Transcript.

Peter: Sahitya, would you mind giving a brief introduction about yourself for the audience?

Sahitya: Yeah, I'm happy to. Peter, thanks so much for having me on. I really appreciate it and happy to do a quick intro. My quick background is, like you mentioned, I grew up around the automotive industry. I was born in Michigan. My parents worked in automotive. They were lifers at Ford, Chrysler, GM. Instead of going to daycare after school, I'd go to the IT department where my parents would work at Ford. And you could imagine there's a toddler, little me just running around. Maybe through osmosis, just being surrounded by all the programming that was going on at the manufacturing plant, I became very interested in computer science. I was coding since I was a kid. By the time I was 11, I found myself programming at the level of most college students.

So one time, I noticed that, in newspaper, FEMA was having some troubles with managing the response. I decided to cold email the chief of FEMA at the time and let them know that an 11-year-old could help them with building an app for them. For some reason, they decided to take a chance on me. When I was eleven, I helped build an app for FEMA that they ended up using in field testing. I like to joke that once one government agency finds out about you, the rest of them do too. The team of FEMA introduced me to the U.S. Army, which funded my education for the next three to four years. By the time I was sixteen, I found myself at the Air Force, at Kirtland Air Force Base, instead of in high school. I was working on autonomous drones there, deep reinforcement learning, working on the cutting-edge AI technology of the day. It was an amazing experience. But around the time I was eighteen, I decided, "Hey, maybe I want a girlfriend," and I decided to go to college.

I went to college, which was an amazing experience. I studied a bunch of things, from business to computer science to AI. By the time I graduated, I was at a company called Palantir—where I was doing things in the manufacturing space—and ultimately decided going around the country to these warehouses and factories and facilities. I loved the space so much. I wanted to go do something myself. And so I took the leap of faith and founded Endeavor. And here we are.

Peter: Sahitya, that was an awesome introduction. Thanks a lot. It basically already answers at least partly the first question I had for you. Because I wanted to ask you whether you could share more about your early experience in the manufacturing sector and how that influenced your decision to start Endeavor. Is there anything you would like to add from your perspective?

Sahitya: Totally. You know, what I can go into a bit deeper in terms of why I decided to found Endeavor—rather than going and working at a manufacturer myself or staying at Palantir—was because I'd go around the country to these Fortune 500 large enterprise manufacturers or wholesalers, and we'd be thinking about AI. This was, I think, 2022 or 2023. This is the boom of generative AI, the foundation. We had the Cambrian explosion of all these models. Instead of thinking about how do we use large language models, it just seemed like AI had become the new thing. People were just saying things that we had had for the last 10, 20 years. Like deep learning or inventory forecasting of computer vision were the same as generative AI. You and I know that that's not actually generative AI. We've had computer vision. We've had deep learning and forecasting for the last 10, 20 years. And so the same projects that you and I have seen get investment—like demand forecasting, like inventory planning, like transportation networks—were getting massive amounts of funding, $10, $20, $40 million contracts. But fundamentally, nothing had changed in the technology.

And so I was like, "Hey, why are we doing these really big contracts? We're getting paid more, but we're not able to drive any better results than we were yesterday?" So I began to think, how do we really use generative AI? In fact, at that time, I knew AI agents were going to become a thing. I began to think about the use cases where you could think about augmenting your workforce with AI labor and truly using generative AI. Because I realize it would require stepping outside of the box. It'd require thinking outside of the way the manufacturing and wholesale distribution has thought for the last two decades.

Peter: Cool. Yeah, so this is basically also what actually then motivated you to start Endeavor, right? So how did your experience at Palantir and AWS contribute to your vision for the company?

Sahitya: Well, I can tell you when I was at AWS, what I was working on was: how do we build code-generating large language models? This was way back when ChatGPT hadn't even come out. I was working on large language models before most people even knew about ChatGPT. The reality was, I was scared when I first saw it. I was talking to this model, and it was like I was talking to another person on the other end.

