Why a Great Idea Is Only Step One in Innovation — Robin Puskas, JPG Resources

The route Robin Puskas took into food and beverage innovation began with a master’s degree in poetry. After close to a decade in the software industry, learning how products get built and how to think rigorously about the user, Robin enrolled in night classes at the French Culinary Institute, keeping a knife bag under the desk during the working day. Volunteer work at New York’s Union Square Greenmarket led to recipe development for farmers, and then to a chef role at a Brooklyn manufacturing plant producing marinara sauce, a business that was later sold to private equity and subsequently to P&G Foods. 

That progression from home kitchen to scale manufacturing now underpins Robin’s work as Chief Innovation Officer at JPG Resources, a food and beverage consultancy of roughly 70 people running well over 100 projects a year. Robin also leads the firm’s AI practice, handling internal implementation and building bespoke tools for client innovation teams.

What does the role of a Chief Innovation Officer actually look like day to day?

Robin Puskas describes a consultancy that serves both global manufacturers and first-time founders, and a working week still anchored in the kitchen.

“My role is really to help elevate and expand the thinking of innovation teams so that they can create better experiences for people through the food that we eat. We are a food and beverage consultancy, and our range of clients runs from the very large, the PepsiCo’s and Mondelez’s and Unilevers of the world, all the way down to teams that consist entirely of a founder and a co-founder.

There are really two sets of innovation teams I serve. One is our internal team: product developers, strategists, and the operations teams that stand up manufacturing and supply chain. The other is the innovation teams of our clients. It is not uncommon for a global innovation team to come to us and say they are looking to build out their pipeline for the next five years, and to ask us to join them as thought partners so that on the other side of it they have built a vision for where they need to go next.

The bulk of what we do is product development and formulation. We have a very large number of food scientists and culinary professionals doing the nuts and bolts of making a cookie, making 20 versions, baking them off, tasting them, grading them, and then going through that review process with our clients. A very significant part of my day-to-day is centred there. We may have a cutting where you take a small bite, but you might taste eight to ten different cookies in a sitting.”

Where does AI genuinely add value in food and beverage innovation?

Rather than treating a model as an oracle, Robin Puskas argues that the quality of the output depends entirely on the quality and diversity of the human references fed into it.

“The maxim is garbage in, garbage out. If you just flop a prompt into Claude or ChatGPT asking how to make spicy popcorn, it will give you a formulation. It may or may not really work, and by the very nature of these tools, it will not be any different from what already exists. It is pulling on the vast body of knowledge it has ingested and giving you the average.

For both strategic thinking and product development, what we need to do is bring the right references so that we are getting meaningful output. There is a lot of conversation about humans in the loop, and of course that is extremely important. My finding is that having a community in the loop, a group of diverse perspectives, gives us the best possible output.

On a recent frozen product, we had one developer who is an expert in flash freezing and freeze-thaw, and another with deep culinary reference for what the flavour profile was meant to be and how it connects to the culinary heritage it came from. Before these tools, we would of course have had those two people talk to each other. Now we bring in other voices too, and not just their conclusions: what are the menus in your mind as you think about this, what are the manufacturing specs you know around freeze-thaw. We build a body of shared knowledge and use that to guide the formulation. It works because the shared human perspective is the starting point, rather than asking for the average way to make something.”

What do people most often misunderstand about innovation?

Robin Puskas sets out why the celebrated moment of insight is the least demanding part of the process.

“One of the big misinterpretations is that if you have a good idea, or even a great idea, you have already won. The great idea is square one, and your whole journey is yet ahead of you. Innovation is not simply the spark of a great idea; it is really embedded in the execution every step of the way.

A great idea then needs to be held by the development team, and they in turn need to be open and creative and curious about how to make it. If we hit a roadblock, if it costs too much, how do we fix that? What are the pivots we make in the middle so that a roadblock is not the end? The distance travelled between the great idea and having something that is really in the world and serving people is very great, and it requires innovative and flexible thinking all the way down the line.

This is also a place where we find AI models are extremely useful, and it is a surprising application: they can be an avenue for empathy and perspective-taking. There is often tension between the strategy team and the person who actually has to make the thing. Using tools that can bring on an expert voice in dialogue can be a good precursor to human-to-human conversation. It does not replace it, but it lets us ask early on what our supply chain person is going to think about our bright idea, and it allows us to shift a seat or two over before we come together to make decisions as a team.”

Will product development become more data-led, or will creativity remain central?

Robin Puskas points out that the industry has been data-led for generations, and locates the real opportunity in the hybrid.

“There is no replacement for real creativity. We can use new tools to expand our point of reference, but we will never reach a time where creativity is not central to building something delightful, enticing and new.

I would also say that strategy, pipeline building and product development have always been extremely data-led. We have collected data on consumer response for decades. It was much more analogue; it was getting eight people in a room with a score sheet. But those points of data were very significant drivers for decision-making, sometimes in ways that worked and sometimes in ways that were not a good signal for performance.

So we will lean into this more, but the real gains will be made in the hybrid: a collective community of individuals in a room trying eight cookies at the same time as we gather the vast amount of knowledge available to us, learning how to sift it and find real signals versus noise, and learning how to use that data for the good of the people we intend to serve.”

What will most change how products are conceived and brought to market in five years?

Looking ahead, Robin Puskas expects social listening to lose its value while tools designed to stretch human thinking take its place.

“Data scraping and pattern making from scraped social media is not going to win. That is mass-available information, and mass-available pattern making gives everyone the same answer, so there is no creative pathway that can be built out of that model. What we will see instead are more robust and well-built tools intended to push the critical thinking of humans and teams.

We built something for an innovation team that started with an exercise we called this or that. Is your frozen product the Eiffel Tower or the St. Louis Arch? Opinions varied. Some people said the Eiffel Tower, because it is iconic and recognised around the world and everyone has an emotional response to it. Others said the St. Louis Arch, because it is homey and Midwestern and more of our consumers have actually been there. That metaphor became the data set we pulled into the tool, so that our team’s creative leaps were part of the idea generation. That is very distinct from asking what people are most talking about on TikTok, which does not mean they are going to buy it.

On trends, I think two currently separate movements will converge. There is a lot of emphasis on macronutrients and on making that highly personalised. There is also a lot of pressure around defining ultra-processed foods, which I think is a semantic debate that will not hold. Is bread an ultra-processed food? Is cheese? What people are really looking for is food they recognise, and food traditions going back generations align in many ways with the macronutrient conversation. I would also be remiss not to mention GLP-1s. They are here to stay, and they have really shaped eating habits. I think we will see continued reduction in products designed for craveability and repeat snacking, with much more focus on small meals.”

What is the one question every food and beverage leadership team should be asking?

Robin Puskas closes on the question that keeps a pipeline honest, and on a lesson learned at culinary school.

“Are we really meeting a need? It is just making sure we recentre on what we are doing here. Are we bringing something that is in the service of the betterment of people’s food experiences? Are they going to have an elevated day for having experienced this?

That question is the antithesis of cynicism. Is it delightful? This comes from my poetry background: one of the definitions of a poem is to instruct and delight. Can we bring that sense of delight into the room, and make sure that when we are making something it is in the service of the people who will be buying it?

When I went to cooking school, Jacques Pepin was the dean at the time. This was when the Food Network was very popular, and a lot of people thought cooking school was a pathway to stardom. He came in and talked to us really beautifully about how you are entering a service industry. We are here to make and to serve, and there is dignity in working in service. It is a deeply human experience and one to be proud of.”

Join the Conversation: The Ortus Club’s Executive Network

As Robin Puskas has shown, the hardest problems in food and beverage innovation are rarely technical in isolation. They sit at the junction of consumer need, manufacturing reality and commercial timing, and they are sharpened considerably by hearing how other leaders are approaching the same tensions.

The argument for a community in the loop applies well beyond the formulation bench. Leaders navigating AI adoption, shifting macronutrient expectations and the operational consequences of GLP-1s benefit from the same thing Robin builds into his tools: a wider set of expert references than any one organisation holds internally.

At The Ortus Club, we host curated executive roundtables that bring together senior leaders facing these exact challenges. Move past the trend report and into the kind of open, high-value conversation that connects consumer insight to what can actually be made, at scale, at a price people will pay.

FAQs Section

Q: What does “community in the loop” mean in AI-assisted product development?

A: It extends the familiar human in the loop principle. Rather than one person guiding a model, a group of specialists contributes their references, menus, manufacturing specifications and sensory experience as shared inputs, so the output reflects genuine expertise rather than an average drawn from public data.

Q: Why is a great idea only the starting point in food innovation?

A: Because the value is created in execution. A concept must survive formulation, cost constraints, manufacturing feasibility and repeated iteration. The distance between a strong idea and a product serving people on the shelf is where most innovation work actually happens.

Q: Do food brands need their own manufacturing infrastructure to launch something new?

A: Not necessarily. Consultancies help founders find right-fit manufacturing partners at the right volumes and cadence. For genuinely novel products, there may be no existing way to make them, which means partnering with a willing manufacturer to build specific capability, typically over multi-year projects.

Q: Is being on trend enough to make a product succeed?

A: No. A product that is on trend but too expensive, or that is delightful without serving a real purpose in someone’s life, will not earn repeat purchase. The stronger question is what consumers will genuinely want or need in three to five years.

Q: What skills will matter most for people entering food innovation?

A: Collaboration with humans, including treating the AI-generated perspective as another seat at the table, and flexibility. Teams frequently build something to roughly 80 per cent, scrap it, and start again, so a willingness to abandon your own preferred idea is essential.