongoing basis at the same time as you’re deploying and getting value out of the tools already in your tech stack.
So having some budget set aside for R&D and being able to say, for example, when Anthropic launches Claude for Legal, “We’re going to experiment with this. We’re going to see, does this hold out? What does it do that’s different?”
Or when agentic AI launches in various products, what can we do with that?
I think that’s really critical for firms as well and will also help them weed out the key products that will be of actual utility.
Legal Tech Bubble, Differentiation, and Client Value
Jen Leonard: One of the things that our team hears sometimes, Nikki, is that there are, as Bridget said, a proliferation of really interesting startups—some of them doing really cool things.
Law firm partners go to conferences, they come back to their firm, and they want to buy a tool because it’s specific to their practice. Then on the procurement side, the IT team realizes the company doesn’t have the security protocols, or they’re not accustomed to working with legal.
There’s a lot of lost time, lost resources, and lost enthusiasm among the partners for AI generally because they want it all in the same category.
So what I’m getting to is the conversation around bubbles. With all of the uptick in legal spend the last few years, and the discussion of whether it’s a bubble or not, are we in for a bubble as all these startups are competing and maybe a lot of them failing?
Or is this the period where we’re starting a totally new era in the legal profession, where technology will lead a lot of the practices, versus ten years ago and before, where—as you said—it was really not central to firm strategy?
Nikki Shaver: I think both. That’s a very lawyer answer.
First of all, if it is a bubble currently, it shows no sign of bursting.
We started putting out our legal AI and gen AI maps last year, and we’ve been tracking market to market the number of solutions net in the market as a result of gen AI.
It’s about 100 new companies every quarter. That has been very steady.
We’re going to be releasing our next one in June, and again, we look on track for about 100 new.
And when I say 100 new, I mean having already controlled for M&A in the market and companies that have gone out of business, which we remove from the map and from our active directory.
So this is a very steady growth trajectory, which is inconsistent with the concept of a bubble.
If it’s a bubble, you would expect a massive ballooning of growth followed by, at some stage, a steep decline. And we are not in year one of gen AI. We are in year three—year four, really.
So to me, that indicates that if it’s a bubble, it’s a very sustained one, which is unusual.
I think there definitely is scope for more technology in law firms and generally in businesses, for the reasons we’ve already discussed around the strategic importance of it.
I like to use e-discovery as a good example of this. We almost never talk about discovery as just discovery anymore. We talk about it as e-discovery. That’s partly because a lot of discovery deals with electronic data, but it’s also because we think of the technology required to undertake that type of work in legal as being inseparable from the work itself.
They are completely integrated with one another. In order to learn how to do discovery in a law firm now, you also have to learn how to use the tools that allow you to do that work. That’s why Practical Law has units on e-discovery and how to actually use the technology—not just the discovery theory around privilege, for example.
I think that’s how a lot of different parts of legal work are going to trend. If you’re an M&A lawyer, when you learn about due diligence, you won’t just learn about the actual legal requirements around due diligence. You will also learn about the technology that powers it.
That will be the same all across legal practice, which means you will need some degree of technology baked into every single practice across the firm.
If you look at what’s likely in terms of how much legal work will be automated, it’s a very significant percentage. It’s up around 80%, which means there just has to be fundamentally a bigger spend and a bigger investment in technology to support it.
On the other hand, I also believe that M&A will increase considerably. We’ll see more consolidation. We’ll see a lot of companies that have come out gradually start to fail.
We’re already seeing pressure on some startups in the market where they’re seeking early exits because it’s just been such a difficult competitive space to be in.
So those things certainly will exist. But I do think overall, we’ll continue to need more.
Bridget McCormack: Nikki, you’ve said a couple of times in a couple of different places that AI enablement across firms is now table stakes.
It’s not a way to differentiate yourself like it might have been two years ago, for sure—maybe even still last year.
Where do you think firms can differentiate themselves now? And are you seeing any examples of that?
Nikki Shaver: Ironically, it’s not that different from how firms would have differentiated themselves before.
You look at what makes one top-tier firm different from another. It’s where their expertise lies. It’s where they invest in their people.
I think that continues to be your key differentiator. It’s how you deliver that through AI that also allows for differentiation in the market.
But again, that’s not just through using the same platform every firm around you is using.
One of the problems legal has had for a long time is that it’s such an echo chamber.
When a firm is looking for a particular tool, one of the things they do is pick up the phone, so to speak, to law firm A on the left and law firm B on the right and ask, “What are you using?” And then that’s how they proceed.
Firms are very reluctant to move first—to be the first mover. That second part has shifted a little bit over the last couple of years, which is good.
But again, you really need to be looking at where you are building, what your build capabilities are, and how you are focusing that build in the areas of your firm where you excel.
The firms that are going to win this are not going to be the ones that broadly try to deploy and create internal workflows across every single practice of their firm and spend all their resources doing that.
They will be the firms that take very considered, deep, probably expensive bets in key areas of expertise, or areas where they have particularly important clients, and that deliver embedded, continuous services combined with AI in those areas of expertise—and charge for that in a way that enables them to actually increase margins.
So this is not about a race to the bottom for automated work that can be done better, faster, cheaper. It is about delivering something highly differentiated because of your firm’s expertise and data for key clients at top-rate margins.
That’s how I see differentiation for firms in the market at the moment. There are certainly some firms that understand that. There is more build happening than there ever has been in the past, partly because it’s easier to do and more fun with AI, but also because firms recognize that’s how they can set themselves apart in the market.
It’s also, by the way, how they can compete head-on with disruptors.
Bridget McCormack: I think that makes sense: building, especially in a lane where you have deep expertise, seems like the place you would want to go. But won’t their clients be able to build just as easily as they can—and with even more relevant data and experience?
What are you seeing as the differentiators between what a law firm might be able to do and what the in-house teams among their clients might be able to do?
Nikki Shaver: No, I don’t think that’s right.
What I’m talking about leveraging is your law firm expertise and data that the client does not have access to, except at the moment by having a conversation with a senior partner.
I’m talking about baking in your expertise to a delivery system that combines your law firm’s data and expertise with key client data and delivers targeted, ongoing advice to them, that is super high value on a continuous basis. So yes, certainly clients will build more internally.
But the other thing for firms to realize—and I think some of them have started to recognize this—is that even in top-tier work, where there has been a ton of fear about clients starting to keep more work in-house, and that is a justifiable fear, there is also opportunity.
The amount of work that clients are now doing with AI—or the amount of in-house departments that are leveraging AI—has doubled in the past year, which is a very significant trend, although they were behind initially.
Certainly, clients are focused on using AI as a way to decrease outside counsel spend and keep more work in-house. But there are also types of work from existing clients that large firms, for example, have previously turned down because it’s not the kind of work they do.
Highly commoditized, high-volume work, for example. Law firms now can take that on. In fact, I’ve had a number of clients say, “We are now taking on work that previously we didn’t, because we can without it taking time away from the high-value work that we want our partners to do.” That’s one way where you can actually increase the work coming from clients.
The other way is by working with your clients to be their trusted advisors around automation and AI. It’s very common for clients to turn to their firms for that, even though you would think they would turn to external providers. They already trust their lawyers and their law firms.
Increasingly, law firms have operations centers or client consulting, or they have access to tools that allow them to co-build workflows with their clients, or build them for them and deploy them into their client instance of AI.
That’s another way of increasing value and also ensuring that the work stays with the firm rather than going in-house or somewhere else. So I think there are a lot of opportunities for firms there.
Agentic AI, Governance Risk, and What Comes Next
Jen Leonard: So, Nikki, the first few years of generative AI felt very challenging to everyone as they started getting their heads around what it was. But this year feels like the game changed again in much more complicated ways with the dawn of the agentic era in AI.
