Amit Shastri, CTO Americas, Digitate, on AI Agents to Handle Outages, More

Amit Shastri, CTO Americas at Digitate, explains how composite AI moves operations from reactive firefighting to predictive and autonomous action, without sidelining human judgment or ripping out trusted systems. Digitate’s approach blends logical reasoning, LLMs, and guardrails to deliver unified observability across IT, OT, and business processes.

• Regional CTO role bridging customers and product
• Autonomous and ticketless operations as the North Star
• Predictive alerts that prevent downtime
• Horizontal observability across procure to pay
• Integrations with ITSM, monitoring, and CMDB
• Composite AI with logic, LLMs, and human approvals
• Action firewalls and role-based controls for safety
• Job shifts from doers to exception handlers
• Build vs buy tradeoffs and enterprise scale
• Natural language interfaces over complex tools
• Leveraging legacy systems for rich operational data
• AI’s global landscape and India’s momentum

Transcript

Speaker 0:19

Hello, and welcome to FoDES, the Future of Design and Engineering Software Podcast. My name is Roopinder Tara. On the show, we will have guests that will discuss tools and technology that engineers will find interesting and useful.

Roopinder 0:33

Amit Shastri, thank you for joining me. Welcome to the show. It's great to have you. You are the CTO of America of Americas at Digitate. Would you like to tell us a little bit about yourself and then we'll get on to what uh what Digitate can do for engineers specifically and manufacturing companies?

Speaker 2 0:54

Absolutely. Hey, thanks, Roopinder. First of all, thanks for having me on the show. Um, it's good to be here. I'm based out of uh good old Cincinnati, Ohio. And my role in Digitate is of the uh regional CTO, which is for the North America Geography. Um, and as the you know the name would indicate, it's uh it's more of a technological bond between the market, our customers, and what they are looking for, what they are doing, and the our product management and product engineering teams, right? So, of course, um increasing the adoption of the product in the market, you know, kind of tailoring the solutions which are suiting to the market needs and customer needs, and of course, taking the feedback back to the product teams for them to have it in the product roadmap and product strategy. So that's predominantly what I do. I mean, the reason I'm here today is of course this one of the marketing, and that's where I work with our marketing teams and the analyst relations and public relations, that's all it's also part of it. Yep.

Roopinder 1:59

Right, right. So why have to ask I have to ask you, why Ohio? So your company is based in Santa Clara, right? And all all all the tech is from there. So, what are you doing over there in Ohio?

Why Ohio And How Digitate Operates

Speaker 2 2:12

Good question. I guess uh you have to be where your customers are, and I think Midwest and the East Coast, uh, there is a good customer base here. I I do come to Santa Clara once in a once in a quarter or so. Uh I mean, if we are really a like federated or a distributed uh setup headquartered in Santa Clara, there is a good amount of product engineering and management that happens back in India, and of course, with the the customers are spread all across, right? The North America, United States, Canada, Latin America, and in fact, all across the globe. So I have like three counterparts of mine who are also doing this for Europe and for Asia Pacific. So they are that's why the regional rule, yep. I see.

Roopinder 2:55

All right, you're not CTO. So you're you're the one in charge of making all this work where we're where everybody's talking about AI, right? Your comp your company is actually doing something, not just talking about it. You're actually so here, all about agents, AI agents, and uh how useful they'll be. So digit digitate is actually in that business of making AI agents, correct?

Beyond Hype: Agents Doing Real Work

Amit 3:20

Absolutely. And in fact, uh we have been here for about this is our 11th year in this business now. It when Digitate was founded in year uh 2015, and it's kind of a uh, you know, the mothership, which is uh TCS, Tata Consultancy Services. And um, I mean, you would know about the Tata brand back back in India, and this is a it's a direct brainchild of uh TIDDC, which is uh Tata Research Development and Design Center back in Pune, right? So it all started back in 2015, and we we have also grown a lot. We have also adapted ourselves with the new trends and the the market demands, right? Predominantly it started as a with a with an automation-heavy focus, and then we we also uh got into the the intelligent insights and providing the um the keeping IT operations as a as a core. That's that's our DNA, and not always being just reactive in the operations, but also how how how do we become proactive and even predictive, right? So that's the that's the main thing.

Speaker 4:24

Yep.

Reactive To Predictive To Autonomous

Roopinder 4:26

You mentioned Pune. We had uh we had a small operation in Pune, and I know a lot of uh computer-aided design companies, CAD companies, to have a lot of up have uh uh people there uh working there. Uh is that where you're from, Pune area? Actually, I'm I'm from Mumbai, it's like three hours from Pune, yeah, yeah. Close enough. So that's where I usually this where I usually flew into Mumbai and then and took that highway. That's a very beautiful go through the mountains, yeah. So okay, so very good. So okay, so we've been hearing now for probably a couple of years how agents are going to do all our drudge work, right? And uh maybe you know the routine work and help us with our jobs, not take them away. They're very careful, everyone's very careful to say not take your jobs away, but help you with your job. So that's good. Uh and you have one, I think you have a product called Igneous your company sells. So tell me how Igneo works. So my impression from reading about it was just that it helps you with the areas of outages. Like if your company has an outage, Igneo can step in and somehow take care of it. Now, what kind of outages and how does that work?

