There’s a question I get asked a lot these days: are you worried about AI? And my honest answer is no, I’m genuinely, deeply excited, but I also understand why people ask. Every major shift in history has come with its share of doomsayers, and this one is no different.
What I find fascinating is that we’ve been here before with this kind of seismic, industry-reshaping change that fundamentally rewrites the business models. For me, that was the dot-com era, in the late 1990s. I was working with one of the first internet banks and people said the same things they’re saying now: jobs will disappear, industries will collapse, nothing will ever be the same. And you know what? They were right that nothing would be the same. But they were wrong about what that meant for people, and jobs.
The jobs didn’t disappear. They changed. And I believe, with everything I’ve seen over the past few years at Thomson Reuters, that the same will be true of this AI era.
Content is our competitive moat
When people talk about AI, they focus on the models, the large language models (LLMs), the generative capabilities, the speed of inference. And yes, all of that matters. But so does great UX and user interfaces and just generally superb software. But what I’ve come to believe, and what our experience at Thomson Reuters confirms, is that now, the real competitive advantage lies in content the AI accesses to produce the results it’s offering users.
We are a content-driven technology company and we’re incredibly fortunate with our franchises, whether that’s Westlaw or Practical Law or our tax and trade content, we have decades of trusted, accurate, expertly curated information. This is what makes our AI powerful in a fiduciary professional context. For a person or organisation that is legally and ethically bound to act in another party’s best interest, having the facts, the law, the regulation, at the heart of the work they are doing is paramount.
Anyone can access a standard foundation LLM. Any organisation can build a product on top of one. But no other organisation has the depth of verified legal, tax, and regulatory content that makes the AI outputs trustworthy for a professional who has a fiduciary responsibility to their client. That’s the difference between a novelty and a tool that professionals can actually trust.
When we talk about products like CoCounsel, the defining feature is that you can see exactly what’s being referenced, you can check the citations, you can have confidence that the answer is grounded in verified law. The lawyer or the tax professional still reviews and verifies the outcomes, but they can get to a high confidence answer dramatically faster.
The transformation is here
When I joined Thomson Reuters in 2020, we had the Thomson Reuters’ Labs function that had existed for decades. Thomson Reuters Labs have been doing genuinely interesting work over this time, but it often operated somewhat to the side of the main engineering function – innovating, experimenting, but not always fully integrated into how we built products day to day.
Now, the Labs team is arguably the most important team in engineering. It has grown by many multiples, the team’s data scientists and AI engineers are embedded across all of our product teams, not sitting in a separate innovation silo, but integrated into our other teams and driving change in how we build our products.
And the results are tangible. When I joined, most of our products were updated maybe once a quarter. Some were on multi-year release cycles. Today, we typically ship updates every couple of weeks.
But the transformation isn’t just about shipping speed. It’s about what our engineers are actually doing.
For our engineers, 70-80% of the code is now written by AI, freeing up our people from just spending their days just producing lines of syntax. We’ve freed them up to now spend their time thinking about strategy, customer value, architecture, and ecosystem integration. The nature of the work has shifted upward, which isn’t a threat to the engineers’ jobs – it’s an elevation of the role.
Reimagining work across every function
There is no role at Thomson Reuters that is untouched by our AI transformation. Our customer support teams are using AI to surface answers faster, making their jobs easier and their service to our customers better. Our sales teams are working with tools that, for example, can flag early warning signals – such as if a customer has had two unresolved support interactions in a week, the salesperson knows about it before it becomes a cancellation conversation. This kind of proactive, data-driven relationship management simply wasn’t possible five years ago.
Even the way I personally prepare for the many meetings I have has changed dramatically. A year ago, preparing for a week of customer and stakeholder meetings meant requesting briefing notes, reading through documents, synthesising information manually. Now, much of that information lives in our ecosystem and I can simply ask questions – “what’s our strategy in this market? what are the key risks? what are customers saying?”. The time I save on gathering baseline information is time I can spend on actual thinking and offering our customers and our organisation better analysis leading to better decision making.
This shift in personal productivity, multiplied across the 26,000 professionals that Thomson Reuters employs globally, is where the real value of AI starts to compound.
How to actually get organisations moving
Another question I get asked, particularly in markets where AI adoption is still finding its footing, is how to move from individual curiosity to genuine organisational capability. This is the gap that many businesses are existing in right now – their people are experimenting personally, but organisations aren’t capturing that value in any structured way.
I’ve worked in ‘change’ for almost 30 years, and it still holds true that people generally don’t like it. It makes them nervous. So the key is to try and find a way to make them excited about it rather than fearful.
Here’s what has worked for us here at Thomson Reuters. First, be clear on ethics and guardrails before you scale adoption and ensure you communicate that with your teams. Telling people clearly what is safe to use, why you’ve built the environment you have, and what the rules are can help remove anxiety and accelerate adoption rather than slowing it down.
Second, make it a low stakes to experiment. In our early days, the rule was simple: if you have a use case you want to try, we’ll fund six weeks of experimentation. If it works, we’ll invest more. If it doesn’t, we’ll shut it down. Both outcomes happened. And that’s fine – because the learning from what works and what doesn’t is also really valuable.
Third, start at the top. I asked our executive team to do a hackathon early in our journey. Some of them were clearly uncomfortable. Some didn’t know their login credentials for our internal AI platform Open Arena. But going through that experience, actually allowing them to build something and see what’s possible, gave them the confidence to then go and ask their own teams the right questions. Leadership curiosity is contagious.
Looking ahead
If I look two years ahead, I think the picture is fairly clear in some areas and genuinely open in others. In engineering, I suspect no one will be writing code in the traditional sense. Engineers will be directing agents, reviewing outputs, making architectural decisions. The team will be a mixture of humans and AI working together, and honestly, some organisational charts are already starting to look that way.
In customer-facing roles, I see growing demand for subject matter expertise – lawyers, accountants, tax and trade specialists who can work alongside AI and help customers implement it effectively. The most complex questions will always need human judgment. But the simple, repetitive queries will be handled automatically, which means our people can focus their expertise where it matters most.
Here at Thomson Reuter’s, the opportunity we’re seeing is significant. We’re growing faster, our customer satisfaction scores have risen substantially and we’re shipping more product, more quickly, than ever before.
The AI era isn’t something that’s coming. It’s here, and for those willing to lean into it thoughtfully, the upside is enormous.
