Inside Joel Hron’s AI Adoption Strategy at Thomson Reuters
By: Gemma Versace
December 9, 2025 23 min read
AI adoption is on every tech leader’s agenda, but turning that intention into real, lasting change requires more than ambitious goal setting.
Joel Hron, chief technology officer at Thomson Reuters, leads a global organization of more than 5,000 engineers. When generative AI began to shift the landscape, he and his team knew they had to lead by example: Before advising customers on AI transformation, their own engineers needed to go first. “If we're building products and putting them in the market … if we're not doing that ourselves, then we're being quite hypocritical about what we're selling.”
In this episode of Keep Moving Forward, Joel shares how Thomson Reuters’ engineering team reached 80% AI tool adoption, what it takes to make innovation everyone’s job and why he believes subject matter expertise is the ultimate differentiator in the AI era.
Turning AI Friction Into Momentum
When AI tools were first introduced to Thomson Reuters’ engineering organization, the results were mixed. A handful of early adopters found real value. But for many, the experience fell short. Expectations were high, and the tools didn’t deliver the magic they hoped for. Adoption stalled.
Joel and his leadership team didn’t see that as a reason to pull back. They doubled down. “We really forced it from top down,” Joel recalls. “And we said, ‘I know that you guys have had some struggles. And I know there are some failures. But we are going to be persistent through those failures, and we're going to keep trying.’”
Instead of chasing instant success, the team focused on creating the conditions for long-term learning. Engineers were given time — sometimes months — to experiment with the tools and figure out where they added value, even if it slowed down their work at first.
And when a team figured out how to meaningfully improve their workflow, those examples were celebrated. Internal success stories became teaching tools. Real workflows became shareable templates.
Skeptical engineers became advocates. Teams that once struggled found breakthroughs. And some of the company’s most experienced engineers saw exponential gains. “Some of our best engineers went from being 10X people to 100X people,” Joel says. “Because once they figured out how to use the tools, they got so much faster.”
He also addressed the quiet concern many engineers had: If these tools are this powerful, what does that mean for my job? “You could be concerned about the future of your job,” Joel says, “or you cannot use them and be concerned about the future of your job.”
It was never about replacement. It was about amplification — and building a culture where persistence, not perfection, powered transformation.
Making Innovation Everyone's Job
Joel believes innovation shouldn't live in a separate team. It should be embedded in how every engineer thinks about their work.
"Innovation can mean a lot of different things," Joel explains. "It could be innovating on a new product. It could be innovating the way you deliver an existing product." What matters most is the culture behind it, like building teams that question their assumptions daily and approach problems with fresh eyes.
That mindset shapes how Thomson Reuters approaches everything from contract analysis to compliance workflows. Thomson Reuters Labs, a three-decade-old applied research group with more than 250 machine learning engineers and scientists, focuses on longer-term exploration. But even product-focused teams are expected to challenge the status quo within their domain.
The key is giving teams the right constraints. Too much freedom without direction can slow progress. Too much rigidity stifles creativity. Joel's approach is to set clear missions while encouraging organic experimentation within those boundaries.
Why Your Domain Expertise Matters in the AI Era
When every company has access to the same AI tools, the real differentiator isn't the model — it's what you build on top of it.
"You need to ask yourself as a business or a company: Why do customers buy from me?" Joel says. "What are they paying me for today that nobody else in the market can deliver?"
For Thomson Reuters, that means combining AI with deep subject matter expertise in law, tax, and compliance. The company employs one of the largest teams of legal and financial experts in the world. That pairing — proprietary knowledge plus emerging tools — is what transforms AI from a generic capability into something customers can’t get anywhere else.
"Figure out how to take what is unique to you," Joel says, "and layer it on top of what this technology is capable of."
Transcript
Joel Hron:
I'm a big believer that innovation should be something that is everybody at the company's job, and that should be something that organically we do. And so we do try to instill that, I would say, into our culture. And innovating can mean a lot of different things. It could be innovating on a new product. It could be innovating the way you deliver an existing product, but this sort of culture and drive for thinking about the problem differently every day and really testing your assumptions and testing your biases every day.
