How can I use generative AI in my legal work?

Former lawyer turned legal engineer Catherine Bamford talks through her views on the top use cases for generative AI when carrying out legal tasks – and when it’s best avoided.

Catherine Bamford

We have all seen the headlines about generative AI (genAI) transforming industries overnight. But what does it actually mean for small and medium-sized law firms? Is it worth your time, money and headspace to explore it now, or should you wait until the dust settles? GenAI refers to tools that can produce text, summaries, or images in response to your instructions (known as prompts). For example – drafting an email or summarising a legal document.

From my work with firms of all sizes, I believe genAI can touch on almost every aspect of legal work – if used wisely. It does not mean it is the right tool for every task. Here are my thoughts on where it can help, where it cannot (yet), and how firms can get started safely and effectively.

How generative AI can help lawyers today

When we think about a typical legal matter, regardless of practice area, the workflow usually involves:

  • client onboarding and triage
  • due diligence and information gathering
  • drafting suitable documents
  • negotiating terms
  • summarising and reporting to clients.

Let us look at how genAI supports each of these.

Client onboarding and triage

Even before onboarding, genAI can enhance how clients find and contact you. For example, AI chatbots on your website (computer programs that simulate human conversation to answer questions and collect information) can triage enquiries, gather initial information and answer frequently asked questions. No client wants to wait days for a reply to check if you handle their type of case. Fast, clear triaging can give small firms a competitive edge by creating a professional, responsive-first impression. While many firms already clearly state their services on their websites, a chatbot provides an added layer of responsiveness by engaging visitors and helping direct them more quickly to the right support.

However, when it comes to formal onboarding, such as conflict checks, anti-money laundering (AML) compliance and identity verification, workflow tools with deterministic ‘if-then-else logic’ remain more reliable. ‘If-then-else’ logic follows clear rules and branching paths. For example, in an AML check, if the client has lived at their current address for fewer than two years, then ask for a previous address. If not, move to the next question. This ensures every detail is captured accurately, in the correct format, every time. It is helpful to picture these kinds of tools as following decision trees to get the correct result. This is different to genAI, which is ‘predictive’ rather than ‘deterministic’.

'GenAI can save hours of manual reading and note-taking. For small firms, this time saving can directly translate into improved client responsiveness and profitability’

Due diligence and information gathering

This is one of genAI’s real strengths. Whether it is summarising a long document, analysing transcripts from client interviews, or extracting and organising key details from bundles of PDFs, genAI can save hours of manual reading and note-taking. For small firms, this time saving can directly translate into improved client responsiveness and profitability.

Drafting suitable documents

Here is where a clear distinction is needed. There is a lot of hype around the capabilities of genAI which ignore those areas where accuracy and consistency are key.

If you are creating simple documents – such as board minutes based on meeting notes, first drafts of witness statements or drafting short client letters – genAI can produce quick, usable drafts for you to review and finalise. Think of it as having an always-available junior assistant. They can create those first drafts, but you can also ask genAI to sense-check clarity and phrasing before you send documents to clients or opposing counsel. For example, one firm I know uses genAI to turn board meeting notes into draft minutes instantly, rather than manually drafting them line by line.

Negotiating contracts

Negotiation is another area where genAI excels. Reviewing contracts received from the other side’s lawyers, marking up suggested changes and comparing drafts to your playbooks can be hugely time consuming.

GenAI tools can analyse the incoming document, identify deviations from your standard positions, and suggest redlines or alternative wording. This does not replace your judgement but it gives you a strong starting point saving significant time. A client comes to a lawyer for their expertise and professional judgement; this incredibly helpful assistant simply provides the lawyer with the information they need to apply it.

Summarising and reporting to clients

We often hear that clients find legal advice too long, too technical and hard to digest. GenAI is excellent at summarising complex advice in plain English, structuring it clearly so clients understand the key risks, recommendations and next steps. This is not just an efficiency gain – it directly enhances client service and satisfaction. But – take note of the points raised below concerning compliance and, in particular, client confidentiality.

Where genAI is not suitable for legal work

While genAI has huge potential, it is not suitable for everything.

Emotional support and strategic reassurance

Clients come to lawyers for professional judgement, but also for reassurance and support during stressful times. Many legal issues are real-life problems and clients want a real life human to talk to. No AI can replace the human empathy and strategic nuance that builds client trust. In fact, by using genAI to reduce the administrative burden, lawyers can create more time for these crucial human conversations.

