Anybody who has worked for any amount of time in the marketing and BD department of a UK law firm is likely to have a visceral reaction to the phrase ‘directory submissions’.
Chambers & Partners and The Legal 500 were once, for those who are lucky enough not to know, pretty much the only way law firms marketed themselves and are still very much part of the mix. The directory publishers require a large amount of information on a deadline. Much of what is required can’t be taken from a database and needs ‘fee earner engagement’.
Unsurprisingly, this can be a torturous task. Typically, fee earners are contacted for feedback, need to be chased, then chased again. Time is wasted. Then, as the deadline looms, minds are focused, mountains are climbed, and submissions are produced. The result is absolutely fine. But not as good as the authors would have liked them to be.
Holly Taylor-Wright, a BD professional at Clifford Chance, looked at this process and thought that there might just be a better way.
AI.
Using Microsoft Copilot, she developed a system where lawyers provide short bullet-point answers to outline their work. Copilot is then used to reference information from the practice area description to align the messaging and create a standardised write-up.
Yes, each response requires editing and honing. Which is good, because it preserves and enhances the human element. But the bottleneck, getting full-length write-ups from busy fee earners, has been eliminated by streamlining the process.
And the time saving?
“Fifty to sixty per cent,” says Taylor-Wright.
“It means that we can take write-ups from one submission and, if they’re relevant, put them into another submission. It all stays standardised, rather than having several different formats or versions of the same work highlight,” she says.
“The time spent chasing and rewriting submissions has been drastically reduced,” she says.
It has taken almost 40 years for tech to solve a marketing problem that came from the era of carphones, filing cabinets, Rolodexes and, for the cutting-edge adopters, Lotus 1-2-3.
AI has done it in a logical and understated way.
The adoption debate
The approach to AI adoption varies dramatically across firms. Hugh James, a Cardiff-headquartered firm whose CTO was recently ranked in the UK’s top 100 AI people, went all-in approximately two years ago. After evaluating various solutions, including Harvey and Copilot, they chose ChatGPT Enterprise.
“The reassuring point was that it was a closed environment,” says Joe Purcell, Chief Commercial Officer at Hugh James. “We were comfortable putting more sensitive information into it, because we knew it wasn’t teaching other people.”
At Simmons & Simmons, the approach is equally aggressive. The firm developed its own in-house LLM technology called Percy and adopted what Camilla D’Arcy, who leads the global client programme, calls a “truly AI-enabled” strategy.
“I say to the team, before you start any task you think ‘What part of this can I delegate to AI so I can add more client value elsewhere?’
“It’s about adopting AI and making use a habit. For example, at every internal meeting, we will use Copilot for the meeting summary and notes.
It creates efficiency and means focus can be given to higher-value outputs.”
At ACCA (Association of Chartered Certified Accountants), where the Policy and Insights team produces substantial thought leadership reports of 15,000 to 35,000 words, the approach is more measured.
“We are still best described as being in the proof-of-concept stage,” Clive Webb, one of ACCA’s senior subject experts, explains. “In thought leadership, you need to make sure that you identify and verify the primary source. LLMs often suggest sources where the meaning of the primary source has become confused. There have been instances, in the broader community, where unreliable sources have been treated as verified.”
Webb and his colleagues use AI to analyse and sometimes translate mountains of research data – hundreds of interviews, global roundtables, thousands of survey responses – and to test whether finished reports communicate their intended messages. Crucially, they use a closed environment. Generic open systems “are false friends”, Webb warns. “They can conflate sources,” which is often the cause of hallucinations.
This “work in progress” is, really, a comprehensive use of AI’s sorting and analysis skills combined with a rejection of its debatable capacity to write.
Top view photo of wall covered yellow sticky note papers isolated grey background with blank space
Transforming tenders
Perhaps the most dramatic AI success story comes from Hugh James’s work on bids and tenders. The traditional nightmare: documents arrive, fee earners struggle to address them while managing client work, and RFPs are hard graft.
The firm created a custom GPT connected to its SharePoint bid library. When a major insurer asked them to tender for a niche area of work – the GPT produced a first draft in minutes.
“Our estimates were that the draft took us 70% of the way there,” Purcell says.
“In terms of hours saved, that’s absolutely vast, because a team of people starting from a blank sheet of paper would have taken two to three weeks just to get it to that 70% point.
It probably took us 40 minutes with the right prompting.”
But the real breakthrough isn’t just speed – it’s institutional knowledge. During one tender for renewable energy work involving very niche types of contracts, the firm could only recall doing such a contract once.
“We interrogated our custom GPT and found seven examples where we had done them four years ago as part of a big deal. No one remembered that that type of contract was in there, but it was referenced in a tender we’d submitted a couple of years ago.”
This illustrates what Purcell calls “the perfect use case” for AI: it’s not writing the tender; it’s assembling it from the firm’s own materials. “We’re very conscious of the need to not pump out any kind of purely AI-generated material,” he emphasises.
The remaining 30% is crucial – experienced human professionals reviewing, verifying, tweaking language and ensuring accuracy.
Supercharge me
At Clifford Chance, Holly Taylor-Wright has extended the approach, developing prompts that generate CV descriptions and deal lists for pitches.
Working with knowledge lawyers, she’s developed an end-of-deal review process where teams identify big deals, hold meetings with standard questions capturing information, then use Copilot to turn the recording into a first draft of directory write-ups.
“It can actually output it in the format of the directory table, which is so much faster than having to do it all manually.”
And all of that will be invaluable during the next directory rotation.
At Simmons & Simmons, D’Arcy’s team uses AI across their entire marketing and business development spectrum. For events, they use AI to profile attendees and match them with appropriate firm representatives. AI also dramatically accelerates analysis tasks, meaning activity is more strategic as it is data informed, monitored and can be adjusted at speed.
“I was doing research on M&A activity in a particular sector, and typically that might have taken me two or three days to pull together. With AI, I had a comprehensive report in maybe two hours, which I could then fold into my BD strategy,” D’Arcy explains. The firm also uses AI to translate technical legal content into business-friendly language for client conversations.
For someone working in a technical practice area without a legal background, AI also helps her understand complex legal developments.
“Obviously, for me, leading a portfolio of clients across five highly regulated sectors, there is a lot going on, a lot of changing regulation, macro trends and different business needs and challenges that I need to be across. Some of the content, even internal briefings, can be quite technical. AI is really helpful in simplifying a lot of it for you, and in making information more commercial and translated for the business,” she says.
The human factor
None of this works without human skill. As D’Arcy puts it: “A lot of AI needs to be directed. You still need someone with vision and someone with the strategic perspective: Where are we trying to go? What are we trying to deliver? And what outcomes are we trying to drive? Then use the tools at your disposal to get the best outcome as quickly as possible.”
Prompting, it turns out, is a craft. Taylor-Wright invested significant time developing directory prompts and doing the firm’s AI training modules, to use Copilot’s ‘prompt coach’ agent which develops prompts in ways AI can interpret properly.
“It does take time. You have to go through several iterations to get the prompt right. But once it works, the time saving is massive. It’s important to keep the human in the loop though.”
The philosophy at Simmons reflects this: “The big message we’ve had across the team is: just be curious and take time to play,” says D’Arcy. “We just have to play with the technology and see what’s possible.”
The client perspective
Perhaps the most significant development is the shift in client attitudes. D’Arcy notes that some clients have reviewed their guidelines to encourage AI use, having originally prohibited it.
“We’ve had clients review those terms. Having originally said no, they now say ‘we want to see the benefits of AI’.”
At Hugh James, this has evolved into specific tender requirements. Purcell describes receiving a major tender from an important existing client: “It’s the first time I’ve seen two very specific questions about AI where they’re not just asking us what’s our approach. They actually said: ‘How will you use AI on two specific types of work?’ In one case, it’s to improve and streamline a niche area of finance work. And the other one is all about improving quality and efficiency.”
This represents a fundamental shift. Clients are no longer asking whether firms are exploring AI. They’re demanding concrete explanations of how AI will benefit them on specific work, from cost and time perspectives.
A tipping point
For Taylor-Wright, the choice is clear. She’s not concerned that AI will take her job, but she recognises the competitive reality.
“I do think people need to get on board with using AI to help their roles, because if you don’t, the people who are using AI are going to have a massive advantage over the people who don’t. They’re going to be seen to be better at their jobs, because they have more time to focus on the big stuff that makes a difference, that the lawyers will notice.”
D’Arcy’s experience supports this. “It’s absolutely supercharged what I can deliver,” she says. “I knew that it wasn’t optional. So, you’re better off to be an early adopter and see the benefits and integrate it as quickly as you can into the way that you do your role.”
It is clear that for marketing and BD professionals, AI uptake is reaching a tipping point.
Before too long, maybe even the fee earners will follow.
James Lumley, freelance journalist, corporate writer, trainer and coach.
Log in
