:Harvey: Power Users: The Purposeful User

It shouldn’t be someone who simply uses a tool a lot. One of the things we say to people is that you should use technology with purpose — you’re trying to do something better for your clients, colleagues, or case.

A power user is someone who knows how to make the best use of a product or tool, both in identifying the right use case and in understanding how it fits within a broader workflow. You know what works and what doesn’t, and you know how to execute. Ideally, you can also explain that to colleagues, teaching them and enabling them to do it themselves.

That requires a combination of law, technology, and practice. The law defines both why you’re doing something and the constraints around it. Technology is what enables it. And practice is how it actually gets done. A power user shouldn’t just understand the technology, because without context, it could be applied incorrectly. It should be someone who can bring all of those elements together.

No. A power user is different from an everyday user. If you focus on the number of power users, you’re looking at it from the wrong end of the telescope. It should be driven by purpose: are you meeting client or colleague needs and helping to achieve their objectives? Some of that isn’t Harvey, and some of it isn’t AI.

It’s not like email, where you expect everyone to be proficient. AI has a wide range of uses and applications. What matters is that if you’re not a power user, you know when to call on one.

Ideally, you teach someone to fish so they can start doing it themselves. But sometimes there are whales so big that you’ll always need a power user (or a small cohort) to handle them. You can’t expect everyone to be able to land it. Catching a fish feeds you for a day — that’s a small, incremental improvement. Catching a whale can deliver far greater, longer-term value. But it requires a different approach, and not just anyone with a fishing rod will be able to do it.

Ideally, you’d have one power user per team so that when people are listening to what that sector or those clients need, they can identify where the technology could help. But it should also be task-based. There’s repeat work across departments: chronologies, dramatis personae, preparing bundles, interrogating large data sets. These are the areas where we should define purpose, workflow, and data, and then identify where Harvey can augment that workflow.

We look at data to understand whether usage is increasing overall and whether it’s becoming embedded in practice. We also look at which teams are using it. If there’s a gap, that’s usually a signal to engage with those teams.

But measuring adoption purely by prompt volume doesn’t reflect quality or outcomes. Someone could use it frequently without addressing the actual problem. In many cases, the solution isn’t solely technological, it’s also structural.

For example, if people complain that I send long emails, the answer isn’t to summarise them with AI — it’s to pick up the phone and ask why I provided that level of detail and to make sure I understand their needs and preferences.

You need to understand the client or colleague’s needs: what’s the objective, and what are we trying to achieve? You also need to understand the legal and technical constraints, so you’re working effectively within them rather than treating the technology as unconstrained.

Then there’s curiosity and openness, both in learning how to use the technology itself and in recognising that most work spans multiple platforms, tasks, and stages. It’s the ability to see the broader workflow.

Equally important is a willingness to collaborate — to recognise what you don’t know, and to know who to involve.

Over time, you build that combined understanding of law, technology, and practice, ideally in real-world environments where you can see how things actually work and learn what delivers results.