Gemma Versace:
Hey everyone, and welcome to Keep Moving Forward, the podcast from X-Team for tech professionals who are passionate about growth, leadership, and innovation. I'm your host, Gemma Versace, chief customer officer at X-Team. In every episode, we sit down with leaders who are redefining how technology teams work, grow, and lead. People who understand that performance begins with connection.
Today, I'm joined by Joel Hron, chief technology officer at Thomson Reuters. Joel leads a global organization of more than 5,000 engineers and technologists, and he brings a rare blend of deep technical expertise and practical leadership experience. His career spans mechanical engineering, machine learning, startup entrepreneurship, and large scale enterprise transformation, including the AI company he co-founded that TR later acquired.
In our conversation, Joel explains how TR approaches innovation in high stakes domains like law, tax, and compliance, and why giving teams room to explore is essential for building AI-powered products the right way. He shares how his teams balance experimentation with rigor and why grounding technical work in subject matter expertise is essential in high-stakes domains. Throughout our discussion, Joel offers a grounded look at what it means to lead through rapid technological change. He emphasizes adaptability, curiosity, and giving teams the space to learn, iterate, and invent new ways of working. Let's get started.
Hi, Joel. Thanks for joining us today.
Joel Hron:
Thanks for having me, Gemma.
Gemma Versace:
Excellent. Let's get straight into it. And let's start with your journey. So you started your journey in mechanical engineering, and now lead a global team of over 5,000 people. Can you take us through what your journey has been today?
Joel Hron:
Yeah, I've had what I would call a pretty unconventional path to this point. As you said, I'm originally a mechanical engineer by background. Started my educational career working in simulation, working in signal processing. I think that was really, I guess, where I cut my teeth, if you could say it like that, in terms of computer science, and sort of just evolved from there as technology evolved.
Both of those disciplines, particularly signal processing, have a lot of mathematical computation elements to them, which I think, serendipitously, was sort of a natural lead into working in AI and machine learning quite a lot from an early part of my career, which I did. As things go, I started a company with a few of my co-founders a handful of years ago in AI and machine learning, specifically around natural language processing and contract interpretation and enterprise search as sort of primary focal points for us.
And I guess about three and a half years ago, we were acquired by Thomson Reuters to really help them expand what they were doing on the contract analysis side of things, in relation to use of their team of practical law experts. Which Thomson Reuters is one of the largest employers of experts in law, but also tax, audit compliance, et cetera, in the world.
And so we had built some technology that was really useful for allowing lawyers, in this case, to build supervised machine learning models in really an end-to-end way from data curation all the way to deployment and evaluation without engagement of data scientists. So think about a Scale AI type application, but really built around lawyers and legal contract interpretation. And so that was really one of the sort of core pieces of IP that we built that we scaled into Thomson Reuters when we were acquired.
And yeah, the rest has evolved since then. I mean, since joining Thomson Reuters, obviously integrating the company and the technology previously, but also really helping the company on this journey of adapting to generative AI and what it's meant for our products, our customers, and what we do. And I think, fortunately, I was in a good position at that time to help be influential to the company. But I was also working with a tremendous team of now 250 plus machine learning engineers and scientists in our Thomson Reuters Labs who that organization's existed for 30 plus years at TR.
And a lot of people don't know this about TR, but we have quite a rich history of AI and machine learning in our products, particularly in the areas of information retrieval and search type problems, given the complex domains that we work in. So we really had a lot of ingredients to be successful here. And it's just been a matter of really doubling down on those ingredients and trying to capitalize on them in whatever ways we could, which I think we've done a pretty good job of thus far.
Gemma Versace:
Yeah, absolutely. And as you mentioned, I mean, the TR name is very much recognizable worldwide. And what a fantastic opportunity for you to, obviously with all of your wonderful experience, but with them acquiring your business, to now lead such a large and globally diverse team sitting there as CTO. It sounds like the next evolution and the next journey for you at TR seems like a really exciting one.
In relation to innovation at TR, do you have specific innovation teams, or is it very much your teams are allowed to, well, allowed's probably the wrong word, but that they can innovate organically as it's coming to them as they're in the creation and the development phase? Or is it more specifically around you really being able to target what and where you specifically want to focus on from an innovation perspective?