To the extent that firms were starting to get comfortable with the chatbot era, now it feels to me like a little bit of, “Oh my gosh, now we have this whole new world to figure out.”
What do you think is the current gap between what agents are able to do alongside lawyers who are managing and orchestrating their activity, and what law firms are willing to absorb with respect to the risk of something that we’re all—including the labs—still trying to figure out beneath the surface?
Nikki Shaver: We are in super early days here.
There’s a big disconnect between what we hear in the market around agentic AI and actual agents being deployed. And certainly, what you said in terms of lawyers orchestrating the work of agents—that’s not happening.
There are agentic tools deployed at law firms, but people are being ginger and moving slowly, as they likely should, around this. The reason for that is exactly what you said, Jen. There’s enormous risk. And from my perspective, firms are already behind on the governance side. It’s something I’m quite concerned about.
Firms leapt quite quickly to implement policies around AI, but policy is more complicated than it has been because the regulatory frameworks around this are so fragmented and moving quickly. So are the rules of professional practice, and various certifications and guidelines coming out in different jurisdictions.
Then firms also have to think about acceptable use policies and responsible use of AI: which tools are okay to use for what types of work, and whether it’s okay to use browser-based technologies or frontier models through a browser. All of that adds huge complexity, and that’s before you even get to the agent.
The way law firms have been handling that, for the most part, is with things like, “Here’s a list you have to look up on the internet of clients who prohibit use of AI. Before you use AI, please go and check that list.” That’s how they’ve been handling it.
It’s not really sustainable as things get more complicated and clients change positions, and you’re looking for far more enforcement across the entire network of your policies around AI—as well as security, breaches, and risks that come from the technology itself.
When it comes to agents, all of that risk profile proliferates significantly. For example, just being able to identify: is that an agent in your system or a person?
I’ve already had conversations with firms where they didn’t consider that ahead of time, because it’s brand new, and later looked at activity and said, “That’s really weird. What happened there?”
And then realized, “Oh, that was an agent.” Then they realized, “We’re in a place now where we need unique identifiers for agents in our systems.”
When you look at something like MCP, where you would have agents connecting with other agents, there are all kinds of downstream impacts you need to consider.
What is the data that a downstream agent has access to, and how is that maintained? What models are being used in the back end? Is that compliant with all of your guidelines and security policies?
It’s a massive quagmire of potential risk issues, but it’s also hugely powerful, potentially. Of course, agentic AI really works more like lawyers work. So it’s about having the right controls in place, rolling out carefully, and understanding the risks involved.
Obviously, when you’re dealing with trusted providers, they will have security and should be able to answer questions. But there are additional questions law firms should be asking their providers if they are using agentic AI, because they introduce a level of autonomy for the system that is previously unheard of.
The thing that is really important for firms to know, too, is that there are now AI governance technologies you can deploy to enforce your policies and other rules internally. Those can help with governance risk around agentic behavior.
I think that’s where firms will start to move. Again, it’s super early days. There are some firms looking at that, but I’m sure we’ll see more of it over the next year.
Jen Leonard: When we’re talking about lawyers not orchestrating agentic workflows and teams, is that maybe not consciously true? If the product is built on top of one of the major labs LLM or agent, it seems to me that those labs are moving increasingly in agentic directions for their flagship models.
Is it possible that lawyers now—or soon—will be unwittingly using agents because of the tool they’re using? And what does that do to some of the momentum behind adoption that we’re starting to see, when lawyers start to realize that there are things going on that they’re not aware of that are risky?
Nikki Shaver: Yeah, good point.
When I said lawyers are not orchestrating, I think we’re probably using “orchestrating” in a slightly different way. When I think of orchestration in the context of AI, I think of an orchestration layer within a platform where it is pulling together outputs from many models into what actually goes to the lawyer or end user.
Or, for example, a layer that is able to optimize the use of a particular model for a particular type of query. In other words, it recognizes, “This is this type of query, therefore this kind of model should be used,” and then combines it with another model that optimizes for tone or output, or for a certain type of format.
