Industrial machinery and AI – Episode 5 transcript
Chris Pennington: Hi, everyone. Welcome back to the Siemens Digital Industry Transformation podcast series. I’m your host. As always, I’m here with Raul Garg, Vice President for Industrial Machinery, Vertical Software Strategy at Siemens Digital Industry Software. Today we are joined by our expert Subba Rao, Director, Manufacturing Industries Cloud, Siemens Digital Industry Software. Subba, could you give us a brief introduction explaining your role at Siemens?
Chris Pennington: OK. Could you give our listeners a quick rundown of Mendix?
Subba Rao: Mendix like the way you should look at like Mendix is it’s a visual development tool. historically a lot of software development is done with high coding and Mendix is bringing abstraction and automation much simpler with visual development. Also, it’s AI enabled, so the capabilities that are within with all the AI advancements that we are seeing.
Developers get assistance while they’re building the applications or software, right? That’s the key element of AI enabled and also build smart applications. I’ll go further detail into that depending upon how we drive the discussion. But like you can see like 4 aspects of Mendix is 1 is visual development.
The other one is the AI enabled, getting all the core capabilities of AI into the lifecycle of software development. Then the deployment is very critical in terms of providing the flexible deployments, either it might be cloud deployment or in the industrial edge. And lastly like multi experience, providing the right level of experience that is required for different users when and where they need be it on the mobile device or on the HMI devices on the shop floor versus your desktop applications, right? Web applications. that is the Kind, of like 4 elements.
All this is to bring abstraction as I was telling you earlier and automation for higher productivity.
That means the speed of building your software solutions or applications, bringing the collaboration between your stakeholders, not just IT or software experts build the applications or the solutions, but more, the domain experts can also build and engage in this process. That’s a collaboration that is enabled in the Mendix and lastly the governance and DevOps which is like its own domain traditional. In the traditional development it’s brings more simplicity and abstraction. That’s at a high level what Mendix brings to the users.
Rahul Garg: : Actually Subha, you could probably say that you’re democratizing software develop with Mendix where you’re making non software developers into software developers as well.
Subba Rao: That’s a good way of putting it actually democratization of software development. Yes, traditionally in software development you need high coding skills and with Mendix if you’re a domain expert you know the process. You know how you want to use the information and also how to take the actions and integrate into the workflows. That’s whereas a non-expert of software development, software skills or programming skills the domain or subject matter expert can even build on applications that fills the process that they are very much familiar with. Yes, I totally agree that it is democratizing that software development for the domain experts too.
Rahul Garg: : That’s great. That’s great based on the adoption of Mendix which from what we know is very well one very well adopted in the industry. Do you all have some metrics on how the customers who have been able to use metrics what kind of speed and performance they’ve been able to get?
Subba Rao: From the speed perspective, we can we have seen that like compared to the traditional coding using Mendix like it taking off accelerator like 6 to 10 times, right. Obviously, it kind of varies from the current maturity of the IT landscape at the organization, but 6 to 10 times is what we have seen as a baseline in terms of the speed and also the collaboration, the inefficiencies that typically come in the traditional development process where there are requirements, and the requirements are fed by the subject matter experts to the IT teams to build it and then a cycle of verifying it.
All those are also kind of simplified so that there it’s a co-creation and collaboration that happens within the domain teams and IT teams to build this applicant. That’s the simplicity and that’s another level of improvement or productivity that we see from our customers.
Rahul Garg: I think that’s really good because I’m sure there is no company in the universe that doesn’t have need for more software, and you basically are making it a lot easier to build all the necessary software applications. So that’s fantastic.
Chris Pennington: So, Subba, in your opinion, how does AI and Mendix fit together?
Subba Rao: That’s a great question. like let’s as I was telling earlier. Mendix has four elements, one of them is AI enabled as part of the bringing abstraction automation to the users for the software development. So what does AI enabled mean? It’s like its AI assistance during the actual development of your application or solutions.
When you’re an application developer within your Mendix environment, you might be an expert with high court. Or you might be, as we talked earlier, a domain expert but not highly skilled in programming. you need some assistance while building an application. That’s what we call it, like AI assistance. It gives all the recommendations and even steps into what the best next step logic that they can use while building the workflow or any process information or even when they pick like the data sources giving the recommendation.