Joel Hron:
I'm a big believer that innovation should be something that is everybody at the company's job, and that should be something that organically we do. And so we do try to instill that, I would say, into our culture. And innovating can mean a lot of different things. It could be innovating on a new product. It could be innovating the way you deliver an existing product. But this sort of culture and drive for thinking about the problem differently every day and really testing your assumptions and testing your biases every day, I would say is culturally something that is important and I think something we try to instill across the board.
I do think it's natural though that a team that has a specific product goal in mind with a specific product function that they're already serving is naturally more constrained than some team that's sort of just like off thinking blue sky. And so we do try to create more specificity for teams so they know what their mission in life is. And is their job to deliver something X% better than current state, or is their job to think blue sky of the problem? And that can mean you have separate teams that have separate remits of innovation within the company. Certainly I mentioned TR Labs, I would say is one of the groups that's more involved with the blue sky or applied research that we might be doing. But I do think the sort of drive for organic innovation is something we try to push through all of our teams in some way.
Gemma Versace:
Yeah, fantastic. Thanks for that. So also you mentioned that, obviously when you are developing products, there's lots of different iterations, and it can be moving quite quickly, but it's so important to be able to do that touch base with the subject matter experts or your clients. But what frameworks have helped you keep pace with your timelines, but also at the same time without having your teams burn out?
Joel Hron:
I would say we've used a number of these technologies that have started to evolve in this place. There are various teams using Pydantic, Langfuse, LangCraft. We've had a number of teams that have started operating more on A2A or the Claude SDK, or OpenAI's new Apps SDK. And so I would say we haven't necessarily standardized on any one. And I think one of the things that we've actually learned is that driving too much standardization early on, particularly as the space is evolving quite fast, sort of locks you in a way that actually slows you down instead of speed you up. And then on the flip side, what we've learned is that if you don't settle fast enough, and you sort of leave it too open-ended, then you also slow down.
So I really do think there is some middle ground there, and you have to be perceptive about when is there enough critical mass or collective action in the market around a particular tool or technology or framework like MCP would be a good example. Where there's enough collective action in the market today around this to say, yeah, we're going to use MCP as a company, and this is going to be something that we all do and we don't really question it. But you have to be an early mover there, but also not too early to the point where the whole world changes up from under you quite quickly. And so we've tried to be pretty judicious about how quickly to standardize and when we do standardize, to have some conviction around it.
Gemma Versace:
Yeah, fabulous. Thank you. And across TR's legal, tax, and compliance portfolios, what are some of the signals that you and your team look for to know whether or not an AI product is really delivering true value to the customer?
Joel Hron:
Yeah, that's a really good question actually, and honestly something we've spent a good bit of time debating and researching a little bit. And I would say there's a lot of objective and subjective ways to measure that. I think, subjectively speaking, I love hearing customer testimonials through our sales team and our customer success team. And I hear these anecdotes or examples of a team of lawyers did something they could have done in a month and they did it in 10 minutes kind of thing. These anecdotes are real, and those really speak volumes to me in terms of giving me confidence that these solutions that we're putting into the market are really having tangible impact.
I think quantitatively we also look at usage of the products. And I get reports on this week on week on week and I see substantial double digit growth week on week on week of users engaging with the tools more and more and more. And to me, that is very objective evidence that many users, and certainly the majority of users, I would say, are finding tremendous value in it. Because they're going back to it a second, a third, and a fourth, and the time to do more work. And that's, I think, one of the best points of validation we look for.
Gemma Versace:
Yeah, I definitely couldn't agree more. The next question that I do have, one of the amazing statistics that stood out during our conversation that you shared with us is that, at TR, you have 80% user adoption for AI tools. Now, across a 5,000 plus strong workforce, this, Joel, is just an absolutely, truly amazing achievement. I think everybody that is listening to this episode just wants to know how did you bring your team along for this journey, and how have you been so successful in getting such a large uptake of 80% of users that are regularly using and adopting AI tools across TR?
Joel Hron:
Well, one of the first things we did as a leadership team, I remember sitting down with my leadership team, we sort of all kind of came to the conclusion that we got to walk the talk. If we're building products and putting them in the market, and telling our customers, "Oh, you need to transform yourselves to use AI and these AI products are going to accelerate your work and improve the quality of your work," if we're not doing that ourselves, then we're being quite hypocritical about what we're selling.
And I think that was the first really salient moment where we said, "Okay, this is something we've got to lean into internally as well as externally." And AI dev tools in particular in the engineering space has been one of the fastest evolving and fastest improving segments of AI in the last few years. And I think what our experience has been is there was certainly this early minority of early movers in the engineering space who had seen some good value from it. And there was an early majority of those users who were disappointed in some way. They expected it to sort of do this magic trick. And it did not do the magic trick that they were hoping it would, and it fell short in some way. And there are hundreds of ways in which it fell short.
And I think what we saw is a really fast influx at first and then quite a big slowdown. And we did two things at that point in time. One is we really forced it from top down. And we said, "I know that you guys have had some struggles. And I know there is some failures. But we are going to be persistent through those failures and we're going to keep trying. We're going to figure out what works. Because there's enough signal that something will work, we just have to learn how to find it."
And the second thing we did is when we saw teams find it, we tried to get loud about it. We tried to be very vocal internally about what worked, how they did it, why it worked. We intersourced it so it was available to everybody in the company to go use and duplicate and build off of. And I think this pattern of sharing the things that were working allowed the organization to learn together a lot faster.
And at some point, I don't exactly know when it was, but at some point, there was enough understanding that there was a path to success here if you spent enough time finding it and sort of building out the systems around it in order to be successful. And we gave teams the space to spend the time to do that. We set the expectation that, yeah, this might slow you down a little bit for the first month, or first two months, or three months, and that's okay, but be persistent because there's likely light at the end of this tunnel.
And I think giving people the sort of space and encouragement to be persistent through that and giving them examples of what it could look like, helped us move through that period of disillusionment, if you will. And certainly there are areas of deficiency still today where we think that something's going to work and it doesn't quite work, and that's fine, but I think we're starting to see more wins than losses at this point on that math.
Gemma Versace:
Yeah, which is great. And I think the two points that you called out there that are really important, and I think given the success that you've had also for listeners to be able to take on board, is that in a world where a lot of leaders are very open to a lot of collaboration and a lot of employee-led innovation and things like that, it's also to achieve really successful results, it's also sometimes totally reasonable for leadership to mandate, and to top down and say, "This is something that our business is going to get behind. It's an investment that we're making, and we expect the user adoption and the buy-in from across the teams."
And then the other piece that you called out was really making sure that you get loud around the successes and really wanting to continually show and highlight what good looks like. But also for those people that are persisting, that are buying in, what some of the individual and team successes look like as a result of that too. I think that's a really powerful and compelling way to be able to help bring people along for the journey.
And I guess the flip side of that is also, yes, with top down and getting loud with the successes, what type of pushback did you get however? Because even though, with all of the right communication strategy and the talking the talk, but then showing them how to walk the walk, did you get much, I guess, visceral reaction from your teams where they were quite hesitant about it? And with a lot of noise around AI is taking everybody's jobs, asking for such adoption within your teams, no doubt there were lots of questions around, are we training these AI tools to ultimately take our jobs? How did you work in addressing those fears across your team?
Joel Hron:
Yeah, I would say in certain pockets, we did get some pretty direct feedback to the negative, and there were a number of negative reactions to that process. And I think the way that we dealt with them was frankly just in stride. One of the observations I made very early on about who was finding success in using these tools and who wasn't, is that it was some of our best engineers in the company. And in fact, those engineers were the biggest skeptics at the beginning. And at the end, they were the biggest proponents, and they went from being 10X people to 100X people.
And it was that realization that I think framed my reaction to a lot of these negative was like, I can point to specific examples of really phenomenal engineers in the company that everybody looked up to as people who were leading the charge there. And I could also point that out to say, look, these tools are not here to make everybody slightly better or shave costs out of the organization. They're here to elevate the best of the best of our engineering talent. And if anything, we are out there trying to hire more phenomenal talent because we find that that's the most accretive accelerant to development that we can get.
So I think for us, pointing to those examples and really elevating some of these people who have been staples of our engineering team for a long time as examples, gave people the right signal that this was the path we were on. And then lastly, I would say it's just being real clear about the counter. It's like, look, you can use these tools and automate things and you could be concerned about the future of your job or you cannot use them and be concerned about the future of your job.
That really is the sort of mental mess to work through because I think that's the reality that we all face today. And I put myself in that bucket too. And I was very clear with my teams that I put myself and my leadership team puts themselves in that same bucket. And none of us are immune to that. But leaning away from that change is not the right way to adapt to it because that's surely not going to end, I think, in a positive outcome for any of us. So that was kind of how we framed it.
Gemma Versace:
Thank you. And it would've been really important and empowering for your team as well to be able to see you and your leadership team also leading by example, not doing do as I say, not as I do, to be able to show them you guys leading the way would've also certainly had an impact on helping the motivation levels across the team for that buy-in.
So with a lot of, obviously TR being such a large business, you would no doubt be very focused on really wanting to be able to work with the individuals within your team in mapping out what the career advancement and the skills acquisition and growth opportunity looks like for them. How are you doing that in a world where AI is really doing a lot of, it can be seen, I should say, as doing a lot of the heavy lifting with this? How are you working with your HR department or professional development department, training department in making sure that you can continue to be able to help your teams with their career progression and skills acquisition in this AI dominated world at the moment that we do live in?
Joel Hron:
Yeah. I mean, in terms of training and enablement, we've put out lots of internal materials on this. We work with our partners, major cloud providers, the major frontier labs, to do a lot of hands-on co-development or workshops or things like that with those teams to educate our employees. We have tried to really be fluid about our talent movement across the organization, particularly from labs into the rest of engineering. So in areas where we had this historical AI and machine learning experience, and trying to bring that closer to the rest of the engineering teams to give them an opportunity to develop those muscles and develop that understanding as well.
We've also instituted new career paths for AI engineering specifically. And again, AI engineering may sort of be like the cloud engineering 15 years ago where cloud engineering used to be a specific title that we called an engineer who worked in the cloud, and now it's just like every engineer on the planet. AI engineering may very well be the same thing and probably will be. But today it sort of stands for something rather specific for us at least.
And we've been specific around what are the skills and expectations of an engineer with that title. And that's been a way to really memorialize what is unique about building AI products versus not. And give people a roadmap of what skills they might need to develop in order to move up that curve a little bit more.
And so I think a combination of all those things have really helped move the organization forward in terms of climbing that curve. I think it also helps that AI, from the top down, from our CEO down, has been a common thread of priority across our entire product portfolio. So it's not just this one component of one application. It's really pervasive. And so almost every one of our teams can't escape the impetus to really do something material here and lean into it. So I think that really helps catalyze the organization in that direction as well.
Gemma Versace:
Yeah, absolutely. And also speaks to the level of investment and commitment that TR has for their staff as well, is such robust plans to be able to make sure that there is that continual learning, but also the continual opportunity that your team members can be really safe in the knowledge that there is that continued development opportunity for them at TR.
Joel Hron:
Yeah. They see that and they see our CEO Steve say, "We're investing $200 billion a year in AI," and they have some confidence that this isn't just like a flash in the pan, this isn't going away. This is here to stay, and something I should get on board with because it's not going away.
Gemma Versace:
If a CTO at a mid-size company came to you and said, "Joel, where do I begin with gen AI?" What would you say? And I guess to compartmentalize it, what's the smallest and the smartest place to start given your experience?
Joel Hron:
Well, I think the most helpful place to start is baselining on what the models are capable of out of the box. And that bar continues to shift, but I think a lot of people maybe miss that as a baseline. Or they take that as a baseline, and they just jump to like I'm going to go build some application on top of what the model's knowledge is. I think those are probably the wrong path.
I think you got to start with what is the baseline capability of the model. And you need to ask yourself as a business or a company like, why do customers buy from me? What are they paying me for today that nobody else in the market can deliver? And you need to figure out how do I take what is unique to me and layer it on top of what this technology is capable of to deliver value that is unique to me.
And I think people, that second part of the equation, which is what is unique to me and how can I add to it, is the hard part. And I think that's where you got to spend a lot of time really being decisive and crisp and inward looking in terms of what is valuable about the proposition of your business. And once you find that, and you double and triple and quadruple down on it, and you make it as big and as meaningful as you possibly can, but don't miss that part of the equation, because I think that is ultimately what makes you get the business.
Gemma Versace:
Yeah, fantastic. That's some really good advice. And looking forward ahead, so the next 12 to 18 months, what do you think are going to be the main challenges or opportunities that engineering leaders really need to be ready for in this AI era? What are some of the challenges, opportunities that you think are the most important for them to focus on?
Joel Hron:
A couple of the challenges. One is keeping pace with technology. I think building teams that can be adaptable, like I said before, and constantly reinvent themselves is a challenge. I think security will be a growing challenge. I mean, it's becoming harder to build secure applications as they become more distributed. And it's also the threats of security, I think, are becoming more pervasive than they have ever been before. So I think the challenges of security are a growing one.
I think in terms of opportunities though, there are many opportunities. A couple that I would pick off. One is really the opportunity, particularly for us, we have a large legacy portfolio of products that do very specific and meaningful things, but they perhaps are quite narrow in what they do. The ability to think of these things as an ecosystem of products that work together with an agent that orchestrates between them, I think is a really powerful concept that really brings new life to old products and applications and businesses. And there's really, I think, a huge opportunity to rethink existing portfolios and products in new ways.
And the second thing that I would say is an extension of that is also rethinking markets or groups of products in different ways. So the ability for third parties and partner companies to have applications that interact with each other more seamlessly than just by API, I think is a growing opportunity, and I think an opportunity for products and services to work in more seamless ways than we could before.
Gemma Versace:
Yeah. Thank you. That was really insightful for not only the challenges that you've identified, but also really importantly as well, what some of these wonderful opportunities are across the next 12 to 18 months as well. And in that vein, with keep moving forward over the course of the next 12 to 18 months that's in front of us. As a CTO that's leading through one of the most disruptive shifts in recent history, what does Keep Moving Forward mean to you, and how do you communicate that to your your teams to keep moving forward?
Joel Hron:
Yeah, it's an interesting question. I would say keep moving forward to me means, A, knowing where you're moving in the first place. Having the goal in mind, having an outcome in mind, having a purpose in mind. To me, it starts and ends with that.
And then secondly, is I think, not being afraid to fail along the path of getting there. I think there are a lot of things that we do every day, that I mentioned a couple of them earlier, where we do them and we realize we did them the wrong way. And the worst thing we can do is do them the wrong way the second time or do them the same way we did them the first time the second time. And I think just being a learning organization is probably the most important thing you can do to ensure that you don't fall victim to being too afraid to fail. Because there are certainly many things you're going to do wrong along that journey. So I would say those two things, like having a purpose and knowing where you're going, and having the humility and willingness to fail along the way, because you certainly will.
Gemma Versace:
I love that answer. I love the answer because sometimes the whole methodology or the feelings behind keep moving forward is just keep moving forward at all costs. But the two things that you've called out that you can't really move forward unless you know where you're going is just such an important and insightful comment.
But also I really also love what you really highlight there is to say, yes, keep moving forward, but also make sure that you do take the time to do that audit of what potentially didn't go 100% right. And make sure, to your point, that you don't replicate any failure into the future. And really learn from it and communicate it and sit in it, and know where and how and why the failure potentially occurred, and make sure that it don't happen again, to get back straight back into that lane to take you where you want to go for the overall purpose.
So thank you so much, Joel. I have thoroughly enjoyed this conversation. Thanks so much for being so detailed and sharing just so openly and honestly your thoughts and some wonderful advice that has been shared with our Keep Moving Forward listeners as well. So thank you so much, Joel.
Joel Hron:
Absolutely. Thank you for having me. I very much enjoyed the conversation.
Gemma Versace:
My conversation with Joel underscored how powerful it is when leaders create conditions for discovery. His approach shows that innovation grows when teams have room to explore ideas, test assumptions, and learn from what they build, especially in environments where accuracy and trust matter most.
Joel also highlighted the importance of communicating clearly during times of change. By setting expectations, giving teams the time to experiment, and celebrating what works, he helped drive remarkable adoption of AI tools inside a global engineering organization. His experience shows that people become more confident with technology when they see leaders using it, supporting it, and sharing real examples of its impact.
What stood out most was his perspective on adaptability. As AI capabilities shift month to month, Joel emphasized the need for teams who are curious, humble, and willing to rethink how they work. That mindset, combined with meaningful goals and a willingness to learn, is what helps organizations keep moving forward.
Join us next time for more conversations with technology leaders who inspire us to grow, lead, and innovate. You can find us on Apple Podcasts, Spotify, or YouTube Music. And if you enjoyed this episode, please share it with your network. Until next time.
SHARE:
