‘Sell before you build’: Chef Robotics Rajat Bhageria on how to de-risk development in the hardware business

In this candid conversation with George Thangadurai of Avataar, Rajat Bhageria of Chef Robotics shares bold aspirations not just to automate food prep but to inspire a broader robotics revolution akin to the recent AI wave, including a growth strategy for Chef and leadership lessons.

In a world increasingly shaped by artificial intelligence, Rajat Bhageria’s Chef Robotics is turning heads by applying AI not to screens, but to skillets. 

As Founder and CEO, Bhageria is leading the charge in bringing advanced robotics into industrial kitchens, tackling one of the most pressing challenges in food manufacturing: labor shortages. What began as a practical decision to choose food over other physical industries, including agriculture and construction, has evolved into a pioneering company at the intersection of AI, robotics, and real-world utility.

In this candid conversation with George Thangadurai of Avataar, Bhageria discusses Chef’s origin story, early hurdles, and how “selling before building” helped de-risk development in a notoriously difficult hardware business. From securing marquee clients like Amy’s Kitchen to embedding teams on-site to refine product-market fit, the company has relied on agility and deep customer collaboration to grow.

Bhageria also reflects on leadership lessons from operating with low ego and surrounding himself with mentors, to obsessively focusing on hiring senior technical talent. He shares bold aspirations not just to automate food prep but to inspire a broader robotics revolution akin to the recent AI wave.

With real robots now live across major food producers, Chef is reshaping how meals are assembled, improving throughput, reducing food waste, and unlocking new production capacity. As the company eyes expansion across food categories, international markets, and adjacent sectors like pet food and airline catering, Bhageria’s long-term vision is to make robotics as inevitable in the physical world as AI is becoming in the digital one.

Equal parts engineer and realist, Bhageria isn’t worried about robots going rogue. “AI is just a bunch of matrices being multiplied,” he quips. What keeps him up instead is the scale of the opportunity and the responsibility to lead it right.

‍

[Watch the full conversation here]

‍

‍Excerpts from the interview.

Founder's Vision & Journey

George Thangadurai (GT): What inspired you to start Chef Robotics, and what gap did you see in the food industry that robotics could uniquely solve? If I’m not wrong, Chef is the first robot to be installed in a food assembly line, and you don’t come from the food industry.

Rajat Bhageria (RB): Yes, that’s correct.

AI is going to have a massive impact on our world, but the big question is: will it stay confined to the digital world, or will it transform the physical world as well?

Around 90% of global GDP is tied to the physical world, so it makes sense to explore how AI can be applied there.

Why food? Honestly, it was a practical decision. I looked at the largest industries by labor force in the US, including retail sales, nursing, and personal care aides, and concluded these roles weren’t likely to be automated by AI anytime soon. Food, however, stood out. It’s a huge, tangible market where AI can have an immediate and meaningful impact.

Speaking to people in the food industry confirmed that there’s a massive labor shortage in the US. Traditional automation simply isn’t flexible enough to solve it. What’s needed is something adaptable that uses modern AI techniques to create the equivalent of a human workstation, thereby helping food companies overcome labor shortages and increase production capacity.

I did explore other sectors like construction, agriculture, and automotive. But in the end, food struck me as the most promising market to tackle first.

GT: What were the earliest challenges you faced in bringing robotics into commercial kitchens, and how did you overcome them?

RB: In robotics, building a single prototype can take six to nine months, so you only get one or two real chances before you run out of money. That’s why it’s critical to skip unnecessary detours and move straight toward building the right product. Our approach differed from the standard Silicon Valley “lean startup” model, which works well for SaaS, where you can build a prototype in a weekend. In robotics, six-month iteration cycles make that method financially risky.

Instead, we sold the product before building it. This way, we secured real contracts up front, ensuring customer demand while reducing their risk. If the technology didn’t work, they wouldn’t have to pay under our Robotics-as-a-Service model. This not only gave us confidence but also helped us gather clear product requirements from multiple customers, which we merged into a single Product Requirements Document (PRD).

Another key move was embedding ourselves deeply with our early clients. With Amy’s Kitchen, for example, we were on-site daily testing robots on their production lines, with their ingredients, run by their team, even before the systems were revenue-ready. This hands-on approach allowed us to fix issues in real time: from placement quality and ingredient spillage to unexpected daily variations in food prep, like how rice cooked yesterday might be different from rice cooked today.