More, it’s assistant to the developer. Whether you are expert or not, you can get into a lot of indications what are the next steps or non-experts as well is almost hand holding for them to build applications. That is 1 aspect of AI and then the other aspect is it’s not only during the development phase, right? You are using these applications which are more like smart applications. And with AI advancements we are seeing a lot, especially in the manufacturing AI where they have been using machine learning for several years. There are machine learning models, and the outputs are sometimes there is customized solution that was built and those models. You can bring into your application and with these additional advancements that are happening with Gen AI, which we can discuss further is how you bring that smartness into your application.
That’s the next level of AI augmentation that Mendix provides for the users. To make the applications smarter with all the intelligence coming from this AI results, either data science results or it might be the models that the different scenarios have been built to drive the human in the loop processes.
Rahul Garg: Actually, if you think about it, if you were to take a step back, Mendix does a level of abstraction where you don’t need a software developer to build the necessary applications that you would want. And then we are further super powering it by bringing an AI into the process as well, so it becomes even easier for a person to use Mendix. A Mendix developer to kind of makes it even easier for them to take advantage of Mendix has to offer.
Subba Rao: Yeah, that’s a nice way to put it to yes, Mendix with the visual development, it is already giving that speed that we talked 6 to 10 times I mentioned earlier, right. And now just augmenting that further with this AI tooling within the same processes.It’s even accelerating further in terms of the whole application building and putting that into the production and then into the hands of the end users who actually are making that impact tactical.
Rahul Garg: Yeah, that’s fantastic, right. And especially with all of the growth that’s happening in AI and as companies are looking to examine closely and put out strategies on how to drive adoption of AI within their own organizations for enhancing productivity inside their companies This can be a great combination of AI and Mendix to really drive and take advantage of what AI has to offer.
Chris Pennington: Agentic AI and Gen AI are both still in their budding stages. For our listeners, could you both explain what Gen AI and Agentic AI is to you, so we get a good idea of like its capabilities for machine builders?
Subba Rao: Let me start with the higher-level definition of Gen AI. Gen AI is definitely what OpenAI started with ChatGPT in 2022. It’s picked up really fast. Everybody is now talking about that and that’s a tipping point to get into the potential of that. For gen AI, all the data that is available generating with content, right? Either it might be like the data might be a text data or image data or the video data, right? Means like generating the content based on certain prompts .That’s at a high level. It’s more like taking to the next step of actual generating a new content. That’s in a simple term.
Agent is a tool that continuously monitors, learns, and takes actions. A chat bot is a gen AI application that responds to a specific prompt whereas agents, especially agentic AI, it takes actions and drives towards an outcome.
that’s at a high level. You can see it is difference and obviously as Gen AI is used further into your agent because it is generating certain actions, like certain information that drives actions automatically or some in some cases humans in the loop.
Rahul Garg: Yeah, I think that’s a good explanation there, Subba. What I would further add into that is when you think about what’s thought about in the context of generative AI today, it’s primarily having the ability to create more text or images and perhaps. There in some code and audio video and relying upon the large volumes of data that’s available on the Internet.
Whereas Siemens is trying to drive this is bringing in AI and intersecting that with industry. That’s where we have this whole notion of industrial AI where we are trying to intersect between what the general-purpose AI capabilities are available and how they can be leveraged even further from an industrial perspective and trying to bring those two words together. Because while AI can be useful for writing better emails or messages or creating better images, we think there is significant value in the industrial world as well and that’s what we are trying to do is bring the value of the two combined together for our customers.
Subba Rao: Yep, I totally agree. And this is where the biggest difference comes in terms of like the general-purpose large language models that we see out there versus what is needed for the industrial perspective for industry if you slice that into two aspects of engineering and manufacturing. On engineering side, the domain know how is in the engineering data which might be in the 2D drawings or 3D information, interpreting that and being able to generate new aspects based on certain previous engineering intents. That’s one dimension.
And then on the manufacturing side is more the data is on the OT, lot of real time data and how do you bring that real time data which is very specific to customer processes and their infrastructure in the shop floor into this gen AI or even agent. That is where Siemens thinks it’s critical for the organizations to understand those differences between these general-purpose large language models versus what is really needed in the industry perspective.