Amit 5:47

Well, I mean, that's that's one of the scenarios, but talking about the product, I mean, it is the core of the product is about enabling enterprises towards uh autonomous operations. And when I say autonomous, I mean autonomous operations and the ticketless, right? This is the of course the Nirvana state or the North Star. And um, when we I think the example that you gave about outages, I mean, the outage happens, and then you know, a bunch of people get together and they start, you know, kind of brainstorming what happened, and there are signals and metrics and data that is available, they analyze that and they react to that, right? So this is a reactive scenario, and that's how it happens in in traditional key operations. So where Igneo comes in is not not just limited to the reaction. I mean, if the if an outage happens or if a problem happens, by the context and the intelligence that Igneo has, IGNIO is going to take an action at a machine speed and resolve that outage or resolve the problem. But even one step further, I mean, it it has this ability to predict about the potential outage or the potential problem by predicting it. I mean, think of a SRE persona or operations or whose main job is to keep the lights on, right? If we start giving him like two hours or four hours notice that, hey, this particular server, I'm watching it and it's been the CPU is trending upwards continuously, you know, and I suspect that this could go down, this could crash in the next couple of hours. I think that is a huge, huge input or huge help for any human or the operations person, right? So that's where the predictive scenario comes in. And I think the one the talking about autonomous is where Ignio, this intelligent machine, with human in loop, of course, it actually goes and takes necessary action. It decides to restart that machine, it decides to add CPUs, it decides to secure additional instance, right? So right from the reactive to proactive to autonomous, I think that's the journey. And that's what IgNIO enables you with.

From IT To OT And Business Signals

Roopinder 7:50

So once it senses a shortage in your server, it doesn't wait for, it just doesn't inform you, it doesn't wait for a human, a slow-acting human to come along and decide what to do. It actually takes all the action. Is that correct? That's right. Right. Now, is it specific? Outages, we're using the term outage very specifically to servers, but outage can happen anywhere. A factory could stop, a line could stop operating, you know, robotaxi could stop in the middle of its operation. That's there's a lot of different outages, right? Is your so is igneo specific to servers and uh computers, or can it be applied across the board to other applications?

Amit 8:36

So it is predominantly for the for the IT operations, right? But when we talk about the areas of manufacturing, or we are talking about energy resources and utilities, right? We are seeing that there is a there is just so much of uh operational technology or this uh IoT devices, like you have cameras and sensors and meters, right? I mean, that that's where it it just not would not just be limited to the uh IT devices as such. I mean, in IT also, I mean server is one example, but it could be network devices, it could be storage devices, right? And everything this IT and OT we are talking about, I think we are still still in that operational at that operational layer, right? But what we uh I think the main fundamental principle with which ignio works or anything, any feature or capability that we bring into product, it's with to address these three problems. Number one is uh IT IT exists because of business. And then anything IT should find an issue before business does. That's number one. Number two is IT should resolve that issue uh before business starts getting a pinch of pinch out of it. And third is it should be proactive to resolve the issue so that business never never gets that or never gets impacted. So with these are three fundamental things. I mean, uh IGNIO is not just limited to information technology or operational technology, but also the business level monitoring, right? It can do the business, we call it as a unified observability, which is not just limited to the infrastructure and storage and servers, but it's also about the business observability, right? For a manufacturing plant, what has been the throughput of that plant over the last six months and what is the trend that I'm observing there? What are the energy levels of that plant? And what so more data you feed to to AI, it it will, which is which in this case it's Igneo, it it will it is capable of making this, you know, establishing patterns, doing the forecast, and you know, giving you notices in advance.

Supply Chain And Horizontal Observability

Roopinder 10:40

Okay. So I gather from this that it was you know trained initially from on IT operations, and it was uh, you know, it understands the processes involved with uh server outages and how to channel resources or switch over to other servers if anything happens. But then it's also then applied the same principles can be applied to any business outage, right? Any outage, like for example, I'll give you an example. What's heavy in my mind is operations based on supply chains. So would you consider that an outage? Like if all of a sudden a supply chain operate breaks your operation, right? Could could it be used, could it be trained on that sort of thing, like to have another supplier jump in where one has failed, just like you would have a server jump in where another server has failed?

