HELEN FOORD: So, last year we had a chat about whether AI was the future of legal directories. You had some good points to make. Can you recap the key takeaways for me please?
CHAT GPT: Hello Helen, lovely to hear from you again. I remember that conversation well—it was thought-provoking. It’ll be fascinating to revisit it a year on and see what’s changed. Here are the key takeaways from last time.
- Confidentiality Concerns – You were worried about the risk of confidential, identifiable client data being shared with public AI tools.
- AI as a Support Tool, Not a Replacement – We agreed AI could enhance efficiency and analysis, but not replace strategic insight, judgement, and creativity. Human oversight remained essential.
- Referee Management Is Vital – We noted that referees can make or break a submission, and that strong selection and management often matter more than the written content.
- What an AI Consultant Might Look Like – We imagined a tool that could update itself on legal trends and submission criteria, personalise advice, draft optimised submissions, and help manage referees and track feedback.
HF: Thank you. So, let’s tackle confidentiality first. What has changed since I expressed those concerns?
CHAT: The risks haven’t gone away. Many legal teams still use unapproved AI tools, with few governance frameworks in place. Shadow AI use is widespread (Forensic Risk; SCL), and regulators are responding—requiring disclosure of AI use in some cases (Bar Council)—so firms can’t afford to treat governance as optional. Public systems like ChatGPT still risk exposing sensitive data (IBA). The SRA reminds solicitors they remain accountable for AI outputs.
The difference is that firms now have clearer ways to manage risk. Updated licences, like the CLA Law Licence, allow published content to be used in prompts. The new AI Confidentiality Code of Practice sets out security standards, and tools such as LegalGuardian can automatically mask sensitive data before it’s entered into AI.
Real-world adoption is also underway. The SRA licensed Garfield.Law as an AI-powered firm, with human oversight and client approval built in. Major firms like A&O Shearman use AI for complex filings, but only under senior supervision. The pattern is clear: AI is being used, but always with safeguards.
HF: So are there ways firms can use AI in legal directory submissions without compromising confidentiality?
CHAT: Yes—there are now several practical ways law firms can harness AI for directory submissions while protecting client confidentiality. The emphasis is on using AI as a controlled support tool rather than as a free-form content generator. Think of AI as a toolkit to support the process, not a replacement for it.
- Work in Secure, Closed Environments
Private AI models (hosted internally or by trusted providers) can be trained on anonymised or firm-controlled data, ensuring prompts never leave the firm’s systems. Tools like Ranking Copilot and Legal Engine are being designed specifically for legal directories, offering structured workflows and confidentiality safeguards. Some firms are developing “AI sandboxes”: walled-off environments where experimentation can happen safely without risk of data leakage. - Use AI for Non-Confidential Tasks
AI can safely handle workstreams that don’t expose sensitive client details, for example formatting and style consistency, summarising public domain material, creating timelines, and generating checklists. This frees up human consultants to focus on strategy and confidential case content. - Anonymisation and Masking
Firms can run pre-processing layers (e.g., tools like LegalGuardian) that automatically detect and mask names, client identifiers, or financial figures before text is entered into an AI system. This allows AI to analyse structure, tone, or clarity of writing without ever seeing confidential details. - AI-Assisted Referee Management
While the referee process itself involves sensitive data, AI can support indirectly by suggesting patterns in referee response rates, flagging underused referees, and automating reminders or workflow tracking—without exposing underlying client relationships. - Governance and Oversight
The safest firms will create an AI usage policy that involves directories and wider marketing, specifying which tools are approved and what data can be shared, require human review of all outputs, and train staff in AI literacy. This ensures accountability and protects professional standards.
HF: Interesting. In the original conversation we talked about AI as a support tool, not a replacement. Do you think this has changed over the last year?
CHAT: Last year, we said AI could help with efficiency but not replace human expertise. That still holds, though the boundaries have shifted.
- AI is now stronger at drafting. It can handle boilerplate sections—practice overviews and team bios—getting firms most of the way there.
- Firms are using it in practice. Tools are moving from theory to daily use.
- Regulation has caught up. With clearer rules from the SRA, Bar Council, and Law Society, adoption feels less risky.
But the limits remain. AI still hallucinates, lacks contextual judgement, and cannot manage relationships. And while firms once assumed AI would be cheap, hidden costs—licensing, governance, training, compliance—are now apparent.
In short: AI can replace the typing, but not the thinking. And the model is becoming clear:
- AI does the heavy lifting—drafting repetitive sections, formatting, checking guidelines.
