What On Earth Is A GenAI Use Case?

Read on for an easy-to-follow guide to finding a GenAI use case.

What On Earth Is A GenAI Use Case?
Photo by Louis Reed / Unsplash

‘Use case’ or ‘GenAI use case’ are terms that are thrown about a lot these days but what do they actually mean, and how do you find one? In this simple guide, I offer a clear and repeatable set of steps to finding the most impactful GenAI use cases for your organisation.

In the simplest of terms, a ‘use case’ for generative AI is a situation where there is a clear opportunity and benefit to replacing human intervention with AI intervention.

When successfully explored, the result is such that outputs previously generated by a human is effectively replaced by outputs generated by an AI.

So, where do you even start to look for these opportunities?

First things, first.

Step 1: Identify a pain point.

What is causing the most friction for your organisation? Take a real close look around you and make a list.

Here are a few examples:

  1. Waiting for legal review on customer contracts;
  2. Answering vendor questionnaires;
  3. Waiting for compliance to perform routine checks;
  4. Addressing infosec questions about;
  5. Preparing documentation for audits;
  6. Tackling employment and benefit questions;
  7. Dealing with questions about existing contracts;
  8. Answering repetitive questions about the same agreement, the same policies, the same legislation;
  9. Reviewing DPAs;
  10. Reviewing NDAs;
  11. Answering questions from customers;
  12. Dealing with partner or customer disputes.

Step 2: Understand the root cause for your paint point.

Is the pain point inserting friction because friction is necessary to protect the business and mitigate risk? If so, discard this as a use case (for now) and select another paint point.

Is the pain point being caused because of an excess of red tape? If so, remove the red tape and simplify the process wherever possible, for example by simplifying the chain of ‘approvals’ or checks needed, or by tackling the problem at source (for example by moving to online terms and dropping deal-by-deal contract negotiation).

Is the pain point exacerbated due to an inefficient use of resource? If yes, move to Step 3.

Step 3: Gather your documentation.

Now that you’ve chosen a paint point to tackle, you can carry on with the groundwork.

A good use case for GenAI is one where there is solid and reliable ‘context’ for the AI to draw from. By now, we’ve all heard of the widely used euphemism for GenAI which is “Garbage In, Garbage Out”.

Good documentation is important and relevant here because the AI needs to have access to all the relevant knowledge it needs to effectively substitute the human in the loop.

That said, ‘good’ documentation looks a lot simpler than you might think. Ideally, you want documentation that is clear, concise and complete.

For example:

  1. A log of answered legal, privacy, compliance or info sec tickets;
  2. A Slack history on the Legal Support channel;
  3. A suite of policies;
  4. A contracting playbook;
  5. A Frequently Asked Questions document tackling the most common queries;
  6. An employee handbook.

If you spot any gaps in documentation, that’s absolutely fine and shouldn’t deter you from going ahead with your use case. Many organisations that I’ve seen roll out GenAI successfully actually started with very sparse or incomplete documentation which they fleshed out over time. This is actually much more effective that overloading your AI with excessive amounts of documentation. An excess of context, especially where the context is not ‘clean’ and contains conflicting information, can adversely affect the output quality of your AI generated responses.

Step 4: Give it a go!

Sure, the prep work is exciting but even more exciting is seeing GenAI in action answering real time questions for real users. Testing and roll out warrant a guide in their own merit but the key point I want to make here is that the sooner you get GenAI deployed, the quicker you can learn, adjust, iterate and fast track the path to success.

What does success look like in this context? It’s that moment where you effectively automate human tasks in a way that’s so seamless it still feels like the business is interacting with the expert teams direct, when in reality they are receiving support just as effectively from an AI but at the speed of generation response time!

Got your use case already and looking to roll out AI? Ask us for tips here!