January 26, 2024

Reduce Non-Billable Hours With These 3 ChatGPT Legal Prompts


Last year AI took off, and this year, there has been a growing number of articles suggesting ways that ‘legal prompts’ can help lawyers to provide their legal services. Trawl through the internet and you’ll find that AI now helps with drafting contracts, performing legal research on judicial practice, forming legal opinions, and much more. It is obvious that AI technologies have great potential to improve the legal services sector. 

However, in our opinion, AI is probably not going to take off in the legal industry the way most people might think: by taking over all legal services. In fact, we think that this initial adoption of AI technologies in the legal industry will not start with legal services, but instead help with other legal-related–but far less-complex–tasks. This is because AI technology has limited capacity to think like a lawyer. We’ll explore this in more detail in this article, as well as take a look at what AI currently can do for lawyers to help with the less logical and “out of the box” tasks in the legal field.

This article is written for–and therefore should be particularly useful for–the following legal providers:

  • those who engage lawyers in sales processes to make calls and prepare offers;
  • those who have specialized in a jurisdiction and area of law (and who collaborate with colleagues from other countries/adjacent areas of law to support complex cross-jurisdictional cases); and
  • those who are interested in optimizing and scaling their sales processes with the help of AI.

A quick reminder: none of this information is to be treated as legal, tax, or any other advice.

Let’s dive in!

Why AI technology is not ready to replace lawyers (yet)

  • The AI technology we have today (in particular, LLMs or large language models) still is rather hallucinatory in nature (they just make stuff up) when faced with  questions to which it doesn’t have answers to. One embarrassing example comes from a recent case in New York, where a lawyer used AI to help prepare for a case and…the AI tool invented completely fake precedents which the lawyer then used. AI wasn’t responsible for the lawyer’s poor oversight, but AI certainly was responsible for fabricating law.
  • In current LLMs there are still very limited contextual windows, which in turn limits the amount of information that AI can simultaneously “keep in mind” while in the process of solving tasks. If AI is given too much context or instructions it may start “forgetting” documents and instructions that were given to it at the very beginning, which is not suitable in the world of law. Work on complex legal cases requires lawyers to remember vast collections of detail covering regulatory documents, judicial precedents, experience of previous client cases, etc, and simultaneously pick out and process the relevant items to apply to the case at hand. Current AI technology just can’t do that. 
  • LLM are models that work well with the content they were trained on, but when the task requires thinking out of the box, they hit a wall. Tasks that require the LLM to use rules of formal logic or think in abstract categories result in rather unsatisfactory outputs from the LLM. Consequently, these necessary tasks must be executed by a lawyer; AI can’t replace the human brain in this capacity just yet. 

So, if there’s huge aspects of legal work that AI technology can’t do, and some work that AI shouldn’t do, should lawyers even consider adopting AI technology at the moment?

Yes. Complex billable legal works does not constitute 100% of a lawyer’s work week. For those non-billable (and some perhaps utterly mundane tasks) AI can already provide a great service.

📚Read more: 5 Ways AI Will Impact the Legal Profession and 6 Tips for Lawyers to Prepare for It

The problem with non-billable legal works

Non-billable work is what lawyers usually have the least motivation to do, mostly because this work does not directly affect the income of the legal practice and, consequently, the lawyers’ bonuses. However, ignoring this work often has an indirect effect on the law firm, which manifests itself in the following:

  • Non-billable hours are unpaid, yet lawyers receive monthly salaries so these non-billable hours translate to costs for the law firm.
  • The more non-billable hours (with the lawyer's salary unchanged), the higher the cost of the lawyer per hour, and as a result, the lower the marginality of the legal practice as a whole.
  • The lower the marginality of the law firm, the higher the prices that the law firm needs to set for its own services, and as a result, the less flexibility the  law firm has in pricing issues.

As law firms struggle to find the balance (and pay the price) of non-billable works, the emergence of AI may be just the solution legal practices need. 

How can AI help lawyers with their non-billable legal tasks?

