Nutrios to Epicuri: The Triple Lutz of Startup Pivots

Story time, kids, because all great startups have a compelling backstory. Ours starts with medical school. Specifically when my wife started medical school early in our marriage and suddenly I had to fend for myself in the kitchen. Hamburger Helper was the extent of my culinary prowess. I was lost. And hungry.

Dating myself, but this was right at the upswing of the Food Network and Chef Emeril Lagasse. I'd watch him cook on TV, with more desire than a man should have for chicken, and just think to myself, "I want someone to bring THAT to me right now."

And thus began my quest to build a business around "bring THAT to me right now" which would take various forms, initially as just-in-time, locally sourced meal kits (a great business, victim of COVID, I WILL rebuild it one day), which conceptually morphed into gen AI recipes (GPT-4 put an end to that), and then Nutrios and our first foray into crypto.

The Emeril Dream Meets Reality

The original idea behind Nutrios was "bring THAT to me right now" where "THAT" was a menu item a chef had posted. It was "unbundled DoorDash" where menu content lived separately from production. A user placing an order would kick off an AI-driven process where the recipe would be matched to a local ghost kitchen for production and delivery. It was cool and it was my Emeril dream.

We had an operational problem, though. We needed a way for our chefs to own their content. Food blogs were a copy/paste/steal wasteland where nary an original idea existed. We didn't want that on Nutrios. We needed a definitive way for content to be owned, verifiably. Enter NFTs. Our operational problem was a perfect fit for what crypto provided: immutable proof of ownership of data. We're likely the only builders who came to crypto because we had a specific challenge that crypto solved.

From Solana to Aptos: The Great Chain Migration

When that revelation hit us, I rolled up my sleeves, found a great tutorial on Solana development from Loris Leiva, and built an MVP. Now look, my intent here is not to chain bash, but development in Solana/Rust is hard (partially from the fact that my entire experience with rust up to that point was battles with my patio furniture). That whole "chewing glass" phrase didn't invent itself. I mean, I'm literally using a calculator to figure out memory allocation. But I digress.

In March of 2022, after venting about the massive Solana outages and the operational risk they represented, a friend at a16z, Porter Smith, suggested I check out a new chain that was just launching called Aptos. He introduced me to the founding team. I had one call with those guys and was convinced. I spent a week doing a deep dive on Move and then made the command decision to dump our MVP and rebuild everything on Aptos Move (I will cover the "why" of this in great detail next week, for now, trust me. It was the right move.)

Identifying the Real Enemy

Now we're in execution mode, talking to customers, ghost kitchens, and restaurants. What we found was a battlefield. Ghost kitchens weren't what they were cracked up to be. Customers were locked into DoorDash/UberEats. Restaurants were getting crushed by 30% take rates, and even though consumers were paying more, the restaurants were getting less. My decades of experience as an armored cavalry officer took over: I identified the enemy, and that enemy was any gatekeeper that extracts value by hoarding information. We set our focus on giving restaurants the tools to bypass them.

But consumer behavior is hard to push. We tested tools that I would personally use daily with very little traction. The reality is when we want food, we start with UberEats and DoorDash—even though we know it costs more and the search is influenced by ad dollars. It's what we do. We were getting nowhere, an outlier crypto/AI company in a niche vertical in a post-FTX world. The night was dark.

The AI Catalyst

But then the world changed. AI happened. ChatGPT happened. Overnight, consumer behavior began to shift. We all began our search for whatever by asking an AI, not reaching for an app. It was the catalyst that gave us an opening. We wanted AI to do everything for us. Why not order food? We just needed to connect AI to the real world to do that.

And then, as my math professors at West Point would say, we had a realization that "should have been intuitively obvious to the casual observer"—connecting AI to real-world data to bypass gatekeepers is not a food-specific problem. It's a universal problem. It's a massive problem. And we had already built the core of the solution: a universal registry for declaring capabilities and a trust layer to verify ownership. We just needed to point it at the whole internet instead of just at the food industry.

The Triple Lutz Landing

And that's how we got here. The journey from wanting a chef to 'bring THAT to me right now' to realizing the entire internet needs a way for AI to do the same was long and winding. We had to learn firsthand that consumer behavior is immovable until a tidal wave like AI comes along. We had to learn that the real problem wasn't just unbundling DoorDash, but unbundling the entire internet's trust model. