AI in law: What I learned as a general counsel introducing a digital agent into the legal team

Q - What prompted you to explore AI in your legal team. Was there a trigger point?

When I joined Fusion5 in early 2025, it was clear the small legal team was operating at capacity. The business was growing, we simply couldn’t keep up with the volume, and I knew that just adding more headcount wasn’t a sustainable solution. There was a particular client engagement that didn’t go smoothly, and it really forced me to step back and ask: what needs to change?

At the same time, AI was becoming more visible in conversations across the business. I was curious how it might be used in-house to make life easier. That moment of pressure made the curiosity feel more urgent.

As a function, we were being seen as a brake on business momentum. I didn’t want legal to be a blocker. I wanted us to be a valued enabler, and that meant rethinking how we worked. Fortunately, being at Fusion5 meant I had access to people with the right capability. I didn’t know how to build an AI agent, but I knew what our function needed to be better at. That was the tipping point that shifted it from idea to action.

Q - What assumptions did you have about AI going in, and how have they changed?

I definitely had some doubts, both about whether AI could live up to the hype and whether it could actually help to solve the challenges we were facing. But I was open to trying something new. I knew there would be limitations, especially in a legal context. I never expected, or wanted, to eliminate the need for human oversight or judgement, but I was curious to see what it would be like to work alongside a digital agent.
One assumption I had was that it would take ages to get something meaningful up and running. And when you’re under pressure, speed matters. It’s one thing to explore a solution and it’s another if that solution won’t have an impact for months.

The speed from problem to action really surprised me. We had something live and effective far quicker than I expected.

Q - How is AI actually being used in your team today?

Right now, agent Claire (there’s a wonderful acronym behind the name – see end of article*) is helping us with first-pass reviews of contracts. We’ve built and trained her on our standard positions, common risks, and preferred language, so she can scan a contract and evaluate what’s in there and where we might need to focus. She’s not doing the final legal review, but she’s doing the groundwork and she’s doing it fast.

We get a colour-coded report we call the doughnut. If it’s wasabi green, that’s a good sign. Custard yellow means there are things that need reviewing. Jam red means there are higher-risk items that need more attention. It’s a fast triage mechanism and sometimes you get a bit of everything in the doughnut.  

What used to take hours now takes minutes, and it helps the team focus their attention where it matters most.

Risk is always front and centre for a legal function and that doesn’t change because you’re working with AI. In fact, I’d say I was more concerned about unwittingly introducing risk. So, making sure we had the right data protections in place, that everything was quarantined and secure, was a big part of how we approached this. We moved quickly, but not carelessly and I never stopped being a diligent General Counsel.

Q - What kind of impact has this had on time, workload, and perception?

Time and productivity are the obvious ones. Tasks that used to take hours can now take minutes. We’re still tracking the data, but our early comparisons of traditional reviews versus Claire-assisted ones show a 50 to 70 percent time saving, depending on the task and type of contract being reviewed. For a small team, that sort of efficiency makes a real difference. It means we’re getting through more work, faster, without compromising quality.

It’s also shifted how the team spends their time. They’re not buried in repeatable review work. They’re using their experience to make more targeted calls, applying judgement where it matters, and getting clearer on the value they bring. So, in that sense, Claire’s giving us space to think and focus.

But the really big shift has been perception. Going to engage the legal team used to come with a sense that you were stepping into a slow, out of touch space.

Since introducing Claire, the response from the business has been incredibly positive. I even had someone say, “Hang on, look at what the legal team are doing with AI. Why aren’t we?” That sparked a few lightbulb moments.

We want to be a great partner for the business, and it’s been rewarding to see innovation actually make a difference. It’s definitely changed how the function is viewed across the business. We’ve gone from being seen as cautious and conventional to being recognised as early adopters. That shift in brand, for a team like ours, is pretty significant and it feels good. 

Q - What did you have to unlearn or approach differently as AI entered your world?

The team is learning to be more comfortable stepping back from doing all the review work themselves and instead take on more of a checking and training role. You’re no longer the one going through the contract line by line, but the one checking whether Claire has done a decent first pass. In a way, you become a second-level reviewer.

I’ve also had to change how I think about the AI itself. Claire’s not a tool you use once and walk away from; she’s more like a junior team member. She still needs input, training, context, feedback. If you want the results to be useful and reliable, you’ve got to put in the time, just like you would with a new hire. I’m still adapting to that mindset, but it makes sense. If you bring a digital team member on board, you owe it to the team and the business to set them up for success too.

It’s not set-and-forget. You’re constantly spotting gaps, adjusting the data inputs, improving how the agent works. And the more we do that, the more useful it becomes. In that sense, I’m learning how to lead a blended team of human and digital resources.

Q - How has your team responded to working alongside an AI agent?

They’ve been great. There’s been no fear or resistance, just a lot of curiosity and a sense of “what else could we hand over to Claire” or “what else can agents do for us?” The team’s workload is big, and this is helping, so it’s never been about replacing anyone’s role. It’s freeing people up and allowing us to grow capacity so, in that sense it’s been a wonderful kind of liberation.

What’s also been interesting is how clearly this has highlighted the value of human judgment. When you’re working alongside an agent, you start to see even more clearly the things only people can do, spotting nuance, reading between the lines and bringing context. Claire gets us part of the way, but the team still brings the final thinking. It’s made that part of their role stand out more. 

Exploring the use of AI has helped us see where human experience and intervention is needed and valued, and that’s been reassuring and helpful for the team.

I think we’re all learning how to partner with a digital colleague, and it’s a way of working we’re embracing as it’s likely we’ll introduce more digital agents in the future.

Q – What has this experience taught you as a leader in terms of mindset, confidence, and approach?

I’ve certainly learnt a lot and made some adjustments in a concentrated period.  Some deliberate, and some that I’ve only noticed in hindsight.

When I joined the team, I’d describe the legal function as conventional. Binary in mindset and process-driven in approach. Introducing AI into that mix probably felt a bit radical, especially coming from me. I still fancy a printed copy, highlighter and stick notes, so I’m not what most people would picture as a tech-forward leader. But we were at a crossroads. We couldn’t keep up with demand, and we had to show the business we were committed to changing how we worked.

What helped was knowing I wasn’t trying, or expected, to make this change alone. Fusion5 is the kind of business that encourages progress. There was a clear message: AI is here, get on board or get left behind, and we’re here to help you. I wasn’t afraid to fail and that executive support made a big difference and gave me the confidence to say, “I don’t know how to build an agent, but I do know what I need it to do.”

I never felt pressure to look like I had all the answers or pretend to be an AI expert. We were probably the surprise runners in the race, the team no one expected to be leading the AI charge. But I knew the problem we were trying to solve, and I had people around me with the skills to help solve it. That was enough.

This experience has also reinforced how I think about my role as a leader. In a business of our size, you have to roll up your sleeves and be hands-on, but you also see how the right use of innovation and technology can help you keep looking across the business and asking strategic questions.

And now that we’re working with AI in the team, I’m learning how to lead in a blended environment, one where humans and digital agents work side by side. That’s new for all of us, but it’s something I’m leaning into.

Q - What would you say to other legal or compliance leaders who are unsure where to start?

Don’t get distracted or put off by the hype. Introducing AI agents doesn’t have to be big or flashy. For us, it started with a real problem to solve and a desire to try something new. The work Claire’s doing now isn’t flashy. It’s just helping us move faster, focus better, and reduce the repeatable noise. That alone makes a huge difference and we’re learning as we go, which helps us get clearer on what to solve for next. 

You don’t need to be an expert in AI to get started. What matters is knowing your function, knowing what’s slowing you down, and having a clear sense of the outcome you want. That’s where you begin.

I’ve been lucky to have access to capability, people in our business who know how to design and build these kinds of tools. Not everyone has that in-house. So, if you’re doing the thinking but don’t have the technical support to bring it to life, the next step is finding someone you can work with. You don’t need a blue sky strategy, multi-year roadmap or big budget, just a partner who can help you move from idea to action.

And don’t lose sight of risk. That was an important consideration for us, especially around data security and privacy, but when you know what you need to protect, it can be factored into the solution. You can move fast, but you still need the right safeguards in place.

The biggest barrier is often in our own heads, thinking we have to know everything or wait until it’s all mapped out. You don’t and the progress we’ve made in a short timeframe is evidence of that.

*Agent Claire (Contracting Legal AI Review Engine)