All non-billable legal tasks can be conditionally divided into two categories:

  • sales-related tasks: including sales calls with clients, follow-up calls and emails, and  preparation works for offers and presentations.
  • administrative tasks: including filling out hour reports, performing team management responsibilities, and taking courses to gain qualifications and hone skills.

Legal provider’s sales processes without the help of AI

To understand how AI can help optimize and scale legal provider sales processes, let's start by analyzing how these sales jobs currently work for lawyers without AI.

Here’s how the lawyers sales workflow looks like to prepare a proposal for a new client (without using AI tools):

  1. Receiving a request from a client.
  2. Conducting an intro (sales) call with a client. Usually, on the call lawyer clarifies the details about the project (market, business model, etc.), clarifies the details about the legal request and analyzes the client request based on what part of the request lawyer can cover. Lawyer takes call notes manually.
  3. Preparing a follow-up after the call. This usually includes confirming the details received from the client, confirming the client's request for legal support, describing the next steps about preparing the proposal for the client. Lawyer prepares follow-up manually. Average time: 30-40 min.
  4. Sending a follow-up to the client.
  5. Preparing a proposal for lawyer's services. The description of the legal works needed to solve a client's request, preparing budget and time estimates for legal works and transforming the list of works and estimates into the final proposal usually go into this. Lawyer prepares the proposal manually. Average time: 40-60 mins.
  6. Preparing a request to a legal partner, which includes preparing a short summary about the client's project, describing the part of the client's request that doesn't fall within the lawyer's specialization, and preparing a request for time and budget estimates. Lawyer prepares a request manually. Average time: 20-30 mins.
  7. Sending a request to the partner and getting an estimate.
  8. Putting together all the information for a final proposal. Lawyer prepares a final proposal manually. Average time: 15-20 mins.
  9. Sending the final proposal to the client.

The average time spent on all these activities without using AI will be around 2.5 - 4 hours. 

How AI can improve legal providers' sales processes

Bearing in mind the status quo for most legal providers’ sales processes, let’s take a fresh perspective and consider how AI might already be able to improve certain processes today.

Here’s how the workflow we described above looks like when a lawyer’s using AI tools:

  1. Receiving a request from a client.
  2. Conducting an intro (sales) call with a client. Instead of taking notes manually, the lawyer uses an AI notetaker tool (like Fireflies.ai) that transcribes the call and prepares a summary.
  3. Preparing a follow-up after the call. Instead of doing it manually, the lawyer  prepares a follow up using a template, prompts, ChatGPT and the call summary generated by an AI notetaker. Average time: 5-10 mins instead of 30-40 mins.
  4. Sending a follow-up to the client.
  5. Preparing a proposal for lawyer's services. Lawyer structuring a proposal for the client based on the list of their services, template of a proposal, prompts and ChatGPT. Average time: 5-10 mins instead of 30-40 mins.
  6. Preparing a request to a legal partner. Preparing a request to a partner based on the summary using the list of partner's services, a template, ChatGPT and prompts. Average time: 5-10 mins instead of 20-30 mins.
  7. Sending a request to the partner and getting an estimate.
  8. Putting together all the information for a final proposal. Lawyer prepares a final proposal manually. Average time: 15-20 mins.
  9. Sending the final proposal to the client.

The average time spent on all these activities using AI will be around 30-50 minutes, reducing the time spent on all these tasks by 5 times compared to when AI tools aren’t used.

So, how can lawyers use LLMs, or AI, to get their processes to look like this? The answer lies in mastering legal prompts. Although not much is needed, there is a skill when it comes to working with AI tools. Learning how to write good prompts is the difference between knowing how to drive a car and thinking you know how to drive a car (before bunny-hopping out of the car park). Figure out your legal prompts, and you’ll be able to simplify and optimize your sales processes. 

What is AI prompting and how does prompting work in ChatGPT?

Put simply: prompting is a command or instruction given to a language model that contains the context of the use case and the goal to be achieved.

But what does that really mean? Compare it to giving a task to a colleague. A superficially stated task will result in a low-quality outcome, whereas a detailed task description will decrease the likelihood of needing iterations (or perhaps you won’t need any at all). Therefore, usually one prompt for ChatGPT is not enough to obtain a quality result. Sets of prompts are needed. Using an example of giving a task to a colleague, let’s map out the kinds of prompts that might be needed to create an effective set of prompts.

  • Set the context. Set the role of the AI and describe the situation, for example: "you are a paralegal with a specialization in commercial/contract law. The client plans to launch a joint venture with a partner, and before starting to share their confidential information, the client needs guarantees that this confidential information will not be disclosed to third parties. The client has approached you with a request to prepare a document that will protect the client from the disclosure of such confidential information”.
  • Set the task. Describe what needs to be done, namely “prepare a draft NDA”.
  • Give clear instructions for completing the task. For example,  state which NDA template to use and what should be taken into account in the process of customizing it. This might include providing a list of laws, court practices, applicable laws, the client's contact information, and document formatting requirements.
  • Explain what the final result of completing the task should look like. For example: state that “the output should be filled with client data, adapted to the applicable law and formatted accordingly to the NDA template, ready to be sent to the client for signature”.
  • Be prepared for iterations. Give comments to the AI on what to add/delete/shorten/expand on in the document.

AI prompts for legal emails, offers, and partner requests to cut non-billable hours in half

The following three prompts offer a means to help scale sales processes and halve the time spent on certain non-billable tasks.

  • Set 1: AI generation of follow-up emails
  • Set 2: AI generation of offers
  • Set 3: AI generation of partner requests

We’ll be exploring the first set of prompts, Set 1: AI generation of follow-up emails in this article. You’re welcome to copy and use these prompts. To receive the checklists for Set 2: AI generation of offers and Set 3: AI generation of partner requests you can sign up to our free webinar: How Lawyers Can Reduce Non-Billable Hours with AI.

During the webinar, we’ll be exploring  prompt engineering for lawyers (that’s just a fancy way of saying how to write prompts to help with legal tasks). We’ll break down these three sets of prompts and each attendee will receive a checklist for all 3 prompts to help get started with building your own sets!

Set 1: Prompts for AI generation of follow-up emails for clients after sales calls

Legal Nodes Ltd is not responsible for the use or outcomes of any prompts. Always use AI technology for legal works at your own discretion and with caution. This information is not legal advice.

What can you achieve with these legal prompts? 

Once you’ve perfected the flow of using legal prompts, you can expect to:

  • Reduce the number of non-billable hours for sales work and as a result, increase the margin of the legal practice.
  • Increase the capacity / throughput of lawyers in sales by helping to process more customer requests and add details to CRM systems faster. 
  • Help significantly increase communication speed and become more customer-oriented. Clients will no longer need to wait for prolonged periods to receive an initial offer or a follow-up from a lawyer. 

What do lawyers need to start creating AI legal prompts?

It’s important to prepare a knowledge base to support the creation and operations of these prompts. The knowledge base should contain the following:

  • follow-up templates for proposals 
  • service price lists
  • a database of legal partners
  • structured CRM with the client data
  • plus any other materials that are needed specifically in the sales process of your practice

Get 3 free prompt sets to halve non-billable hours and scale sales processes

Sign up for our free webinar on legal prompt engineering: How Lawyers Can Reduce Non-Billable Hours with AI. We’ll share our experience on how to build these prompt sets effectively and use them to substantially reduce non-billable hours. All webinar attendees will receive a checklist with the following 3 prompt sets:  

  • Set 1: AI generation of follow-up emails
  • Set 2: AI generation of offers
  • Set 3: AI generation of partner requests

Join us on Jul 5, 2023 at 5PM BST. Sign up now.

Sign up for the webinar on legal prompt engineering

Book a seat

Nestor is a Co-founder & Head of Web3 Legal at Legal Nodes. Having over seven years of legal consulting experience, Nestor loves working with innovative startups and Web3 projects, helping them navigate the regulations and scale on global markets.

Explore popular resources