Amit 11:34

Absolutely, absolutely. And in fact, that's where we we have this. I mean, I spoke about this unified observability, right? I mean, the ability to observe, we believe it will not just be limited to your application stack or the technology stack, but there is something we call as a horizontal observability, right? I mean, giving an example of a supply chain or a let's say procure-to-pay process, right? Where the need is identified, right? You have to generate the um, you know, you know you have to generate a purchase order that gets approved, then the goods are received, then the invoice gets generated and the payment gets made. So if you think of this procure to pay as a let's say six-step process for a particular company, right? So by business observability, we say that at each hop of this process, we would want to monitor this. And at a particular every hop of this process, of course, there are going to be underlying services, applications, and the related IT components, right? But we say that out of these six steps, if we know that it's, hey, it's a step number five, which is invoice generation, that is causing the problem, and that's affecting my entire procurement to pay process, then I would like to kind of focus on that hop. And then I also get an ability to kind of drill drill down into it and say that, hey, is it because of underlying application? Is it because of the server that is crashing? What is the root cause? And that's really the business and information technology. That's the it's a convergence of both. That's what that's what Ignio does.

Context Via Integrations And Adaptation

Roopinder 13:04

Interesting. So tell me, I know this is in your area sales, but uh how would it work? Would it be sold on the on the uh on the basis of like, hey, okay, we understand your, we're gonna try to understand your business, and then we'll tailor Igneo to handle your outages and your pro. We'll learn your process and then we'll train Igneo to use to uh to work for you. And then you kind of leave the Igneo in place. Is that the business model?

Amit 13:35

Yeah, I mean, at the at the end of the day, it it is a product with a certain certain features and capabilities and and functionalities and area which in which it is very good at some of the areas where it it cannot do much, right? But uh again, I think the the good part here is that if you look at any enterprise, right, there are going to be these uh systems which have grown over the years. Like there will be uh IT service management, there will be monitoring solutions, there will be uh a CMDB, which is like entire inventory of assets that you have. So Igneo comes with this inherent ability to kind of interact with these tools, or we call it as integrations with these products. And that's what that's how it understands the context, that's how it understands about your particular company and the organization. A procure-to-pay process, for example, the six steps could be true for company XYZ. If I go to another manufacturer, they said they say that hey, I have a couple of additional approvals in this process, and instead of six, it is a 10-step process for me, right? So Igneo, this particular context, that's very important for any AI technology, and Igneo, IGNIO is no exception. But how it builds this context is by integrating with what you already have, the knowledge sources or data sources in your organization. And that's how it starts getting its intelligence.

Roopinder 15:00

So from where I sit and I see, you know, I see a lot of news, I see a lot of press releases, I go to conferences. Uh of course, the subject is always AI. It's always like leads with AI, ends with AI. And people that go to a conference, for example, are then left with, okay, AI, let's make it work, let's find it. So I imagine here's you, your company like raising their hand and saying, yeah, we got that, right? We got we have that solution. Yeah, because uh there's a huge gap between understanding a technology is is uh necessary or useful to actually have an app that creates it. I don't think a normal person knows how to make an AI agent.

Amit 15:43

I mean, now nowadays with you know, but it's it's get getting better and better and it's getting more and more democratized with uh you know, with uh with every I mean you would have watched the Super Bowl a couple of weeks or last week, and then yeah, every other advertisement during the Super Bowl was on AI and the big force, right? The cloud was there, the Google was there.

Roopinder 16:03

It was, yes. I was surprised to see so much AI. I I don't I'm trying to remember another time where uh maybe when the internet was getting hot or overheated, uh that then uh that that maybe there was some internet company saying they were okay, look at us, and it was in the super poll, right? But that was weird. That was weird to me.

Amit 16:24

That's right. And I mean, uh I mean that helps in kind of democratizing this, but that gives a that brings another challenge, right? Um other day we I was talking with my wife, and then um we have a teen teenager, teenage daughter and a son who is like three years younger. And I mean, we were just debating about when to give them the cell phone, and you know, I mean, of course, my teenage daughter already has it for a few years now. Yeah, but about this, there is so much of peer pressure involved, right? I mean, they are going to come back and make an argument. Hey, my this, my XYZ friend has it for last couple of years and he takes it everywhere he goes. Why can't I do the same thing? So, kind of same thing for about AI, also, right? With all this democratization and more people getting to know about is one thing, but we feel that there is also a sort of peer pressure that because the technology is available, what can I do with it? Right. I mean, the actual question should be that hey, this is what I need to do, and how AI is going to help me doing that. I mean, you don't want to do something because technology is available, so it's it's like uh that horse and carriage kind of thing.

Composite AI: Logic, LLMs, And Humans

Roopinder 17:28

But uh it is, it is. You you mentioned that the children using you know using cell phones, but uh I would have to say that the uh it was actually the youngsters, a youth that first really picked up on AI and they use it, they used it for bad purposes, of course. They used uh ChatGPT to cheat on their homework assignments and uh you know then later to uh write essays for their entrance uh applications at universities. But when we first heard about AI, that yeah, that was that was what we heard. I mean, the latest version of AI. I'm talking about LLMs, of course, and that just like burst into popular uh view. Uh and then the industry then started noticing, hey, we could do something with this, right? AI could actually be useful in the industry. Kind of kind of what happened in gaming, actually. You know, the uh gaming made computers graphics very brought was the the kids were the first people using uh gaming technology, and then that technology is now used for AI, but it was gamers that brought that chips into into play, the GPUs. So the kids are we have to thank the children, like they were they sort of introduced us to this tech, and then we come on the scene later.

