How Banking Teams Should Actually Be Using AI

Lots of exciting news going on in the AI space. Ron Shevlin just did a What’s Going on in Banking report, and one of the key themes was AI. Check that out in the comments below.

But we want to talk a little bit more about some of the practical applications for day-to-day use of AI in the organization as you use it today. So I’ve asked Jack Strauss to join us.

Jack, what are some of the key takeaways, key tools or ways to maximize the use of things like ChatGPT and Claude?

You want to focus on being practical, like you said, first. That’s going to lead you to three different things. I think the first is going to be personalization of the model. Going into the model and allowing it to get tailored outputs to your job description as an employee will help you create more value.

So go into the core system, into the profile of the LLM. Tell it who you are and what you do.

Yeah, exactly. That’s going to help you get more tailored outputs that can really, like I said, help you create more value. They’re going to apply to what you’re doing and what you’re trying to get as an outcome.

Exactly. The second one is going to be effective prompting. I think this is where a huge skill gap is because, you know, we’ve talked about this a lot, and it’s about wrestling back and forth with the model.

It’s going to save you more time upfront. Instead of spending hours going back and forth with the model, now you’re spending five to 10 minutes upfront telling the model the context of the situation, the task and the role that you need it to play.

So take your time with your initial prompt. Put as much information and context around that, in conjunction with your personalization. You’re going to get a much better output instead of having to reprompt and reprompt and reprompt.

Exactly.

All right. Number three.

Number three is going to be project folders. Project folders are going to help you be more distinct with all of your information and keep it organized so that you can continuously follow up on a project or a process that you have.

For us, for example, we’re using our client projects in our project folders. If you’re an employee at a financial institution, maybe it’s a process or a workflow that you continuously go through where you’re adding a project folder and continuously putting in this information.

So it’s customized to a specific project you’re working on. If I’m doing a process flow in deposit ops, but I’m also doing a process flow in lending...

Exactly.

If I put those in separate project folders, I’m going to get better outcomes from both of those because it has the context of maybe meetings that I attended or other content that’s out there.

Yeah. So it’s not only going to help you become faster, which is what everybody thinks about when they think about using AI, but it’s also going to help you be more detail-oriented.

Yeah. How are you all using AI today and getting the most out of the LLMs?

Three things to think about. Make sure it’s personalized to what you do and what you’re trying to accomplish. Make sure you’re prompting effectively. Invest the time upfront so you’re not reprompting all the time. Then leverage project folders for specific outcomes to make sure you’re getting the best possible results based on the work that you’re doing in that specific instance.