So going into 2022 and 2023, I knew the world was going to change. Because here was a thing that felt like artificial general intelligence. Here was something that was smarter than I was at coding. I only knew how to do Python, Java, and Rust. This thing knew every single programming language, and it knew it better than I did. I didn't even know how good these things would really get. When I went to Palantir, and we would do these projects for measuring things like overall equipment effectiveness or doing inventory forecasting, downtime and maintenance prediction, I was like, this is great, right? This is fine, but it's not truly groundbreaking. There's nothing here that's leveraging generative AI technology to really push the boundaries of what's possible.

And so what motivated me was saying, how do we think of ways to bring generative AI? Frankly, the way it came about was, I decided to actually just go work at a factory for three months. That's what I did. I worked at a steel factory, and I was doing all sorts of things, including being on the shop floor, looking at the purchasing department, and calling up people to buy raw materials. I was in the inside sales department looking at quotes and orders and figuring out how to price things. I was doing all sorts of cool stuff around the factory. I was the odd one out. I was the only guy under 30 years old who wasn't from the industry. But I learned a ton to the point where I realized the changes aren't going to come through these massive forecasting projects. Generative AI is not going to help with that, nor is generative AI going to help with robotics and automation on the floor. What it will help with was the back office and the front office and automating the highly labor-intensive processes in most of these factories, especially at enterprise scale.

Peter: Okay. So how does Endeavor's platform specifically address those challenges faced by the traditional manufacturing industries?

Sahitya: When we look at our clients—many of whom are really large enterprises, often public or privately held but doing several billions in top line or turnover—the big thing is: they're often dealing with very bloated back office or front office departments. Right? What do I mean by bloated? I mean, they've just got a ton of people. The reality is that there's a lot of manual work, right? You've got people doing manual things, and you've got people waiting for other people to do manual things.

So, for example, if we look at supplier onboarding, very common workflow, right? You need to bring on a new supplier, whether you're a manufacturer or a distributor and we're talking about your raw materials or products. I mean, the process could look like you need to have them do clients. You need to have the supplier fill out some forms. You need to look at some documents. You need to have someone internally go fill something out and send it to R&D, who then goes and sends it to the product information management group, who then goes and sends it to the inside sales department to set up a customer relationship. You can imagine you have a chain of these processes that need to happen—one person waiting for the other person, waiting for the other person, waiting for the other person. And before you know it, a process that should take a day has now taken three months. The impact is that you're not able to drive enough revenue, right? You're not able to sell the products that you want. Your suppliers are incredibly frustrated because it's taking so long to get up to date in your system. You're not utilizing your ERP system correctly. So it's extremely costly. And it's taking up resources from your labor, right? You're hiring high-value people to be on the phone with the customer, not doing data entry into SAP or Oracle. This is ultimately what happens. You'll get this more than any other person. So many people just doing data entry into the ERP system nowadays. So I answered the problem that we talked about.

Now, where I would say our platform fits in is all the problems that I've mentioned: the back-and-forth transactional communication, the manual document processing, information processing, of voicemails and emails and phone calls, the back and forth that occurs when someone needs to reference data in an ERP system, perform various lookups and calculations, and the data entry aspect of porting things into supplier portals and customer portals and into the ERP system itself. All of these, we place them to this whole bucket where, today, modern generative AI technology and in particular, agentic technology enables the full automation of these class of median complexity problems.

Historically, when companies have bought software, it has been for low-complexity problems, where you could apply an order-to-cash solution inside of SAP or a procure-to-pay solution inside of Oracle. What would happen is that you would have software that comes in and can automate fairly easy processes, like filling out a form. However, the problem that results from that is the medium complexity still requires a person to exist in the company, and they are working with the Software as a Service to perform the task. We are now moving to a paradigm where the agent can independently perform that task and interact with the various software as a service system—the SAP, the Oracles, the customer portals—and do the manual workflows that the person is doing. That's fundamentally what we're doing, which is bringing AI labor and AI workforce, you could think about it, to many manufacturers and wholesalers.

Peter: Cool. Okay. I'm getting a sense of it. All right. That's really cool. So looking at the foundation, what are your plans for using the US$7 million funding you have received? So how do you see Endeavor expanding its operations in the coming year?