Amit 18:53

So all right, so that's good. So it's AI Roopinder. Just one analogy on that. I mean, I always tell my kids, I mean, when you mention about this chat, the use of chat GPT in a negative way, right? Uh I mean, we always say that we want AI to help us with our with our mopping the floors and laundries and dishes and those kind of low-level activities. When it comes to creativity, right? Uh writing an essay for a college or generating music, I mean that's that's where I would like to use the human intelligence. I mean, that's I think that's always going to be a that's always going to be a topic of discussion, I think. But I think that's where the the right use of AI, I think the more of a creativity, I think that would still be, it could be assisting humans in doing that, but ultimately it should be, it should stay with humans. Yeah.

Guardrails, Trust, And Action Firewalls

Roopinder 19:45

It could be, I mean, it could be uh allow us to have more creativity if we can get used to the used to it. I think the initial fear is that it's going to be used for bad purposes and used for cheating. Uh, we had somebody from the Florida uh university that was big into AI. It's like promoting itself as an AI, and yet the same university had problems with its students using AI. And I thought that was ironic. You know, like you don't want your students to use AI, and yet you're thinking AI should be used somehow into the curriculum. I just thought that was strange, but you're right, it's it's uh first used for it could be used for bad purposes, but it could also be useful. But we're talking about LLM specifically here, right? Like what's right, right? That's the industrial need for AI, you know, it can use LLMs, of course, but it needs physical AI, also, right? It needs what uh Jensen Hong promotes as the next level of AI is physical AI, which is can be used in robots. Which if you could give AI the sense of not just language, but also sense of physical objects, physics forces and and uh mass, and you know, and uh you know, it's just stupid right now that computers don't understand that one object can't go through another object. Like a robotic hand has to stop when it senses pressure, right? It's we do that, but AI can't do that yet. LLMs can't do that, but physical AI can do that, right? We see that also as a next step in AI, the physical AI.

Jobs Shift: From Doers To Exception Handlers

Amit 21:31

Absolutely. I think in in some of the industries, like particularly like manufacturing, right? I think where the automation has been big, and we are already hearing about the extent of automation where companies like there are um you know some German automobile makers who have like these dark factories, or even China is one of the prominent ones into that. And they have achieved so much of such such a level of precision in automation that there is not much light required for this uh physical AI. And then this the I think the humanoids and the uh quadrupets and all these things where the or even even the computer vision, where if it's a process of a of a welding, right? And then if a computer, I mean, compared to a human naked eye or human eye, computer would be in a much more position, uh better position to to observe that, to monitor that, right? Physical AI, of course, that's I mean, predominantly for these industries is going to be more and more prominent. Going back to LLMs, I think uh one thing that separates uh our product is the um, you know, I mean I'm talking about digitate and ignio, of course, is that the use of composite AI, right? So we do not just rely on on LLMs, right? So IGNIO, I mean we we we put this into uh into kind of three three layers of intelligence. One is the logical reasoning, the logical intelligence, which is like the context that I spoke about and the IGNIO's own machine learning algorithms and the intelligence it builds. And it can it could suggest very deterministic uh reasoning and actions, right? Wherever IGNIO feels, so that's the first layer, wherever IGNIO feels that hey, a certain problem or an outage that has happened is little beyond my understanding, right? Rather than just pointing it to human, it it is going to get into that analogical AI where the LLMs come into picture, where the determinism is a little lesser, but there is a huge amount of creativity. So LLM is always going to come back to you with certain answer. How precise or how applicable that answer would be to your scenario, that's still a question. But there is an increased amount of efficiency there. So that's the second layer. And third layer, I mean logical, analogical, and third is humans. Like we cannot can never keep humans out of the loop. So there are certain decisions where humans have to approve it, you know, or there are uh certain things which even action suggested by LLM, if Igneo says that, hey, really go and restart my database, right? And if it is safe to do, or if it is safe to do from a business point of view, if human approves it, then only AI is going to perform. So I think that's that's one thing which is about this composite AI. That's what we preach and practice. And um see right now, uh like we were discussing, I mean, every this is kind of a honeymoon period for for all the companies with this big LLMs and the you know, uh, and I mean everybody's thinking about what they can do, but what about the energy consumptions and the cost by these LLMs? I mean, we cannot solely just rely on the LLM and for each and every input or every query, if I start going to LLM, of course it's going to be very, very expensive proposition. So that's where how much of it can be can be answered or resolved within the products context. That's the first priority. And anything that is little beyond, then we go to go to LLM.

Roopinder 24:56

I see. Okay, you I want to clear this up for anyone who's listening because initially you had said that India Indio can sorry, did I get that wrong? What's it? Yeah, so it can it operates without a human in the loop, and but you're saying there's also it's important to have humans in the loop sometimes, right? That's right.

Build Vs Buy And Enterprise Scale

Amit 25:18

Yeah, so let me clarify that. So one concern when we talk about any AI technology and we we are speaking with our customers and we are pitching for this and we are even implementing it, right? First concern that comes to comes to their mind is about hey, are you talking about taking an action on my servers or my infrastructure, my applications, right? And it's it's really a journey where you experiment with AI, you test it, you build trust, the explainable AI, right? And I would put it in into a spectrum, right? Like there are, let's consider the three actions that if you know that if one of the servers it has to be backed up every night. So that's a very mundane, repetitive task, and that's a very definitive task, right? So this kind of action definitely Igneo can do, or it can be given to AI, right? There is a second action that okay, I need to uh bounce a certain machine or I need to restart up a certain server. But what is the business impact of that, right? If someone does it for a bank, if someone does it in the in the middle of the day, that that it's going to be hindered to the hindrance to the operations and it's not good for a business. So that's where we say that okay, do these actions only if human approves. So we are keeping human in the loop. So you're you're talking about restarting a server, but you should only do it when human approves it. So that's the second thing. And the third is like we humans are saying that hey, no matter what, AI has no business touching this application or touching this system because it is too critical for me. So these are the three levels that we see. And of course, like I said, with the explainable AI, that's how you build a trust. You put certain guardrails for the AI actions, and same thing is in our product. We have certain features like uh rule-based, uh role-based access control or even the uh action firewall, right? I mean, if no matter what, if uh if a delete command for a particular file, that should be blacklisted. It's a action firewall which is never going to let AI perform that action, right? That's how the explainable AI and the trustworthy AI is there. And that's how humans uh grow confidence with uh with AI. But you, I mean, yeah, to answer your question, I mean, we are not taking humans out of the loop. We are saying that let humans focus on a more critical strategic, um, you know, high value, high-value actions and high value decisions rather than spending time on repetitive mundane tasks.

Roopinder 27:46

Do you have to answer to we were talking about this briefly, but do you have to answer to a question like, are you uh does it ever come to you that, hey, are you is my job safe? Is AI gonna take my job? Because I hear that a lot, right? People are especially white-collar workers, professionals, right? Even engineers, they are now, even engineers are now starting to realize that, oh, AI could cost me my job, right? Do you your your company seems to be in the business of not in the business of doing that directly, but could be seen as that sort of that's having that sort of effect. Do you have an answer for that for the if you're asked that?

Ops As AI’s Sweet Spot For ROI

Amit 28:29

Yeah, yeah. I mean, that that that's an apprehension. That's uh, but but what what we have seen, uh what we experience is that, like I said, the roles of humans, I mean, of course, the humans will have to know, or the the any practitioner, engineer will have to understand AI, will have to know about AI, will have to know how it works, right? And use AI in their daily jobs. But again, coming back to what Igneo does and the and in this operations area, right? We have seen that really the the job profile of people change rather than just being the doers of the work, they they become more of a exceptional handlers. The scenario that I mentioned that okay, uh IGNIO did not know or does not know what exact action to be taken here, and it goes to LLM, right? And LLM gives it, suggests it something, but without human approval, that should not be should more should not be performed. So in this scenario, humans' role was to of that exception handler, right? So they become the handlers of the exceptions, they they start doing more more creative or a strategic work. And on top of everything, human job is to train AI. I mean, particular manufacturer, how how procure to pay process runs, the person who has spent like five years or seven years in that in that domain, he's the best judge of that. And if he does not like something that is AI is doing, or does not like certain thresholds or alerts or actions that AI is proposing, it's his job, it becomes his job to train AI to take the right decisions. So in short, I think you can you can never take uh humans out of the loop. Uh their job profile changes, but uh they're they're pretty much part of the equation.

Roopinder 30:14

It changes, right? I I I interviewed somebody from a um an agricultural company. It was actually, believe it or not, a person, a company in Iowa that was uh getting into drones and AI. And uh I questioned that whole concept, like, wait, you're out in the Midwest, again, like I did with you. Like, what are you doing out in the Midwest? And what could you possibly be doing? We consider ourselves uh center of the universe here in the Bay Area, right? So what in the fringes, what are you using AI for? And he told me, well, you know, farmers were the first users of technology. And I said, Oh, then it real then I realized, yes, those tractors and combines, right? They did a lot of use a lot of technology initially, but the end effect of them put a lot of farmers out of work. You have to admit, a lot of farmers didn't get them out of work. And so I'm looking, I'm thinking of this example which you have provided, is like, okay, yeah, I'm gonna train my AI agent, kind of like I would train an assistant, right? I would take an AI agent, I would make an assistant. An assistant is gonna be pretty stupid for lack of a better word, initially, right? But it stays on the job, it never gets tired, never leaves, and it keeps learning, it keeps getting smarter. Wouldn't you say if I take that to the logical conclusion, it's going to replace me? Like the tractors and combines did for the farmer, would you would you say that? Am I gonna program myself right out of a job?

