Approaches to defining and documenting Gen AI use cases

Introduction

This is a part of the proposed output from the Litig AI Benchmark Initiative that started in July 2024.

The aim was to come up with a simple and effective way to help define and document use cases for legal technology products and services that have an element of GenAI. Within the Litig AI Benchmark Working Group it quickly became clear there are many ways to do this, from creating very high-level use cases of one or two sentences through to mini business case documents that contain more details and may have a wider purpose. And then there is everything in between.

We are conscious the range of individuals and organisations who may pick this up is very wide. WHAT you want to do WHERE, with WHO, WHEN, WHY and HOW needs to work for you, your stakeholders and align with the objectives of the session(s) and your organisational needs.

Therefore, with flexibility and adaptability in mind we have set out below two examples of use case approaches that you may want to adopt and amend as required.

Use case definition – example approach 1 (high-level)

Some organisations, or teams within the same organisation, may be more mature than others in terms of knowing what use cases are, and how to create and apply them. This example is a good starting point to create high-level use cases.

Model/rubric wording:

[WHAT] – [WHO] in [WHERE] are [WHAT] by [HOW] to [WHY].

Model/rubric explanation:

[WHAT – the high-level categorisation of the type of AI use case, e.g. CREATE, SUMMARISE, COMPARE etc.] - [WHO - role] in [WHERE – team / department / office / jurisdiction] is / are [doing WHAT], [HOW/WHAT are they using to do it], and [WHY- the value proposition].

Examples of where the simple approach could be useful include:

engaging with clients and colleagues to show them what information might go into making up a use case, and letting them use it to consider and create some use cases they find interesting, and
preparing for or working with suppliers of Gen AI tools who are keen to hear what use cases may be interesting or relevant to your organisation.

Some further examples of use cases derived using the simple method are set out below:

1 - [SUMMARISE] - [Lawyers] in [London Commercial Contracts Team] are [extracting key data from contracts] by [using LLMs and prompting] to [accelerate delivery and enhance consistency of client due diligence reports].

2 - [AUTOMATE] - [Securities lawyers] in [Singapore] are [extracting information from trade term sheets] by [applying Named Entity Recognition models] to [automatically populate pricing supplements and accelerate trade times].

3 - [CREATE] - [M&BD team members] in [UK] are [creating relevant and targeted pitch responses and content for client RFPs] by [using LLMs and prompting] to [tailor the content, format, and language to meet client needs and improve the win rate for pitches].

4 - [COMPARE] – [Legal ops] in [Global team] are [reviewing entity step plans across jurisdictions] by [using LLMs to flag structural or timing inconsistencies] to [pre-empt approval delays and make audits easier].

Use case definition – example approach 2 (more detailed use cases)

This approach may be appropriate when you are considering ause case in more detail (or drafting a business case). It includes the opportunity to describe the value proposition, inputs and outputs, personas, and what good could look like. Again, how you use it and how much of it you use may be influenced by factors including your requirements, ways of working, governance, and strategy.

The template is set out below and examples of a completed template areincluded at the end.

Example template for detailed GenAI use cases

1. Use Case Reference Details:

Author(s): [ ]

Date: [ ]

Use case ID & Version: [ ]

[NOTE: the above table contains basic use case meta data to help with versioning.]

What you want to do:

[very high-level statement]

2. Business outcome:

[brief description of what the business wants to get out of this use case]

3. Overview:

[more detail about what you want to do and how you will do it].

4. Data/knowledge sources:

[list of the relevant data and/or knowledge sources]

5. Value proposition:

Examples of anticipated benefits for this initiative include:

[list of the anticipated benefits]

6. Proposed technology (if known):

[describe what technologies, elements, and/or methods may be involved]/ [leave blank if unknown]

7. Use case personas:

Persona 1 – [role/job title]

Use: [what they will do for the purposes of this use case]
Objectives: [describe what they stand to gain]

Persona 2 – [role/job title]

Use: [what they will do for the purposes of this use case]
Objectives: [describe what they stand to gain]

8. Pilot user fears / expectations:

[Complete as part of pilot planning]

9. Next steps:

[Business decision, e.g., Proceed / Iterate / Pause / Stop etc.]

END OF TEMPLATE

Peer feedback

The following optional sections have been suggested by peers. It is great to get this input, and they are added for the purposes of collaboration and knowledge sharing.