Whether lawyers are actually using agentic AI—yes, I’m sure they are. As you say, two very major providers have recently made announcements about having agents. But again, that’s within the context of a provider that already has been fully reviewed from a security perspective.
I think the key is that even if a law firm signed up for a license with Harvey or Legora a year ago and underwent security review at that time, that was prior to there being significant agentic launches. Then there are agentic launches in the platform.
Really, law firms should be going back and revisiting security with the provider to ensure that additional questions are asked about how autonomy is handled within the system, what access the agents have to other systems, and so on. That is something I would imagine is happening—the kind of secondary review upon agent launch.
The other thing—and this continues to be a problem—is that the word “agent” is used very frequently in circumstances where it doesn’t necessarily mean the same thing as we understand by agents.
An AI agent has autonomy and access to tools. It is able to undertake a full end-to-end workflow using those tools, deciding which tools make the most sense within that workflow, and understanding when it has come to an end.
A workflow itself is hard-coded in some way, whether by system prompts or otherwise on the back end, and there is consistency and no autonomy in the way it is carried out.
Obviously, from a security perspective, an agent carries more risk because of that autonomous element, and because it has access to other systems, data, and tools. It is able to decide when and if to use those within particular workflows.
So different levels of security and different questions need to be asked.
That’s certainly something law firms need to be thinking about. And I’m sure they are thinking about it, but it does add another layer of intensity to the whole security around AI.
Jen Leonard: Sadly, we’re near the end of our time with you. We could do four episodes with you because I think we’ve only scratched the surface. Hopefully, we’ll have you on again.
Imagining that we have you on a year from now, what predictions do you have about the developments that will define the year ahead in legal?
Nikki Shaver: I think we will see interesting consolidation in the market.
We may see at least one major provider fail in a way that we didn’t expect.
We might see an IPO in the legal tech market.
We’ll see far more agentic AI, and we’ll see that become far more common.
We’ll see MCP or similar standards come into play in a way that will be really interesting, where we’ll have frontier models with MCP connections to all kinds of legal providers.
And we will see the large AI providers, like Harvey or Legora, have their own MCP environments with integrations or connections to many—
Jen Leonard: Can I just pause you, Nikki? For the audience, can you define MCP?
Nikki Shaver: MCP is Model Context Protocol.
It’s a standard. It’s kind of like a form of API, but it just makes it easier for agents to connect to other systems. It’s making it much easier for providers everywhere to connect one system to another without building a bridge, which is much more technical and harder to do and used to take quite a lot of time.
Now it’s really easy to build an environment of integrations into a modern AI platform that allows the user to leverage capabilities from other providers through whatever they’re choosing to use as that front layer.
So I think we’ll see Harvey as the front layer, with the ability to pull capabilities from a lot of legal tech startups. I think we’ll see Anthropic Claude as a front layer. I think that’s where we’re headed.
Jen Leonard: Amazing. Well, that’s a lot on the horizon, and I have no doubt much of it will unfold as you see it—and probably a million things that we don’t see headed our way in the world of AI.
But one thing is clear: the profession will continue to look to you, Nikki, for guidance and expertise in figuring out how we get to the next phase of this.
I know you and your team do a lot of thought leadership. Where can listeners find you, follow you, and all of those things?
Nikki Shaver: Head to LegalTechnologyHub.com. That’s our platform, and you’ll find our insights there. You can sign up for our newsletter and also reach out to us if you want any support around advisory, AI strategy, technology selection, and so on.
Thank you so much, Jen and Bridget.
Jen Leonard: Thank you so much, Nikki. We’ve been dying to get you on, and we’re delighted to share your expertise with the audience.
And thank you to everyone out there for tuning in to AI and the Future of Law. We look forward to seeing you on the next edition.
Until then, be well.
Nikki Shaver: Bye, everyone.