00:00 Overview
Welcome to another episode of Driving Innovation, the podcast dedicated to exploring cutting edge solutions and technologies driving the automotive industry into a new era of software defined innovation.
Today, we’re diving into how artificial intelligence supported by the right SDV infrastructure is empowering OEMs to deliver seamless and enjoyable vehicle ownership experiences.
Joining me for this conversation is Steve Stoddard, product manager for Sonatus’s AI technician builder.
Steve will help us understand how this innovative solution leverages powerful AI and real time vehicle data to simplify vehicle ownership and maintenance.
And we’ll also explore the foundational SDV technologies that make these advanced applications possible.
Let’s get started.
01:00 Complexity of advanced vehicle tech
Welcome to Driving Innovation, Steve. Thanks Sanjay. I’m glad to be here. Glad to have you on. Vehicles today are packed with a lot of technology.
And they’re also getting very complex. What are some of the challenges that this complexity imposes on people?
Yeah. That’s a great question. I think for drivers, one of the problems you often see is a misunderstanding of vehicle, telltales, the symbols that show up on the dash leads a lot of OEMs to be building these digital owners manuals to help understand those things, but it kinda misses out on what do those symbols really mean and what should you actually do about them. I think for service technicians, a lot of the digitization leads to you have to replace entire parts and more confusion around the proliferation of DTCs and signal codes and what do those things mean. And for OEMs, a lot of times they’re running into issues where across the entire fleet, they can’t gain insights into what’s actually happening and what sort of, issues are emerging and so that they can, avoid recalls.
Well, we’re here to talk about the Sonata’s AI technician builder. So I’m assuming you’re gonna tell me that a lot of these issues can be resolved by something like an AI technician. So help us understand what the AI technician is and also help us understand what the builder part of this solution entails.
Yeah. Absolutely. So one of the things that we found out is that, many of these different, people, different users of of these sorts of tools don’t have good insights into real time information about what’s happening with their vehicle. You know, there’s a lot of advances in AI out there today.
And I think if you have a problem with your vehicle, you can go ask Google or query ChatGPT, but the problem is you’ll get answers that are based off of sort of the Internet, and what they think is going on with your vehicle. What’s really missing is that vehicle connection part, the data connection, and, the access to that information underlying there in order to answer that. And so one of the things we find is the underlying technology that’s required to enable these AI technicians really relies upon software defined vehicles, the ability to collect data, things like, Sonatus’s product, collector AI product, as well as the ability to take action in the vehicle really helps to advance and make this capability available for for a vehicle.
03:09 Introducing Sonatus AI Technician
Before we talk about some of the, underlying technologies, you mentioned SDV. Obviously, AI is involved in it. Let’s focus on what the AI technician can do for the different types of users that may potentially interact with it. I’m talking about the owners, the service technicians, and even the OEM engineers. Yeah, absolutely.
So those are three very good use cases, intended users for this type of product.
For users, for drivers of vehicles, the owners of the vehicles, one of the things that an AI technician can help them do is to understand things that may be going on in their vehicle that are not as readily available or easily understandable. So you can imagine the scenario where you have an issue going on with your engine, but a check engine light doesn’t come on. AI technicians can help gain insight into what’s actually happening there and provide recommendations for an owner whether they should bring it in for service or not. And when they do that, they can also provide data to the service technician ahead of time so that they can more easily diagnose what’s happening with the vehicle and have the right part on hand when you bring your vehicle in, as opposed to having to come in three weeks later, once the part finally arrives and, you know, being out the use of your vehicle for for a long period of time.
And for OEMs, it enables the ability to gather that data from that, from the fleet level all the way at the service locations in order to see these issues that are emerging, and maybe better understand what’s happening with their fleet so they can make improvements for the next vehicle generation or avoid the recalls, that might happen otherwise if the if the issue proliferates.
I can see how this would incentivize me as an owner to visit the OEM branded, service shops, given the fact that they’re gonna have the information that the AI technician collects. What are some of the overall benefits for not just the user, but also the OEMs that a solution like this would provide.
Yeah. Absolutely. So you you hit the nail on the head, definitely. Having this kind of insight makes the service experience at a dealer owned service shop much more streamlined, much more efficient.
So their service operations can be more efficient. But also, I think for the OEM, it really leads to a feeling of satisfaction, customer satisfaction with that process and the whole ownership experience. So it leads more to customer loyalty and the knowledge that, hey, I’m gonna stick with these guys into the future because I have such a good experience when I own this vehicle. My next one, I’m gonna buy from the same company.
Right. Right.
05:31 Creating AI Technicians with Builder
Now we talked about the AI technician itself. And just to be clear, Sonatus isn’t actually building the AI technicians. That’s where the builder part comes in. So help us understand a little bit more about that part of it.
Yeah. Absolutely. So one of the things I mentioned earlier is that, a lot of OEMs are exploring these digital owner’s manual type of applications or the in vehicle virtual assistants. One of the things that we find there is that you have to customize these things for each vehicle because there’s a lot of specific domain knowledge for that vehicle, whether you go from one model to the next, one trim level, etcetera.
So in order to make that scalable, you really have to have a platform where an OEM a platform where an OEM can repeat this process over and over. And, ideally, they don’t have to be an expert in AI or even the Vehicle itself in order to put together the technician that that has the capability to know this information. So that’s why at Sonatus, we built a builder platform that enables various users at the OEM to actually create these different technicians depending on how they wanna actually use it for their customers or internally within their own, engineers or even at the, customer service level.
So they can basically create these technicians that are very specific to their vehicle models, their trim levels, potentially even down to the owner level?
Absolutely. So one thing, that they’ll provide is the documentation about the vehicle itself. So these will be things like the owner’s manual or for something that’s maybe more internal facing. For service technicians, it’s gonna be like a repair manual, wiring diagrams, etcetera.
For an OEM engineer as a user, typically, you might have an FMEA document or other fault trees and things like that. So that constitutes sort of the engineering knowledge about the vehicle. Then you wanna pair up the actual real time data coming from the vehicle, and that’s where you start to connect in things like the actual data from the vehicle that’s gonna be oriented from the vehicle VIN and the specific owner, And then also things like service records, your manufacturing history. All of this is possible.
It’s really dependent upon the different data sources that an OEM wants to actually connect up. And that’s gonna depend on what are the use cases and the specific users they’re going to build that particular technician for.
So it’s almost like having your own personal AI technician. I love it.
07:36 Enabling SDV Technologies and AI
So we’re we’re starting to get into how this actually works. Obviously, there is an AI piece of it, but you can’t just plop AI onto a vehicle. Right? There obviously needs to be some precursors or prerequisites, that enable these types of applications to actually be deployed. Help us understand what some of those foundational technologies are.
Yeah. Absolutely. So one of the items that’s critical, we think, is the ability to get the data from the vehicle itself. So, products that, sit on the vehicle in the vehicle network that communicate across the different ECUs, having access to all of the vehicle information, not just specific DTCs or specific areas of the network.
So that’s one as a foundational technology around the software in the vehicle for gathering that data, in real time and and being able to access that across the different nodes. A second aspect is actually getting that data back to the cloud, which is more of a data collection product, something like Sonatus collector, AI, which which brings the data through very lightweight policies to bring those back to the cloud. Once you have those foundational components, then you can actually start to make more, holistic capabilities and bring in things like specific AI models that could run on that vehicle data.
So you know, something that’s very specific or domain expert knowledge required, so that it’s not only relying upon the AI technician itself, but it can call upon these different agents that can access those different tools and, be able to infer information about the vehicle.
09:06 Implementing an agentic model
There’s a lot of discussion nowadays about an agentic model where you’re not just analyzing the data and and inferring things out of it, but you’re also sort of closing the loop. Anything in, AI technician builder or the overall infrastructure that, lends itself to an agentic model?
Absolutely. So all those different data sources that we just talked about, typically, you want to build those into discrete agents. So they’re sort of AI specific modules that are really, really good at one specific thing. Going and looking at this owner’s manual data, for example, or going and creating these databases to understand the signals coming from the vehicle.
So not only does it have to be able to access the data, but it has to be able to interpret that and know, hey. These signals, especially at this time, mean this thing. And relying upon that data, combining it with the knowledge about the design documentation, it can know then, oh, when I see these types of things, it’s this sort of a problem, and that’s what I need to recommend to the user. So those different agents generally connect up to these different data sources, and through that, you can leverage these different tools.
One of which, for example, is to connect to a Sonatus collector AI product where you can deploy new data collection policies based on the, interpretation of the, AI technician. So it could actually look at the data coming from the vehicle based on the user query and actually initiate a new data policy, deploy that to Vehicle, and start collecting data that was otherwise not even available.
It seems like the AI technician is almost thinking in real time in terms of what its next steps are, what additional information I need to make the right, diagnostic, response.
10:42 AI Models involved
Let’s talk a little bit about some of the the AI specific technology.
Are these your traditional AI ML models? What kind of models are you using for this?
Yeah, absolutely. So typically, we’re using LLMs just like, any many of the other applications you start to see today. Ultimately, it’s based on the foundational LLMs, but we prefer open source, where we then fine tune and and bring make those more, in house models. And part of the reason for that is because they can then perform better on the specific automotive domain. A second reason is that a lot of OEMs, they wanna host their own models and know that their data is protected, both from the end user perspective for PII data, but also for the proprietary data that an OEM might be providing as knowledge sources for the AI technician. So we, build on top of those custom LLMs and then bring in techniques, around retrieval augmented generation and then the agentic framework, which overall, we bring in the know how of the system prompting to make all those things work together seamlessly.
Got it.
11:43 Summary and Wrap up
So just to summarize, you’ve got the the AI portion of the solution, but you also have the underlying infrastructure that goes in it. And you talked about having a more, accessible network, having, you know, common data exchanges for the models to to feed the data into the models. I would imagine this is where things like service based architectures or service oriented architectures also come in. So that sort of is the sort of the SDV part of the the overall equation.
Yeah. Definitely the in vehicle EE architecture plays an important role And having access to all of those different data sources, it allows you to bring in things like, you know, just the general telemetry, but also DTCs, log files, all those sorts of things, even the network traffic data that can be useful, particularly if you think about, like, cybersecurity and some of those types of applications. So it’s really important to have you know, the more modern the architecture is, the the sort of more capable that these AI technicians can be, definitely.
Excellent. And, this is all described in the latest solution brief that we’ve just published, the AI technician builder solution brief. So thank you, Steve. This has been very informative and we look forward to having you in another driving innovation episode.
Awesome. Thanks, Sanjay. It was a lot of fun. Sonatus’s AI technician builder is a great example of how SDV technologies and vehicle AI can help OEMs transform vehicle ownership experiences, taking the hassle and stress out of vehicle maintenance and repair while building brand loyalty and earning lifetime customer value.
Download the AI Technician Builder Solution Brief at sonatus dot com for more details and schedule a demo to see it in action.
Thank you for joining us for this episode of Driving Innovation, and we look forward to seeing you in another episode.