1. Risk and readiness:

[An opportunity to name assumptions or dependencies early — whether around data, user behaviour, or handoff points]

2. Measure of success:

[How success will be measured, e.g., time saved, accuracy, trust, adoption, or handover quality]

Table of examples of detailed GenAI use cases

	Example 1 (Learning & Dev)	Example 2 (Private Practice)	Example 3 (In-House)
	Author: A Turing Date: 1 June 2025 Use case ID & Version: UC0001v003	Author: KM Team Date: 10 July 2025 Use case ID & Version: UC0082v002	Author: Risk Team Date: 11 August 2025 Use case ID & Version: UC0010v007
1. What you want to do	Create training materials using Gen AI.	Use GenAI to support faster and more consistent contract drafting through a clause bank assistant.	Deploy GenAI as a legal Q&A assistant to support business stakeholders with contract interpretation queries.
2. Business outcome	To use LLMs to help with preparing training materials and plans in the early careers team.	To improve drafting efficiency and quality across transactional teams, enabling lawyers to produce first drafts and mark-ups more quickly and consistently.	To reduce legal bottlenecks by empowering business teams to answer low-risk contract interpretation queries independently.
3. Overview	We want to use LLMs to help the early careers team plan and deliver training and work experience. This may include tweaking already drafted content to suit a different audience, ideas and content generation, brainstorming or asking questions on a topic/content. We will use SMEs from the early careers team in the pilot.	We want to pilot a GenAI-powered assistant that retrieves appropriate clauses from an internal clause bank and suggests edits based on client templates, sector, and commercial context. The tool would support fee-earners in generating initial drafts and reviewing mark-ups, particularly on high-volume contracts like NDAs, supplier T&Cs, and software licences.	We aim to pilot a GenAI assistant that can provide plain English explanations of specific contract clauses or defined terms. The assistant will be trained on a controlled dataset of frequently asked questions, internal policy documents, and standard contracts. It will be positioned as a self-service tool for commercial and procurement teams.
4. Data/knowledge sources	· Draft notes or publicly available information on a topic. · Previous training plans or materials.	· Internal clause banks / precedents. · Matter-specific templates and mark-ups. · Fallback positions and standard comments from playbooks. · Knowledge team guidance notes.	· Internal contract templates (e.g., NDAs, MSAs, SOWs). · FAQs maintained by legal. · Internal commercial policy documents. · Legal team playbooks.
5. Value proposition	· Be quicker to plan training as preparation time will decrease. · Be able to produce more content. · Be able to better tailor / produce more content tailored for specific audiences.	· Reduce drafting time for common agreements. · Promote consistency across client-facing documents. · Enhance junior fee-earner confidence in clause selection. · Free up senior lawyer time for more strategic review work.	· Reduce legal team time on repetitive queries. · Faster response times for commercial teams. · Improve consistency in contract understanding. · Increase confidence in low-risk decision-making.
6. Proposed technology (if known)	We will use LLMs and Gen AI to create and summarise content, plus [TOOL/SUPPLIER] to transform this to presentations.	GenAI platform with clause retrieval and comparison capabilities (e.g., [TOOL/SUPPLIER] or [TOOL/SUPPLIER] integration).	GenAI chatbot interface integrated with internal [TOOL/SUPPLIER] or [TOOL/SUPPLIER].
7. Use case personas	Persona 1 - Subject Matter Expert Use: Creation / editing of training documents. Objectives: Spend less time on content creation/editing (so they can do other things/create more content) without hampering quality of output. Persona 2 – Learning and Development team member: Use: Reporting and understanding usage. Objectives: Have clear readily accessible usage data and other analytics.	Persona 1 – Associate (1–3 PQE) Use: Drafting and revising contracts using clause suggestions. Objectives: Work more efficiently and independently, improve quality of drafting. Persona 2 – Senior Associate / Partner Use: Reviewing draft outputs and aligning with client preference. · Objectives: Reduce time spent on mark-ups and ensure consistency across teams.	Persona 1 – Business Stakeholder (e.g., Sales / Procurement) · Use: Asking questions about specific contract clauses or obligations. · Objectives: Faster answers without needing to wait for legal input. Persona 2 – Legal Counsel Use: Maintaining oversight of the assistant’s scope and accuracy. · Objectives: Reduce time spent on low-value work while managing legal risk.
8. Pilot user fear/expectations	[To be completed in pilot planning, e.g., content/tone of output, over-reliance on output].	[To be completed in pilot planning, e.g., accuracy of outputs, trust in clause suggestions].	[To be completed in pilot planning, e.g., concerns around legal accuracy, misuse beyond low-risk questions].
9. Next steps	[Business decision, e.g., Proceed — identify suitable training topics and draft materials for the pilot, select SME users, and set evaluation criteria (e.g. time saved, output quality, user satisfaction).]	[Business decision – e.g. Iterate - select pilot teams (e.g. Commercial, Tech/IP), collate standard clauses, define evaluation criteria.]	[Business decision – e.g. Proceed to scoping data sources and define sandbox pilot boundaries].
[and other fields you want to add]	[ ]	[ ]	[ ]