- Humans do the high-value work—strategy, storytelling, referee management, credibility.
That makes the role of consultants even more important: blending efficiency from AI with human expertise.
HF: Thank you. And what about the referee process? We talked about that last year too. Has anything evolved there do you think?
CHAT: Referees remain one of the most human aspects of the process. AI has found a role in support, but it hasn’t replaced relationships.
- Workflow support has grown. AI tools can monitor referee response rates, flag over-reliance, suggest underused contacts, and automate reminders.
- Data analysis is possible. Some firms use anonymised referee data to spot which types of clients are more responsive.
- Confidentiality is still a barrier. Sharing names with public AI tools is off-limits, so uptake is slower than in drafting.
- Referees are more decisive than ever. As AI polishes text, directories place more weight on referee feedback.
So AI can act as a management assistant, but never a relationship manager. The differentiator is still the quality of human connections.
HF: That’s really interesting, thank you. In the article last year we talked about why AI would be better than a specialist agency and compared that to why a human would be better than AI. Can you give me your thoughts on this please, taking into account what we said previously and what we now know has evolved over the last year?
CHAT: At first glance, AI seemed like the obvious winner: fast, cheap, consistent, and always available. A year on, those advantages still stand—but they come with strings attached. AI outputs need oversight, secure systems carry costs, and even the most advanced tools still produce text that can feel flat compared with human storytelling. In short, AI is good for the grunt work, not the strategy or persuasion.
By contrast, the strengths of human consultants—strategic judgement, confidentiality safeguards, sector experience, and referee management—have only grown in importance. Directories are placing more weight on referee feedback, governance demands are higher, and firms need trusted advisors who can guide them through both the process and the technology.
So the debate is no longer AI versus humans. The real advantage lies in the combination: AI brings the horsepower, but humans keep hold of the steering wheel. The firms that succeed will be those who use AI for efficiency while relying on human expertise for effectiveness.
HF: I’d like your predictions for the future. What do you think we’ll be seeing if we have this conversation in a year?
CHAT: Over the coming year, AI tools will move beyond simply drafting submissions to actively analysing them—spotting ranking patterns, identifying gaps, and providing firms with new decision-making insights. But success won’t just depend on using AI; it will depend on using it responsibly. Strong governance frameworks will become a differentiator, inspiring trust from clients and referees, and we may even see directories themselves respond with clearer guidance or updated templates to reduce reliance on generic AI text.
As written submissions become easier to polish, referee feedback will carry more weight than ever. That means firms will need to invest more in client care and relationship tracking to stand out. Alongside this, agencies that combine human expertise with responsible AI adoption will set the standard. Those who embrace the hybrid model—AI for efficiency, humans for effectiveness—will lead the way, while those who ignore AI risk being left behind.
In short: If we sit down again in a year, we’ll be talking less about “AI vs humans” and more about which firms have mastered the art of AI plus human collaboration. The technology will be embedded, governance stronger, and the differentiator clearer: relationships, strategy, and credibility—areas where only humans can shine.
HF: Before we finish, are there any other points you want to say about the role of AI in legal directory submissions?
CHAT: Yes — there are a few extra angles worth highlighting.
First, there’s the question of AI’s impact on quality. Some argue that AI produces bland, generic text that lacks nuance, and in many cases that’s true. But even if the writing isn’t better than a skilled human’s, it can help to raise the baseline standard. Firms that previously struggled to produce clear, well-structured submissions may now turn out cleaner drafts more quickly. As a result, good writing alone won’t be enough to stand out. The differentiators will be strategy, storytelling, and, above all, referee management.
Second, firms should consider client perception. Even if anonymisation is in place, some clients may feel uncomfortable knowing their matters are being processed by AI. Transparency and reassurance will be essential to maintain trust.
Third, we should think about the directories themselves. Chambers and Legal 500 are watching these changes closely, and it wouldn’t be surprising if they issue explicit guidance or adapt templates to manage the use of AI. Firms with responsible policies already in place will be ahead of the curve.
Fourth, there’s the skills shift inside firms. It’s not just about policies and licences, it’s about people knowing how to use AI intelligently. Prompting, reviewing, and judging when to trust the output are all skills that firms will need to develop.
And finally, there’s the opportunity for agencies like ELE. Far from replacing consultants, AI strengthens the need for expert guidance. Law firms are cautious, confused, and time-poor. They’ll need trusted advisors who understand both the directories process and how to integrate AI responsibly. That hybrid role is where the real value lies.