Sahitya: It's a great question. Because right now, when we look at all sorts of industries, there's been such a great explosion of funding towards trying new innovative things in manufacturing, but also in healthcare and in finance and in legal. In particular, when we look at those other industries like healthcare, finance and legal, you have really large startups that have raised hundreds of millions of dollars. But in manufacturing and in wholesale distribution, we don't really have that. We don't have giant companies that have raised hundreds of millions for the pure purpose of bringing agentic technology. The reality is it's because, largely, the attitude that makes sense that most people have is: "I'll just wait a couple years for my ERP system to roll out AI agents." However, what we are seeing is that there's an adoption curve, the same way hundreds of people lined outside Apple's first launch for the iPhone. We had early adopters. We are seeing something very similar, where we work with transformative CIOs, CTOs, or digital transformation folks who want a seat at the table. They want to change the business, and they don't want to be doing technology for technology's sake.

What we are seeing is unprecedented demand for new technology. The last time most of these guys have procured new software has been their ERP system. But for the first time, they are ready to try something new, like our agentic platform. To us, that gives us a confidence that we are going to go invest $7 million in becoming the generation-defining company in the manufacturing space of the 2020s. We want to go build the next Oracle, SAPs that are going to be defining companies that will last in this space for the next several decades. That in 30, 40 years, we'll look back and we'll see most of the Fortune 500 industrial firms use Endeavor for their call centers, for their inside sales departments, for their finance departments, for their supply chain departments. That's fundamentally the goal, which is really built towards owning market leadership here.

Peter: Looking at the vision, how do you see Endeavor contributing in the broader revitalization of manufacturing in the U.S.? What role do you see AI play in this process, in a broader sense?

Sahitya: As of, I think, yesterday or the day before, new tariffs went in. The new administration has definitely put forward a willingness to implement tariffs. Broadly, it seems like the culture and environment and, largely, governmental attitude here as well has been focused on re-industrialization. We want more activity in the United States after decades of outsourcing. I'm not going to take a political stance about that. But what I will say is that there's going to be more manufacturing plants being built here, right? You're going to have new semiconductor activity. You're going to have new automotive activity.

What we see is that a lot of these are new businesses that need to be built up, and there's just not enough people to hire for them. There's just not enough people who can fill in for the labor because of the amount of retirements. The American industry has faced this wave of retirements. They call it silver tsunami. Folks have come into the industry, say, for decades. Now they're leaving the industry, largely hastened because of COVID and just age. But as we build more factories, we'll need people to fill in. When we're lacking labor, the answer becomes very clear, right? We'll need AI. We'll need AI workers and AI labor to work at these factories that are being built. But we want them doing things that really no one wants to do and no one wants to be hired for—like the order processing, like the supplier onboarding, like the accounts payable and the invoices and the deductions and so on, the manual workflow that we were talking about. So that's our vision for how we can help with the mission, the broader collective idea of re-industrializing America.

Peter: Yeah, have you faced challenges? I mean, traditional manufacturing, it's not always that open to latest tech. So what kind of challenges did you face?

Sahitya: Oh, certainly. Lots of challenges, I think, both culturally and technically, right? I think, culturally, when we look at it — I touched on this before, but things like, do we even need to get this agentic platform right now? The technology is moving so quickly. Should we just wait for our ERP system to roll this out? I think what some folks realize right now and what others do not is that the people that really invest in agentic technology now will be the winners in five years. Because we'll look back on this year, 2025, and we'll be like, "This is the year that things changed."

The reason being, it is such a force multiplier. If you could just have 600 extra employees on demand that are working 24/7 around the clock for you, but they're genius-level IQ robots that love doing sales order entry, or they love doing invoice matching, or they love doing quoting or shipping notices, right? That's the only thing they care about, and they work around the clock. And your competitors, if they choose to say, "Hey, I'm just going to wait for the guy next door to do the same thing," this is not like ERPs where we're building a database. This is fundamentally shifting the landscape of labor. Because if I'm a company that invests in agentic technology and I decide to onboard this AI workforce to my company, I may be able to reduce my prices way lower than my competitors, eat up market share. And in five years, I'm not going to have any more competitors. I'm going to have just acquired them all for pennies on the dollar. It's kind of scary, but it's also exciting for those ambitious manufacturing companies and wholesalers that want to tap into the AI workforce to really drive a competitive edge for their business. That's the first one.

