Is It the Death of the Dashboard?

Estimated reading time: 2 minutes

In 2024, companies are leveraging AI agents to automate data-driven decisions, bypassing the need for specialized data science skills. In this interview with Tim Burke and Jennifer Davis of Arch Systems, they say these AI tools can handle straightforward tasks, significantly improving factory efficiency by addressing numerous small issues. This shift allows existing staff to use data effectively without extensive retraining. Both large and small language models are used to support real-time decisions on the production floor, integrating various data sources to create a comprehensive digital twin of the factory’s operations.

Nolan Johnson: Tim, to quote “Star Wars,” there seems to be a disturbance in the Force when it comes to data analytics.

Tim Burke: Yes. This reflects things coming directly from our customers. Until last year, people were saying, “Data is good, but I'm collecting a lot of it, and I don't know what to do with it.” This year, they’ve been saying, “Wait a second. I thought I needed staff with a new skill set to make better data-driven decisions. I didn't have that skill set and you’ve been telling me that's the biggest thing I need to hire for. Now I’m even more out of luck.” As we head into the new year, I hear, “A lot of the tasks that need data aren't that hard conceptually when I factor in this new rise in AI agents.”

The idea is to use a large language model (LLM), not necessarily in helping you write a document, but to help automate a task or to support a decision by using data. This is one area where these tools are getting pretty good.

What we're generally hearing is that an AI agent makes so many decisions that have a large delta on how efficiently I run my factory, even with standard, not special data models. There are a bunch of  fairly straightforward questions to answer from data with a little bit of prompt engineering to the agent, and they can add significant value.

We’re seeing customers believing that a previous blocker to their getting value from data was the skill set in their workforce to know how to use data to make decisions. That is still a challenge. There's an end-run around that. I don't have to wait. I can have an AI agent assist, fill in that skill set and uplevel my existing folks without requiring them to retrain as data science people. Now I've got a direct mechanism in which data helps me run the factory better. When it's coupled with an AI agent, I can turn that directly into the actions that make sense.

I know these were in the correct order here before my change, it aligns with the transcript, but I think it’s intended to read this way, from how he spoke. And, it reads more directly and doesn’t change the construct, so made a quick swap on order for this phrase.

To read the continue reading this article, which originally appeared in the November 2024 SMT007 Magazine, click here.