We also encountered challenges we hadn’t initially considered, like food safety compliance and potential human safety hazards. By being there every day, we learned these requirements firsthand and adapted quickly. That early immersion in our customers’ environments taught us more about the food industry than we could have ever learned in isolation, and for any company building physical products, I think that’s invaluable.

GT: Did you face any challenges when trying to convince customers in the initial stages?

RB: Absolutely. In the beginning, most customers couldn’t even imagine that what we were building was possible. They were used to traditional food automation setups like big hoppers, mechanical dispensers, and very hardware-driven systems. So when we introduced the idea of a software-based robotic system that could make food, it felt alien and a little strange to them.

To build credibility, we started by working with marquee customers from day one, including Amy’s Kitchen and Sun Basket. These are well-known brands, and having them onboard helped us earn trust. The partnerships worked really well, and both companies ended up scaling with us. They were open to writing case studies about their experience.

Those case studies became a game-changer. They allowed us to walk into future customer meetings and say, “We get it, you’re skeptical. So were they. But here’s how it worked for them.” That social proof made a huge difference.

Another thing that helped was how we structured our contracts. We typically asked for a small non-recurring engineering (NRE) fee at the start, not a large amount, just enough to confirm they were serious. That fee covers our basic costs like flying out, lodging near their plant, and deploying the robots. We’re not trying to profit off that part.

Then comes the Robotics-as-a-Service (RaaS) fee, which is where our business model kicks in. But we only ask customers to pay that if we meet certain predefined criteria. These aren't criteria we impose, but something we define together with the customer. Things like throughput, consistency, placement accuracy, food safety, and whatever matters to them. We bake those into a site acceptance test, and if we pass, we move forward. That reassures them that they won’t end up stuck with something that doesn’t work.

And finally, there’s nothing quite like seeing it to believe it. In the early days, we invited customers to our office. We’d ask them to send their ingredients and trays, and we’d set up a demo, a mock assembly line with their materials. We’d walk them through the entire process: how to use the robots, sanitize them, switch between meals, and change ingredients. We even tested edge cases like what happens if a tray rotates, or if there’s no tray at all. They got to see their actual assembly line simulated and working.

That hands-on experience built a huge amount of trust. It made the whole thing real. This wasn’t some futuristic gimmick, it was their process, now automated and working in front of them.

Technology & Innovation

GT: Chef Robotics operates at the intersection of AI, robotics, and the food service industry. How do you balance technological sophistication with the practical realities of busy kitchens?

RB: There is a lot of traditional automation. But George, as you said, it tends to be fixed.
The robot is very good at doing the same thing over and over again. I mean, you probably have seen these robots in car factories, right? They can take the exact same door and put it in the exact same place, but if there is a little bit of difference, then the robot will fail, or if the car is in a slightly different place, the robot will fail.

The food industry has different challenges that are inherent in it, which make these traditional pieces of automation not work. One is, of course, any given ingredient. Just pick an ingredient like chicken. Every single ingredient in it changes day by day. So that's just within an ingredient. Now, of course, there are thousands of ingredient classes, and because of all this, you really need to leverage artificial intelligence (AI) to make sense of it and do something different every single time.

So what do we do? Well, we have a RGBD camera, D as in depth, that's pointing at our ingredients, and that's useful to understand not only the topography, but also things like what kind of ingredient. And then what we can do is we have different AI policies on how to manipulate different ingredients. For example, if I'm picking up blueberries, I probably need to apply less pressure for my end defector, so I'm gonna switch to blueberries. Similarly, for something like mashed potato, the pressure will be different.

Similarly, on the placement side, we have some computer vision models that are detecting and tracking the trays, and this allows us to do things like still be able to place food on a try that is misaligned in a way that's aesthetically clean.

These AI models allow us to make sense of that and dynamically react as opposed to just doing the same thing over and over again.

‍Impact on the Food Industry

GT: I'm curious to know how you see robotics reshaping labor, efficiency, and quality in food production over the next five years? It’s not just about replacing human labor, right?

RB: That’s a great question. And honestly, it goes beyond just replacing humans, it touches on something even more fundamental: a deep and growing labor shortage in food production, at least in the US.

A couple of months ago, we were visiting a customer site in Ohio. We took an Uber from the hotel, and as we got chatting with the driver, she mentioned she used to work at that very plant. Seemed like a fun coincidence until she added that the company had essentially hired almost everyone available within a 50-mile radius. There were just no more people left to hire.

And that’s the reality. These are tough jobs, physically demanding, repetitive, and usually done in cold-room environments. Many of our customers tell us, “If we don’t solve this in the next five to 10 years, we won’t be able to hire people to do this work at all.” It’s not just a labor issue, it’s an existential risk to their business.

So when we talk about robotics, it’s not about replacing people. It’s about survival. Automation has become strategic. There’s simply not enough labor to go around and the future doesn’t look like thousands of humans assembling food by hand every day. That’s just not sustainable.

We hope Chef Robotics is part of the solution, helping make that future possible. Now, if you look at it from a practical ROI perspective, there are four main benefits our customers see. One is unlocking revenue. Let’s say Line 10 can’t run because there aren’t enough workers. Now, with Chef, Line 10 can run, and that directly translates to millions in revenue. That’s the biggest ROI, turning idle capacity into production.

Second is cost saving. Our model is Robotics-as-a-Service. The annual cost of our system is typically less than the cost of manual labor doing the same job. So yes, over time, it is cheaper than the status quo.

The third is consistency and reduced food waste. Roughly 50% of a food company’s revenue goes into raw materials. Humans, as a general trend, tend to over-deposit, whether due to fatigue, error, or habit. That means food waste and higher costs. Robots are far more consistent and that reduction in waste has a major impact on margins.

The fourth benefit is increased throughput. Robots maintain a steady pace and don’t tire out mid-shift. Humans might start strong, but five hours in, fatigue kicks in and productivity drops. Chef Robotics operates like a metronome, it is consistent, reliable, and efficient, which means we can boost average throughput across the board.

So when we think about the future of food production, it’s clear that robots will play a critical role. And we’re working hard to ensure our robots are the ones helping shape that future.

Growth & Strategy

GT: That’s interesting. What’s a scaling strategy for Chef Robotics?

RB: Great question. For now, we’re very focused on a direct go-to-market strategy. That’s because each of our customers, even at the smaller end, represents a pretty substantial opportunity.

We’re not talking about customers buying just 2 or 4 robots. Even the smallest customers typically have potential for 50+ robots, and our largest customers could scale up to 500+ robots. So for us, it's not about winning the first sale but it's about building a relationship that scales. We think in terms of starter packs, maybe a customer begins with 4 robots, but we’re already planning for how they can grow to 50 or more. That requires close partnership management.

That said, we are developing partnerships with adjacent players in the ecosystem. For example, wherever you have Chef robots, you usually need conveyors and vice versa. So we’re exploring partnerships with conveyor system providers to share leads, collaborate on marketing, or even set up referral programs. They may not do the deployments themselves, but they can definitely help bring in leads at the top of the funnel.

Looking ahead, especially as we expand internationally, we know we’ll need strong local partners, particularly in markets like China or Japan. There are many cautionary tales of companies trying to go international without local support and failing. So yes, as we scale globally, partnerships will be key, especially in Asia. But for now, in the US, Canada, and likely Europe, we’re aiming to do most of it directly.

GT: In the next 12–24 months, what are one or two exciting developments on Chef’s roadmap? You are at the intersection of two bleeding-edge fields — AI and robotics. What should we be watching for?

RB: Absolutely. There are a couple of things we’re really excited about.

First, the advances in foundation models and Transformers are powering a new wave of innovation in robotics. Transformers are the architecture behind large language models like GPT, but they also enable things like diffusion policy, which powers imitation learning and learning from demonstration.

A great example is Tesla’s Full Self-Driving (FSD) Version 12. It’s trained on thousands of hours of human driving data. Now, imagine applying that concept to food robotics: a human demonstrates a task once, and Chef learns to mimic the motion without us writing any custom code. That’s a huge leap forward, and we’re actively exploring how to make this possible with our systems.

The second exciting area is more specific to Chef's product evolution. So far, we’ve focused on scoopable food, which accounts for about 70–80% of food processing. But now, we’re starting to explore piece-picked items, things like individual chicken breasts, salmon fillets, and cherry tomatoes. In these cases, it’s not about volume, but about precision and count.

This work is enabled by recent breakthroughs in hardware (end-effector design) and AI-powered perception systems. For example, tools like Meta’s “Segment Anything” model and modern deep learning segmentation techniques have made it possible to detect and isolate individual ingredients with high accuracy even in complex, dynamic environments like a kitchen or assembly line.

So we're very excited about combining this new vision technology with advances in safe food-grade hardware to bring automation to a whole new category of food handling, like produce, bakery, and ready-to-eat meals that require delicate, precise handling.

GT: I know your initial and primary focus has been on food assembly in industrial kitchens, and I was fortunate to visit two of them, thanks to you. That was a really enlightening experience. So what’s next in terms of growth plans? You’ve focused a lot on frozen meals, you know, the ones you see in trays at grocery stores but there are so many adjacent food assembly markets, like airline catering. As you grow more confident in your core segment, how do you see Chef Robotics expanding both within existing verticals and into new ones?

RB: Yeah, great question. There’s a ton of opportunity ahead of us.

Right now, we’re already working across multiple segments: frozen prepared meals, fresh prepared meals, direct-to-consumer offerings, and contract manufacturing for meals. These are all within large-scale industrial kitchen environments.

But even if we did nothing else, just scaling within our current customers is a huge opportunity. Many of them operate across dozens of facilities, for example, one of our customers has 13 plants just in the US. If we can go deeper with each of these clients, the market is massive.

Then there’s expanding our use cases within the same facilities. For instance, if we add capabilities for piece-picked items like salad dressing packets, croutons, forks, or fruit like diced apples, we can deploy additional robots across existing production lines. So even within a single customer’s operation, there’s plenty of room to grow.

We want to increase both breadth and depth within our current customer base, adding more robot types and automating more steps in the food assembly process.

After that, international expansion is a natural next step. Food manufacturing is a global industry, and there's a lot of opportunity in overseas markets.

Then there’s the potential to move into adjacent industries, like produce packaging, where workers still manually count and place items like pears into boxes. Or meat processing, where companies like Tyson and Smithfield employ tens of thousands of people. We’re also looking at large-scale bakeries, not your local bakery, but big industrial manufacturers of bread, pastries, and snacks.

You mentioned airline catering, which is a perfect fit. Meals are pre-portioned, tray-based, and assembled at scale. We’ve also seen interest from pet food manufacturers, which is another huge space.

The beauty is that we can build a massive company just by staying in the industrial kitchen space. There’s so much unmet need for automation.

GT: And then maybe even go downstream into service environments?

RB: Exactly. Once we’ve built out our capabilities in manufacturing, we can start moving into lower-volume but high-frequency environments. For example, ghost kitchens, fast casual chains doing large delivery volumes, school meal programs, hospital food services, stadiums, music venues, and cruises, anywhere with a commercial kitchen. 

But right now, we’re focused maniacally on industrial kitchens because the scale and repeatability are so high. That’s the fastest way to build a scalable automation platform.

But the big picture, and the reason I started Chef in the first place, is that food is one of the largest markets on earth. It’s something people in Silicon Valley often overlook because they’re so focused on software. But when you step outside that bubble, you realize that the physical world is orders of magnitude larger than software.

GT: Totally agree that Physical AI is the future. Chef is a great example of what happens when AI moves beyond the screen and into the real world.

RB: Exactly. And here’s the kicker, each of these industries is massive in its own right. Pet food alone is a multi-billion-dollar space. So are bakeries. So is produce. You take any one of these segments, and it could support a $10B company.

Take a company like Dole Foods, about a third of their revenue goes to labor. That labor spend, that’s our addressable market. Of course, we won’t capture all of it, but it gives you an idea of the scale we’re talking about. Labor is the single largest market in the world. It’s worth over $45 trillion globally, and food is one of the biggest pieces of that.

So, the way we see it, if we can automate food assembly well, we’re not just solving a niche problem. We’re tapping into one of the biggest opportunities on the planet.

Leadership & Lessons

GT: You’ve built a really impressive company, and the proof is in the pudding. Your robots are live in real-world environments, customers are happy, and your growth path is exciting.

But let’s take a step back. Starting a robotics company is no small feat, especially when you're dealing with both hardware and software. Building the team, designing the product, handling deployments, it’s all very complex.

As the CEO and founder, what have been some of your biggest leadership lessons along the way? Any mistakes or challenges that shaped you?

RB: Absolutely, great question. And yes, I’ve made plenty of mistakes, probably all the ones you can make in the book.

But I think one thing that’s helped me is that I try to operate with a very low ego. If I don’t know something, I’m not afraid to ask for help. Early on, I made it a point to build a group of advisors and mentors across different areas, including, engineering and product management, enterprise sales and hiring, and leadership and culture. 

For each of these functions, I found someone I genuinely admire and asked them to guide me. I’d meet with them regularly and just soak up as much as I could.

I’ve also had a few core mentors I’ve leaned on deeply. One of them is the founder of Indiegogo, we’ve known each other for nearly a decade now, and he’s been a constant source of perspective and wisdom. I also worked with a leadership coach, and I’ve built a network of founder friends who are going through similar journeys.

I surround myself with people I respect, and I’m not afraid to say, “I don’t know.” That humility, I think, has been one of my biggest assets as a founder.

GT: That’s a rare and powerful approach.

RB: Thanks. The other thing I’ve realized is that people often think you need decades of experience to build something great. And sure, experience helps no doubt. But there are plenty of counterexamples: Zuckerberg, the Carlson brothers, even Elon Musk in his early days.

The common thread? They were insanely curious and relentless about learning.

Most decisions aren’t that mysterious, the real challenge is having the courage to execute. Like, yeah, you know you need to grow sales. Or you know you need to raise that next round of funding. But actually doing the work, navigating the uncertainty, and taking the leap, that’s the hard part.

So I think the biggest lessons have been: stay humble, keep learning, be brave enough to take risks, and remember, there are no shortcuts. It’s just doing the work, day in and day out.

GT: What advice would you give to other founders trying to introduce disruptive technology into a traditional industry? 

RB: Great question. We touched on this earlier, but I’ll emphasize it again, sell before you build. That’s one of the most important things. The number one reason startups fail is that they build something no one wants.

If you sell first, you’re validating that someone actually wants what you're building. You’re getting real product requirements, and in many cases, even a signed contract. Once you have that, your job becomes execution. You know what the customer wants, and you’re building to meet that.

The other big piece of advice is, obsess over hiring.

A lot of founders are told to "hire your friends," and yes, there’s truth to that. Your friends are likely smart, motivated, and hardworking. But if you’re building something deeply technical like robotics, advanced AI, or hardware, you probably don’t have senior engineers in your friend group, especially if you’re just out of school.

So, you have to go beyond your immediate circle and put in the work to find senior, high-caliber talent. In the early days, I spent 100% of my time on hiring. Even now, it's still 20–30% of my week. I treat these hires like investors, they’re investing their time, which is even more valuable than money. Time is non-renewable.

The Vision for Chef Robotics

GT: What’s your long-term vision for Chef? Are you aiming to automate the entire kitchen, or are you thinking even bigger?

RB: I’d love to see thousands of Chef robots deployed around the world, making meals at scale. No one has really done that yet in robotics. If we can be one of the early success stories, it would be game-changing.

More than just growing our own company, I want Chef to be a lighthouse, the kind of company people point to and say, “See, robotics can work!”

Think about Waymo in autonomous vehicles, it’s magical to see it drive itself around a city, and it inspires people. The same goes for ChatGPT, it inspired a massive AI wave. I want Chef to be that kind of inspiration for robotics.

If we succeed, I hope we can help shift the narrative and make robotics the next great trend like what’s happening in AI right now. Success has a ripple effect. When one company proves it’s possible, it opens the door for many others to follow.

How Avataar Helped Us Scale

GT: Avataar hasn’t just funded Chef, we've partnered deeply on growth. Can you talk about that relationship?

RB: Absolutely. Avataar has been a fantastic partner. Beyond the capital, which itself was strategic and well-timed, they've really doubled down on us, and that commitment means a lot.

Avatar has helped in several key areas, including,

• Debt Financing: We recently closed a significant tranche of debt with Silicon Valley Bank, and Avataar was instrumental in getting that deal across the finish line.
  
  
• Strategic Advice: Whether it’s discussing sales strategies or how to balance different growth levers, they’ve been great mentors.
  
  
• Fundraising Support: They’ve helped us build investor relationships for future rounds.
  
  
• Talent Acquisition: George and the team helped us close key senior hires, introducing top candidates and helping them see why Chef is the right company to bet on.
  
  

I’m really grateful for the partnership, it’s been meaningful, strategic, and values-aligned.

GT: Alright, one last fun question. Have you ever imagined a scenario where your robots go rogue? What would that even look like?

RB: Not really! I know there’s a lot of fear out there about robots taking over. I’ve heard it a bunch from people outside the industry. But when you're the one building them, you realize just how dumb they actually are in some ways. I mean, yes, they’re doing intelligent things but at the end of the day, AI is really just matrices being multiplied, over and over and over again. That’s it. So no, I’m not too worried about a robot uprising anytime soon.

‍