The second one that I think you might find interesting—having done ERP services and being in the implementation world yourself—is interacting with a new age technology like agents and your relational database management system, or a hierarchical database, or these older systems that the manufacturing system is used to is very hard. Right? It's hard to connect the technology that has been developed in 2025 and a technology that has been developed in the late 1990s. There are some interesting things that I can go into there, if you think it's worth it, about, what is it like for an agent to interact with the legacy IT system?

Peter: Makes sense. Yeah, we really need to innovate. That's absolutely right. You already touched briefly based on that. What kind of technologies besides AI, what do you think will be the most influential ones in shaping the future of manufacturing?

Sahitya: I'm a little bit biased, but I definitely think AI is going to be super important. When I kind of look towards it, I think there's a variety of technologies and cool things that I've seen in the realm of now people developing generative AI kind of world models for the realm of robotics, right? I think there's this world of, for example, drones now being used in construction and crawling robots being used in inspections and asset management. There's humanoid robots now, right, that are beginning to work in packaging and logistics. A lot of them are being powered by and amplified by generative AI.

To me, it's glaringly obvious that AI is going to be the centerpiece of this. It's the biggest macro trend that the industry has seen. I could see things that we have been talking about for the last five, six years, like drones or even AR and VR—just being so much better now that we have a better understanding of the world around us. Because what that genitive AI world model does is it allows us to take all the information about the factory floor or the warehouse and really understand it, that traditional software that was driving the robot or driving the AR/VR program was not able to understand before. So I'm quite bullish on all the different sorts of technologies that are now going to be augmented by AI.

Peter: Yeah, make full use of what's available. Right. Yeah, makes sense. I mean, you have been very successful now. So what kind of advice would you give to others who want to start their companies in the tech and manufacturing sector?

Sahitya: Yeah, it's a great question. Because despite kind of growing up around the industry, I'm not someone who spent most of their life in manufacturing. I'm not someone who's worked at a large manufacturer for 30 years and became the CIO of the company and is then starting the company. I'm someone who graduated college and was like, "I really think that I can make a difference in the space."

Part of that is humility on my part, a willingness to come and learn. I think the first thing is, a lot of people in the valley, Silicon Valley here where I'm based out of, come into an industry. They come in with almost this attitude that's like, "You know what? I have this attitude of first principles. I'm going to break the rules and do things that no one has done before. That's why no one has succeeded in this space. I'm going to do things that are very different."

I think while that's admirable to think "I'm going to be different. I'm going to do things contrary to popular belief," I think there's an element of respect that has to go towards the people that are in the space. When we interact with CIOs and digital transformation folks who have been in the business for the last 20, 30 years, they know more about their business than we do. The reality is that what we can help them with is leverage agentic technology. But fundamentally, they are the ones that have a seat at the table. They know what the problems that their business is facing. They want to help the business the most, and they have the most to gain, right?

I'll give you an example. One of our early customers, we were working with a VP-level individual in their IT department. Over the course of last year, that person has gotten promoted to a C-level position because of the work that we've been doing with them—where we have helped that company use AI and, in particular, begin their first ever usage of generative AI. They have been able to gain so much by basically being a trailblazer in the organization. And if we came in and we said, "Hey, man, we don't need you. We know better than you. We know AI and you don't know," that would be wholly unsuccessful. I mean, most of these are partnerships. That's our approach, which is an advice to folks coming in, is be humble. Try to work in a partnership model with most of the folks in the industry because they know a thing or two.

Peter: Yeah, I actually agree with you on that part. Yeah, it's very important indeed. Yeah, the partnership model especially, I believe. Because you can benefit in both ways. Yeah, that's true.