Amit 31:48

I mean, this is see, with going with that analogy, right? I mean, there it was always the like with the invent of internet, you know, that major change was there when the mobile phones um came, when the cloud computing became prominent. I mean, with all these industrial changes, we have seen that it always ends up generating more opportunities, right? It may not be the exact same thing that you have been doing so far, but as long as you're willing to learn, willing to adapt, and willing to progress with technology, use the technology in the right way, it's always going to create more and more opportunities for you. And right now, honestly, we are just scratching the surface, right? We are who knows that how many more, how much more deeper this thing breadthwise and depthwise, this thing is going to get, and how many opportunities it's going to create.

Roopinder 32:36

Right. Now you mentioned children.

Amit 32:37

I have a daughter and son.

Natural Language Over Complex Tools

Roopinder 32:38

How old? 17 and 17 and 14. Okay. What do you think their world is gonna be like? I because I already hear about people like getting entering in the job market, and it's especially in tech, but in lots of other fields, actually, a lot of white-collar jobs. It's tough. Like we had somebody on this week already, and he's a veteran SOLIDWORKS user. And he said it took him three months to find a job. That used to be the golden ticket, by the way, in our industry. You use it as SOLIDWORKS, you were good at SOLIDWORKS, you could go into any design firm and use that, right? Now I don't he doesn't know what to do. So, how do you what do you tell your children? Like you probably 10 years ago, you probably said software, you know, software is the way to go, right?

Amit 33:27

Um, yeah, I mean, it's a we we have of I mean we have been telling them that they can do whatever they want and whatever is their liking. And then uh particularly, I think I I believe that uh, or if they were to take my guidance, I would say that stick with the fundamental branches of of science, right? Or it could be arts also. But for example, my daughter is into into biology and biochemistry, and okay, um, son wants to pursue the the mechanical side of it or mechanical slash electrical. Those are the inclined so I think these fundamental um branches of of science, right? They would always stay. I mean, the the domain knowledge would would always stay. That would not be in a direct uh direct threat, right? I mean, the AI would always be uh we'll have to to to to kind of um work with it, but the fundamental knowledge would would always continue. I mean, who knows? I mean, that's that's my answer right now, but uh like one year or three years down the line, the way things are progressing and the new things are coming to to the forefront, who knows?

Roopinder 34:31

Yeah, things are moving fast. Uh, what about um so you mentioned democratization of AI implementation? I hear about vibe coding, I hear about uh I've heard about uh what is that low L low code, low code, low code, which which makes me think low code came around like a few years ago and it was sold as like an ordinary engineer, even an engineer who hasn't programmed for 10, 20 years like me, could use low code, right? And now with vibe code, all I do have to do is talk and I generate code, right? Was that the demo c democratization you were talking about, like that? I can make my own AI applications now. And then if I can, what do I need digitate for?

Keep Trusted Systems, Add AI Interfaces

Amit 35:20

Um what you need uh, I mean, the for the products like uh igneo and and digitate, right? It's very the area of this this operations or the IT operations. I think that's what uh that that's what the expertise are. And then of course, with uh with the years and years of experience uh serving different customers with our mothership or the TCS background, all that context is is already there. So see, talking about wipe coding or you know writing applications, I mean, as the complexity and scale grows, right? If you if you ask any developer, right, uh creating the the MVP or the the shorter version of it, it's it gets uh pretty quickly. We get get through it very, very quick. But uh as the complexity grows, the scale grows, that's where the challenges start start popping up, right? Yeah, and debugging something which is uh which is AI has built from scratch, that's that's a nightmare, right? So there are certain kind of uh gaps or negatives of that too. Now, what what Ignio does or what what difference Ignio makes, right? This is all already done in in Ignio, right? Talking about the key operations or the bad jobs and the ERPs of the world or the S4 HANAs and ECCs, right? IGNIO has this inbuilt knowledge. I mean, we have about uh 10,000 plus uh inbuilt automation components that that we bring to table. So this is and so the context is there, the composite AI is there, the inbuilt automation is there, and the combination of these three, it's it's kind of a ready-made product for you that is that is ready to to to scale up to your enterprise operations and enterprise IT demands.

Roopinder 37:01

Got it, got it. So it seems like okay, even though I could make any any, I could make AI work for me uh as I'd be really an amateur at it. I wouldn't have I might make an initial attempt, it won't be industrial scale, it won't be it won't be usable on an enterprise level. It might help me as a point solution, right? Yeah, but it's yeah, but it's not going to be a robust application, like right.

Amit 37:29

Yeah, and and to be honest with you, I mean, we we do get in these discussions with our customers also, and this is where it's always the the build versus buy, right? And then um, I mean, like the peer pressure or what's happening outside in the industry and what they get to hear, more often than not, they would have subscribed to one of the large LLMs, and you know, they have this confidence that they can build everything now. But as we start demoing our product, or as we start uh doing the pilots and the POCs for the product, the value comes out really quick. Yeah, and the complexity and the scale, but the rate of change at which it's happening. I mean, you might write a program today, which is uh with the LLM, which is good for today's needs or what what you but the technology is is changing so fast, right? To to keep keep pace with it. I think it's it's good to have a centralized product which is which which brings these features in build. So that's what it's like.

India, China, And The Race In AI

Roopinder 38:25

Yeah, I get it. I get it. I it's like it's a parallel thing to what's happened, what happened in publishing, where after a while, for a certain time, every company thought they would publish their own content. And uh veteran publishers said, oh no, no, that's we're gonna do a better job of this. It happened in lots of industries, happened in news, right? All of a sudden people got their news from people putting uh stuff on Twitter or X, right? And professional news organizations who were much better at it were ignored for a while. Now it's kind of coming back, right? Let's let the professionals do their job. We tried it, but we were amateurs, right? We're not as good as it actually hurts us to get our news from individuals who are not good at producing news, right? So we better let the pros do it. And I think that's happening with with developers, like coding companies. And yeah.

Amit 39:20

And and with this, with this operations, I think it's a it's a it's such a sweet spot for to start with AI, right? I mean, operations, we know that it has abundance of data, which is one of the key things that any intelligence mechanism needs, right?

Speaker 3 39:34

Yeah.

Amit 39:35

I mean, we see we we know that it's a garbage in, garbage out. That's what we talk about, any um system as such, right? But if you have this data, which is there is abundance of data with monitoring tools and ITSMs and CMDBs that I mentioned, if this data can be given to AI, this centralized engine, and yet there is a good amount of dynamism that is there in this this operations, right? IT operations world. That's where it becomes kind of a sweet spot to start with IT operations. And also there is another trend we are seeing is that a lot of companies, a lot of organizations uh wonder that, hey, where am I going to get budget for to invest in this AI? So all such uh optimizations and the cost savings and efficiencies that are that are brought in the areas of operations, we see that they invest that into other areas to towards the research or towards to do to better in the with the help of like physical AI or different branches of AI, which are specific to their domain. So that's why the IT operations or the business operations, that is such a such a sweet spot for AI. And that's where we that's what we concentrate on.

Roopinder 40:43

You mentioned ERP solutions. Uh in the in my space, we talk a lot about POM, which is similar. Uh, and we talk about design, CAD and CAE, computer-aided engineering. These are all a lot of these programs are enterprise-level programs. They're difficult to use. They take a lot of training to learn them, to be proficient in them. What is your view on AI being a natural language interface to bypass all of the complexity and learning and application like that? I realize that's probably not exactly what NDO does or what uh digitate does, but uh I see a world where I can talk naturally to an application. And maybe the application is so far uh submerged, I don't even know I'm talking to it, but I can tell I can tell an operation to continue to okay, fix this outage, and it goes and finds whatever the AI app or the ERP app or whatever app it needs to make it all happen. Do you do you see a world like that?

Amit 41:53

Where I I I do, and I I I do believe that uh all this um like the you know the CADs or the the PLMs and uh ERPs of the world, right? I mean, now you you would have noticed a trend that there are companies who are releasing AI native operating systems or AI native laptops, databases made for AI. So I think going by that, even these enterprise softwares, which are very niche in their area, they are going to be AI native. They are they would say that, hey, now rather than knowing this complex logic or in the ERP or for SAP, you have this complex transaction codes that you need to know. Rather than doing that, okay, I'm going to give you an ability where you can talk to me in a in a high-level language, in a in a simple English, and I'm going to take those actions. So having AI natively kind of converge into the the core of their product, I think that's, I mean, none of these none of these vendors can can deny that or delay that, right? I see that that is happening in near future. Yeah.

Roopinder 42:58

Do you still see those applications maintaining their role although in a submerged level? Or do you do you feel like AI is just going to replace what they do? What I'm talking about specifically is like I still I don't like, I may think my CAD program is very difficult to operate, and I like it to be easier, so I like a natural language interface, but I still trust that it can handle geometry really, really well. And my simulation program, I like ANSIS. I know ANSIS can do simulate everything, but I hate using it because it's already, I gotta learn all the commands. I come back to it after a month, I gotta learn all its commands. I but I still trust that it can give me an accurate solution. An accurate, reliable solution in both cases, can and simulation. It doesn't do this crazy stuff that AI does, where it gives me a different answer for the same problem, right? It's reliable, it's correct, I trust it. I just don't like using it. So I don't want a world where it figures out a uh their simulation solution or an engineering solution, right? I just want it to be easier to use. So, but in my world, it's like keep those apps, just make them easier to use.

