An interview with Mike Branch…

One the hottest topics in the business world in 2024 has been intuitive AI (Artificial Intelligence). While it can sometimes be difficult to separate the reality of this technology from the hype surrounding it, there are a growing number of use cases where it’s being applied and used effectively. In the fleet management space, intuitive AI is being used to deliver faster insights, reliable data protection and easy-to-use interfaces, making it a highly valuable tool for many fleet operations. Geotab Ace is prime example of this new frontier in AI-driven data analysis that’s really helping fleets to take their operations to the next level. To find out more about how it’s doing that, Autosphere sat down with Mike Branch, Geotab’s Vice President of Data & Analytics. Here’s what he had to say:

Autosphere: What do you think are some of the biggest challenges today for fleets when it comes to analyzing and purposing data?

Mike Branch: I think one big thing to consider is the volume of data. There’s just so much of it coming at fleet managers today, who operate all kinds of different vehicles in all kinds of different industries. If we look at Geotab, we have more than 50,000 customers, that operate in everything from construction to government operations, to rental fleets. And while there are commonalities such as fleet safety, there are specific questions that are domain specific when it comes to data. And what we’ve been finding, is that a one-size-fits-all dashboard just doesn’t work for every use case. Therefore, what ends up happening is that people create a little dashboard for different kinds of operations, such as construction, government fleets and rental vehicles. The challenge for many fleet managers is the ability to make sense of all this data, plus when you want to dive deeper into the analytics, you often require a data professional to help you and often these people are hard to find and expensive to hire. There’s also the time it takes to cycle that data which adds further complexity. In that context, the ability to reduce time to gain the insights you need and get to that piece of information that you want is essential, but as we’ve said earlier the sheer volume of data that’s out there and the specific customer use cases where that data needs to be analyzed effectively often presents a real challenge.

AS: What were some of the factors that led to the creation of Geotab Ace and the partnership with Microsoft?

MB: Our view was that people would want to have a conversational interface with their fleets, which is why we started our Project G last year, in beta. We ran that with a handful of customers. And a major difference between project G and Ace is that project G only understood a handful of data, which wasn’t enough to satisfy the needs of our customers. After running through that Project G pilot program, it became abundantly clear that the demand for a conversational interface with the vehicle fleet was there but that a data scientist was needed to help fleet managers understand and analyze this data. And that’s really what drove us to creating Ace—driving down time to insight. If you’re able to do that, then you’re able to empower people to actually leverage and use this data without having to wait a day or a week or for somebody to do the analysis for them. And that’s where you can really impact change, whether it’s the fleet manager, the CEO or even someone in a role such as Chief Sustainability Officer who is wanting, for example to understand and clearly see the organization’s carbon reduction goals. With Ace you not only have the ability to use the data without requiring a data analyst, you also can leverage AI technology such as asking questions through Chat GPT. Essentially you can ask Ace a question and it will provide a response by performing queries on the back end, pulling data from across thousands of different vehicles to get the insights you need very quickly. Today there are multiple different language models from different sources and when developing Ace we tested a number of them. Our conclusion was that for our requirements, Chat GPT 4.0 was the most suitable and best performing. We liked the relationship with Microsoft, the fact that they were hosting it through OpenAI and that there were a whole host of security measures and services wrapped around it that enabled us to have some guarantees when it came to response times.

AS: Can you tell us a little about how Geotab Ace works?

MB: At Geotab, we do a lot of work, not only in the private sector, but also the public sector. Security is a huge component of everything that we do, and privacy is of the utmost importance. We therefore wanted to make sure that we maintained those standards when it came to Geotab Ace All of the data that’s behind the scenes still sits in the Geotab ecosystem. When we ask Ace a question, effectively what it’s doing is creating a code for us that we run behind the scenes, with a number of security measures that we put in place. Besides making sure it’s executing the right codes and the right permissions; we also infuse the language model with a lot of the Geotab knowledge. The goal is to make it understand what safety means, how utilization is calculated, to make it understand the internal workings of how we’ve connected all of our data, and we send that off to the Large Language Model (LLM) without the customer’s data. And then it replies to us and gives us the query that we would run against our ecosystem to obtain that. The important thing to remember however, is that when you ask Ace a question, it isn’t going to get it right every single time. When you ask same question there might be some slight differences. We put a lot of work in a testing framework to make sure that the answers generated are as repeatable as possible. We provide the Microsoft Structured Query Language (SQL) code that gets generated, so if we ask a question, and it interprets something, we also have the ability to check that. For example, if you have a bit of a coding expertise, you can look at it and see that it didn’t compute an average the way it should have, and, when you don’t get the response you’re looking for, you can rephrase the question. Ultimately, we’re trying to be as transparent as possible and alongside Ace, have released an AI guide to show what we’re doing and internally, working as much as possible to ensure Ace doesn’t answer questions that it shouldn’t. Another important aspect with Ace is that if anybody is using our MyGeotab platform, effectively, they are translating their line of business to MyGeotab. By doing this, they are essentially creating geo fences for customer locations, such as distribution centres. Customers can also create specific rules that are important for their fleet, such as idling of vehicles within a specific zone and receiving notifications when that happens. By doing so, they’re conveying something about their business that goes beyond merely understanding their fleet operations and Ace is able to take the telematics data from the vehicles and correlate it with other metadata that customers are inputting. The result allows Ace to put together some very detailed and advanced reading solutions.

AS: What are some of the biggest benefits fleet customers are likely to see from using it?

MB: The biggest one is putting power in the hands of the people using it, so decisions can be made faster and more decisively. For example, if you’re looking at electrifying your fleet, you might have been trying to pull together all sorts of different datasets which is costly and time consuming, or you might be looking to implement a safety program for your employees or drivers. With intuitive AI you now have the capability to predict things much more accurately, such as the cost and timeframe of electrification or what the highest risk probability is of a vehicle and driver being involved in a collision. With Ace, you can ask the question and then, if you’re looking to coach drivers and improve fleet safety you can reach decisions much faster because you can ask a question and quickly get the response you need. Of course, it’s still up to you as a human to make sure that you fact check and go through that due diligence, but Ace is there to assist in helping you speed up that process.

AS: How is Geotab Ace designed to evolve so it can help fleets as their requirements and demands continue to change?

MB: One of the things that we put into Ace is a monitoring and testing framework, which we can convey to our users as well. What it means is that when you ask Ace a question, we’re able to monitor those questions and the feedback resulting from them to make sure we’re getting the right responses. As Geotab and Ace continue to evolve, we want to make sure that we’re keeping track of the kinds of things that fleets are interested in and where there’s a huge interest that’s spiking productivity. In those kinds of situations, it’s imperative to make sure that the productivity questions are getting answered in the way that we expect them to be. This gives us the opportunity to seed Ace with thousands of questions that we feel our fleet customers are likely to ask and, as those questions change over time, we’re able to monitor them and adapt our infrastructure to support them. It’s also important to not only be able to ask data-oriented questions but also have the ability to infuse them with more domain level knowledge. So, for example, if you think about what it means to be a safety professional in the fleet sector, it’s not just about the number crunching. There’s a lot of other things, such as policies that you need to understand and how they are implemented. That’s why it’s important for us to be able to take that unstructured data and make sure that Geotab Ace is not only a great analytics champion, but that it can compare with the best-in-class knowledge within specific domains. This is really how we’re starting to see Ace evolve, and how it will continue to get better and more intelligent over time.