Sahitya: Yeah, and I'm sure you do a lot of that too in the systems integrating and services role as well, right?

Peter: Yeah, it's similar. It's similar. You just have to have to stay up, you know. Don't lose touch to the ground. It's important, yeah.

Sahitya: Right.

Peter: So what makes Endeavor's approach different from other companies using AI in manufacturing? There are a few out there. I've seen stuff coming from Germany even. How does it set you apart? What's the difference?

Sahitya: You know, when we think about AI—we don't say this to just boost ourselves—but largely, if you were founded before 2024, 2025, it's very hard to be an AI company, a truly native AI company. Certainly, you can use AI features and incorporate them into your legacy architecture, right? If you're an ERP company, certainly, you can take an open AI, API, or a Microsoft Azure API and integrate it on top of your existing platform. You can say, "Hey, we have a co-pilot." There's a chatbot that you can talk to, and it'll look at your maintenance manuals, or your standard operating procedures, or your documents.

However, the problem that arises is that these applications are not natively architected for AI in mind, right? AI architecture and AI native architecture is fundamentally going to be different than an AI-optimized architecture. An AI optimized architecture is fundamentally going to be playing into the old world of things, the manual world, the world in which workflow is sequential and requires people to intervene and verify and check things and do lookups. Whereas our architecture is fundamentally AI native. It is built for an AI to use, not for a person to use. There's obvious drawbacks to that, which is, when you think about how people use it, it's like, yeah, our tool is not necessarily built for people to come in and use to make themselves go faster. But it is built to drive outcomes.

When we're asked, "How do you price," we say we price on outcomes. We price on the number of sales documents that we automate or the number of ERP systems that we migrate from one system to the other, right? We price on the number of invoices that we reconcile. This is a fundamentally different way of thinking, where we're not charging on a seat-based model, or we're not charging on the number of applications. Because we don't accrue value like that. We only accrue value on our AI labor providing outcomes directly to the customer.

Peter: Makes sense. So, closing the story, what would you like to give to our listeners about Endeavor and your vision for the future of manufacturing to just round it all up?

Sahitya: The one thing I'll say is that the world is dramatically changing, right? I think this is probably obvious to you and I and a lot of folks out there. I was just talking to a gentleman who was a C-level executive at one of the world's largest chemicals companies. He uses ChatGPT for advice on how to talk to his kids about certain topics, or how to write an all-hands email for his employees, or how to go about having a difficult conversation with this partner, right? There's all these very weird, interesting things. Where it's like if you told someone three years ago, would you be talking into a website about your family business, or how to talk to a difficult conversation with your employees, we'd probably say no. Right? We'd probably be like, "What are you talking about? You're crazy." But this is the new normal, which is we interact with AI on such a deep basis, where it's gradually becoming more and more intertwined with our lives.

The reality is that we shouldn't expect it to be any different in our professional lives as well, where if we are using AI so much in our personal lives, perhaps there is also a place for generative AI today in the industry, right? Perhaps we don't need to wait two years before we use AI. Perhaps we can just take the jump and try it and use it out today. A lot of that doesn't necessarily have to be FOMO. It can be FOMO that the guy next door is going to start using it, hire a bunch of AI workers and compress their prices and take on market share. But part of this is that there's tremendous benefits to it, right? We're talking about retaining employees. We're talking about providing a better experience to your customers, one that's touchless and fast. We're talking about making it easier to do business with your suppliers and reducing the errors in their processes. There are so many benefits that come from it. Part of this is by having a culture of experimentation and thinking forward about it. Because honestly, the way we think about it, there's not much downside to just using AI today.

Peter: Awesome. Awesome. Thanks a lot, Sahitya. Very, very good. It's a very good discussion today. Really appreciate your time and everything you could share with us. It's very valuable. I hope the audience will appreciate it as well. If you got any comments or any thoughts, please share them with us. We are glad to receive any kinds of feedback. All right. Sahitya, thank you very much. We'll keep in touch. Thanks a lot.

Sahitya: We will let you know. Peter, thanks so much for having me. Appreciate it.