Complex legal drafting with trusted precedents

For complex agreements with high levels of conditionality, such as leases, sale and purchase agreements, or detailed commercial contracts, genAI is not currently the best tool. Here, document automation with deterministic logic is far superior. For example, if you have a strong precedent lease, document automation will ask structured questions, such as: ‘Is there a rent review?’ and insert or remove the clause accordingly. The outputs are consistent, based on verified templates and do not require full line-by-line checks in the way genAI outputs currently do.

However, looking to the future, I believe genAI will gradually become more capable of handling complex drafting. Already, the best systems combine genAI with document automation, using automation for structured clause assembly and genAI for narrative drafting. This hybrid approach is something I am working on now and see as the future of legal drafting.

What about the compliance risks?

The biggest compliance concern with genAI is client confidentiality. Never use public-facing versions of ChatGPT or other free tools to process client data. Doing so risks breaching confidentiality and professional obligations.

It is important to understand how different genAI tools use and process data. Some tools may store prompts and responses to improve their models, while others offer enterprise-grade options that do not retain or reuse data. Firms should carry out due diligence, just as they would with any third-party supplier. This means understanding what a tool does, where the data is stored and whether it meets professional and regulatory requirements.

Thankfully, solutions are now widely available. Firms can use enterprise versions of genAI. These versions are designed for business use, offering enhanced security, data privacy and administrative controls. Examples include ChatGPT Enterprise and Microsoft Copilot. They ensure that inputs and outputs are not used to train the model, making them more suitable for legal work. Firms can also invest in legal-specific genAI tools (many under £100 per user per month), which are designed to keep data safe and compliant.

'Never use public-facing versions of ChatGPT or other free tools to process client data. Doing so risks breaching confidentiality and professional obligations'

Practical first steps for small firms

If your firm has not yet explored genAI, here is how to start:

  1. Identify the problem you are trying to solve: start by thinking about specific pain points in your daily work. Are there tasks that are repetitive, time-consuming, or hard to delegate? This will help you focus your efforts and choose tools that genuinely add value.
  2. Educate your team: run a workshop or recommend a short online course to build understanding of what genAI is, what it can (and cannot) do and how to use it safely.
  3. Develop an AI policy: set clear guidance on which tools staff can use and for what purposes. The SRA has useful pointers to help you draft your policy.
  4. Start small: encourage staff to try using genAI tools for safe tasks (for example, drafting non-confidential internal emails, summarising non-client documents) to build confidence and see the possibilities.

Many of your lawyers are probably already experimenting with tools like ChatGPT or Claude. Bring these conversations into the open, set safe parameters and share examples across your team. Regular team catch-ups to discuss how people are using AI can accelerate learning and build firm-wide capability quickly.

Addressing fears about quality and expertise

It is natural for lawyers to fear that genAI might reduce quality or undermine their expertise. My advice is:

  • Test and verify: try it out for yourself to understand where it adds value and how to check outputs effectively.
  • Explore multiple tools: consider experimenting with more than one version of genAI. General-purpose tools like ChatGPT or Claude can be helpful for internal tasks, while specialist legal tools may offer better results for drafting, summarising or reviewing contracts.
  • See it as an assistant, not a replacement: genAI is a tool to help you deliver your expertise to clients more efficiently and effectively.

The goal is not to replace lawyers, but to give them more time to focus on high-value, strategic and client-facing work.

‘The goal is not to replace lawyers, but to give them more time to focus on high-value, strategic, and client-facing work’

Looking ahead – the mindset needed

To make the most of genAI, lawyers will need a curious, willing-to-learn mindset to experiment and find what works. They will need an AI-first approach – whenever you start a task, ask: could AI help me do this better or faster? And an open mind about business models – genAI will enable new ways to serve clients and create revenue. Those who embrace this change will thrive, while firms that do not adapt may find it harder to remain competitive over time.

Final thoughts – should you invest now or wait?

My advice is to start now and experiment. You do not need to make huge investments. Start small, educate your team, develop safe processes and learn together. The firms that begin this journey now will build the skills and confidence to harness genAI effectively in the months and years ahead and ultimately deliver better, faster and more client-focused services.

About the author

Catherine Bamford is the CEO and Founder of BamLegal, a leading legal technology consultancy. As a former real estate lawyer turned legal engineer, she now advises law firms and in-house legal teams globally on implementing technology to transform the delivery of their legal services. Catherine is passionate about making legal work more efficient, effective and client focused. Please note, the views expressed in this article are Catherine's.

Share on social