Uncovering High-Value AI Use Cases in Legal @ FutureLaw 25

Just over a week ago, I was invited back to the FutureLaw Conference in Tallinn, Estonia, organised by Valentin Feklistov and Alexander Bitskov, where I ran a workshop titled “Uncovering High-Value AI Use Cases in Legal”.

We had 60+ lawyers from over 20 countries in attendance. There was standing-room only at the workshop.

I wanted to share a write-up of the workshop for the benefit of those who weren’t able to make it in person.

Educate - What do I need to know about AI?

We spent the first part of the workshop introducing a few core ideas and concepts about AI.

If you’ve been following along with these newsletters, you should be familiar with them by now, so I won’t go into too much detail about these.

We defined our terms, explained the Technology/Application/Use Framework for making sense of AI, and laid out how fidelity is an important measure for use cases that gives rise to a spectrum across Microwaves, Pizza Ovens and Personal Chefs.

We outlined the two approaches to uncovering use cases; either from the domain or from the technology, and introduced the Legal AI Actions.

Fundamentals established, we moved into the interactive part of the workshop to focus on uncovering high-value AI use cases in legal practice.

Ideate - How do I come up with ideas?

This segment of the workshop started by asking participants to simply write down 4 problems they come across in their day-to-day work where they think AI could help make things better.

This was a silent brainstorm, so everyone had a chance to think carefully about some specific pain points and bottlenecks where AI could be used.

I then asked the participants to choose a single problem from the 4 they had come up with and cast it into the form of a Problem Statement using the following template:

WHEN [situation] WE WANT TO [motivation] SO THAT WE CAN [expected outcome]

Here’s an example of what a Problem Statement looks like in practice:

WHEN researching our obligations and duties to clients, WE WANT TO quickly and easily see all the different obligations and duties (certificates, etc.) in one place SO THAT WE CAN deliver and meet those obligations.

Problem Statements at the ready, we asked everyone to share their Problem Statements back to the group using an interactive Slido poll.

We had 30+ high-quality Problem Statements come through.

Here are some of the top picks:

1. WHEN ensuring internal compliance and aiming to support employees in understanding what to do without reading full codes of conduct and lengthy processes, WE WANT TO provide quick and easy access to essential information and guidance, SO THAT WE CAN ensure employees consistently act correctly without needing to consult the legal team for every annoying question.

2. WHEN we receive a counterparty's written brief and want to inform the client WE WANT TO create a qualitative insight, including an assessment regarding viability or the facts/arguments presented, SO THAT WE CAN show the client that we quickly address their matter and so that we can assess the amount of work required to reply to the brief.

3. WHEN we have to issue our legal opinion in a cross-border inheritance case, WE WANT TO quickly find (1) all applicable laws and case-law; (2) locate relevant existing court documents; (3) see our previous advice given, SO THAT WE CAN provide comprehensive, coherent, valuable advice.

With a short-interactive activity, we were able to source 35+ Problem Statements directly from practising lawyers of real, tangible opportunities for where AI could be applied in the space of 15-20 minutes.

How’s that for effective workshoping?

Prioritise - What idea should I focus on?

Armed with a shopping list of Problem Statements for use cases, we have to be realistic and appreciate that we can’t do everything on that list.

We need to be strategic and pragmatic about which use cases to prioritise and pursue.

How do we do that?

We evaluate each use case according to two dimensions:

1. Value/Impact

2. Difficulty

In order to evaluate Value/Impact of a use case, we can pose the following questions:

1. How many hours could be saved per month?

2. Is there a direct revenue upside to solving this problem?

3. Is there a risk or compliance gain from solving this problem?

4. Is there a competitive advantage gained from solving this problem?

5. Does solving this problem improve the employee experience?

In order to evaluate the Difficulty of a use case, we can pose the following questions:

1. Is there data available in a usable format?

2. Is there low process variance in how this work is currently carried out?

3. How likely is it to get stakeholder/partner buy-in for this use case?

4. Is there an off-the-shelf model/vendor that can solve this problem?

5. How easy/hard will it be to validate the value of this use case?

These questions are to help demonstrate the kind of thinking needed to assess a use case and aren’t an exhaustive list.

For each use case, we can score Value/Impact from 1-5 and Difficulty from 1-5 to help us create a clearer picture of what to prioritise first.

After scoring all of the use cases, we can plot them on a 2x2 Use Case Prioritisation Matrix to help visualise what to prioritise and how.

Plotting the use cases on the 2x2 Use Case Prioritisation Matrix will give rise to four quadrants for our use cases:

1. Quick Wins - High Value/Impact & Low Difficulty - Use cases that are easy to implement and also translate into high value and impact can be tagged as Quick Wins. These use cases should be prioritised immediately to create momentum and demonstrate initial progress with AI initiatives.

2. Low-Hanging Fruit - Low Value/Impact & Low Difficulty - Use cases that are easy to implement and translate into low value and impact can be tagged as Low-Hanging Fruit. These use cases can be implemented quickly but don’t create significant impact. They are no brainers.

3. Back-burner - Low Value/Impact & High Difficulty - Use cases that are hard to implement and also don’t offer much value and impact can be tagged as Back-Burner. These use cases should be ignored, left on the back-burner and revisited in the future.

4. Strategic Bets - High Value/Impact & High Difficulty - Use cases that are difficult to implement and also translate into high value and impact can be tagged as Strategic Bets. These use cases are harder to implement but pursuing them can be a means to create strategic differentiation since others will also struggle to implement these use cases.

AI Roadmap

With the use cases now organised on the 2x2 Use Case Prioritisation Matrix and tagged according to the type of opportunity, we are now ready to construct our AI Roadmap.

The AI Roadmap consists of Three Horizons:

1. Horizon 0 (0-3 months) - Focuses on delivering Quick Wins and Low-Hanging Fruit use cases to generate momentum, prove value to stakeholders and demonstrate early success

2. Horizon 1 (3-6 months) - Focuses on finalising and scaling Quick Wins and Low-Hanging Fruit while beginning to explore Strategic Bets in more detail

3. Horizon 2 (6-12 months) - With support and momentum established, shifting focus to more risky Strategic Bets use cases through running pilots, PoC and conducting rollouts where there are strong signals of adoption

With a quarterly cadence of surfacing new use cases and assessing progress, an AI Roadmap can be actioned, and we can start to take a more structured approach to AI within their legal teams.

We don’t need to talk about Gen AI, LLMs or even Agentic AI. We just need to get clear on what the opportunities are and assess them in a robust way to create and drive towards transforming a legal team with AI.

Thanks,

Uwais

Founder @ simplexico

PS - AI Roadmap Sprint - Get Your AI Roadmap in Just 10 Days

At simplexico , we are rolling our expertise in AI to help legal teams jumpstart their AI journey with our new AI Roadmap Sprint package.

In just 10 days, we’ll work with your team to educate them on AI, uncover practice-specific use cases through our signature workshops, align stakeholders, prioritise an actionable plan and deliver a tailored AI Roadmap.

It’s still early days, so we are looking for 4 budding legal teams to work with on our AI Roadmap Sprint across the summer and it’s first come, first-served.

Interested in learning more?

Comment ‘ROADMAP’ to this article and I’ll send across more details on the AI Roadmap Sprint package.