Amit 44:07

I'm with you. And I think that's the earlier point I was saying about the fundamental branches of science, you know, or the the core, core knowledge or the expertise. I think they're going to stay, but it's about our ability to converse with them, about our ability to get things done, right? Our ability to learn things quickly and even kind of monitor them. Like one area, uh one complaint we hear a lot about manufacturing is about the legacy systems that I have this system which is running for last 25 years, and then now how do I make it AI enabled? Or how how can I ask AI to monitor it and give me insights, right? But if we if I really tap the right at the right places, I mean, these legacy systems are so rich with the operational data and the resources it provides, right? If we figure out a way to to observe them, to monitor them, go against the with their logs and get what we want, go with the native database tables and get start getting the data. I think that data can be fed to to to you know products like Igneo and to AI. And that's where that intelligence could come in. So it's not always the kind of rip and replace when it comes to to the legacy systems. They are they have been there for reason.

Roopinder 45:23

Yeah, we like that. Engineers, we we trust them, we like them. We we got used to them. So we we agree, those are good systems. Let's let's leave them in place, let's not reinvent that wheel, let's not rip and replace what it, you know, because I still don't trust LLMs for almost anything that I need as a uh that's mission critical, right? Yeah, I mean I asked it initially I asked it like, oh, find an IBM for me, and uh, you know, and it and it screwed that up. But I guess like, oh my god, this is like not how I beams work at all, and uh it's not how they're labeled. It gave me strange answers for gave me different answers for numerical problems. I was testing physics problems, it uh totally and things like that have given it a bad name. And I think because LLM was was called on to do everything, which it was never supposed to do, it's just supposed to give an answer based on language, right? Not engineering, not processes, not business, right? It needs to be more reliable. So I think you and I agree. Like let's leave these core systems in place, they work, let's just make them accessible and and uh work work with us in our language. We don't want to learn their language, we want them to correspond to our language, right? Yeah, so Emmanuel. So we said before, we think the Bay Area is the center of the universe, all the tech is here. You mentioned China, right? China is, I think, is some in some ways is catching up to us in technology and in AI. It's like it's right at the doorstep, right? It was like they're doing so much work and of the national might behind it to propel it. I think it's quite a danger of it coming to the lead. Well, you and I, you're Indian, I'm Indian, right? And okay, where the hell is Indianists, right? Here we have like two technology leader in the US, we have a government-inspired leader in China, but you're not the first Indian we've interviewed, and uh and I just wonder all the time like where's all that brain power, right? Now that we in AI, right? Do you think India is going to take over here a little bit? Not just individually, right? Not just you and I and other people, but yeah, as a nation, right? Why can't India get behind this thing?

Amit 47:55

Yeah, I think uh, I mean, more more from the I think the information technology point of view, I think um, you know, I think we have always been in that more predominantly the in the services game, right? But now it is it is changing and it it has to change, right? I mean, there are there are these um GCCs which are being established by all the major enterprises across the world. They are doing a lot of things. But I I would not say we are we are behind there. I mean, in fact, uh, I don't know, Rooinder, when was the last time you were in India? When I go to India, I feel such an out of place because I always carry cash with me because I don't have Indian uh like the nobody uses the the cash or the or even the credit cards there. It's all mobile app based. This uh Google Pays, and you know, there are like four or five predominant ones, and it's like they are really ahead compared to the rest of the world in terms of this uh electronic payments. And I think, of course, pandemic and COVID had one big re which was one of the bigger reasons for that. But there are distinct advancements that have happened in that space, and I don't think why it can't be done for the uh in the space of AI and and you know the the areas we are talking about. Yeah, yeah, yeah. So there have been certain miss opportunities in the past, you know. I think uh I I remember growing up as a kid, I was reading a newspaper, and one statement I distinctly remember by Mr. Our great uh Mr. Ratan Tata that India is has missed an opportunity to be to be a factory of the world, right? And that has gone to uh to China. So I don't think uh I mean, yeah, like I said, for various different reasons that might have happened, but when it comes to the technology AI, I think uh yeah, it's it's it's there, it's right there.

Closing And Listener Invitation

Roopinder 49:40

I think there is should there should be a hook there. Eventually it uh yeah, I missed the manufacturing thing. It caught on with IT a little bit because we import a lot of US gets a lot of IT help from India, but yeah, here's another opportunity, AI, which uh I think India should jump on. Okay, that's great. Thank you very much for that. Amit, it was wonderful to meet you and thank you for talking uh on what did you made as well as in general about AI and and the world, as it turned out. So thank you.

Amit 50:14

Thanks for having me.

Roopinder 50:16

Thank you for listening to FoDES, the Future of Design and Engineering software show, brought to you by ENGTechnica. I hope you have learned of a new application or technology that will help you with your job. If you have an application you think would be of interest to other engineers, please let me know by emailing me at roopinder at engtechnica.com or message me on LinkedIn.