Skip to content
Child rocketman

Expectation management: AI reality check

With any major technological advance, particularly one as major as generative AI,  you can safely assume that law firms will go through the following cycle:

  1. They create policies to ensure that users stay within agreed guardrails.
  2. If their competitors have a solution, they want it.
  3. Users will assume any new solution can do way more than it can, much earlier than it can.
  4. There will be little in the way of training on how to use the new technology.

The consequences? Expectation will be (and is) out of kilter with reality. And, dare we add, many lawyers who are excited about using generative AI tools for all manner of legal work, may still grapple with basic software such as Microsoft Word.

The Adoption Curve

Law firms are, unsurprisingly, at very different stages in their generative AI adoption curves. Some were quick off the bat to create policies that encourage (safe) usage of generative AI tools, with innovation teams well down the road of proof of concepts and rollouts. Others are nowhere near that yet. They are still considering the ramifications of the new technology on their business, including what it all means for their IP and their brand.

One common thread among many firms that we speak to is that they are basing much of their Gen AI roadmap and strategy around large, established vendors such as LexisNexis, Thomson Reuters and Microsoft. Clifford Chance this year adopted Microsoft Copilot globally, with many other firms at various stages of adoption.

Copilot is a perfect example of why and how firms need to manage users’ expectations as to what the technology is good for, and what it is not.

Copilot use cases

Early adopter law firms report that Copilot is excellent for the following use cases:

  • Help with summarising emails and drafting responses
  • Document creation – if you have some bullet points you can ask Copilot to create a first draft that you can edit
  • Excel users can use Copilot to produce charts and generate visualisations
  • Copilot is excellent at helping to generate PowerPoint presentations
  • One of the most effective use cases is in Teams, where if you enable transcriptions, it will give you a summary of the meeting and action points
  • For OneNote users, Copilot has a great search function, and it can find information that’s sometimes buried in your text
  • In Microsoft Power Automate, you can ask Copilot to automate steps as a starting point.
  • It is really good at translation.

What Copilot is not excellent at – and lawyers’ expectations need to be managed accordingly – is drafting legal clauses based solely on the content in your Microsoft environment.

What Copilot is not excellent at – and lawyers’ expectations need to be managed accordingly – is drafting legal clauses based just on the content in your Microsoft environment .

Legal Clause Drafting

Law firms spend decades building up their intellectual capital and precedent-based knowledge banks, and while there is interesting integration work going on with the likes of iManage that may change things going forward, right now Copilot ‘out of the box’ should not be used for legal drafting.

While Copilot, which is based on OpenAI’s large language model GPT4 Turbo, is embedded within your Office 365 environment and is not open to the internet, most firms do not use Microsoft-based document and knowledge systems, meaning that the content required for good drafting – and maintaining the firm’s competitive edge – will not be available to Copilot.

Yes, you could extract data from a precedent or precedents and ask Copilot to help with a first draft based on that specific document. But fee-earners need to understand that Copilot is not a magic tool that suddenly has the capability of a junior lawyer. Like with any other technology, what you get out is based on what you put in.

Again, as with all technology, training is really important. Assuming that lawyers will understand these sorts of nuances because you have a policy is like assuming that we read the Ts and Cs when buying a new iPhone. Some will. Most won’t.

Contract drafting made simple

We launched Draft Plus, our AI-powered drafting tool for lawyers at the start of 2025. It is simple and safe to use and sits within law firms’ own environments, only accessing their existing documents and drawing on their precedent libraries.

In trialling Draft Plus, the scope of their burgeoning precedent libraries has surprised many firms. There is often a higher volume of data sitting in their environment than they have been aware of, including duplicate clauses, which they can then manage appropriately.

Additionally, the ease of use and minimal training required for Draft Plus has been very well received and a welcome change from often more complex tech tool adoption.

Find out more about Draft Plus here

Where next with AI in Legal Tech?

The excitement around AI isn’t going anywhere.  In our view, the route to successful adoption within law firms lies in identifying the right tools, ensuring that their end users understand their capabilities, and (crucially) can use them seamlessly in their day-to-day work.

The key for any vendor, as for any end user, is to be transparent, strategic, and above all have realistic expectations.

 

To find out more give Rob Simcock a call on +44 (0)333 202 0971 or email rob.simcock@kutana-brochet.co.uk or use the Contact form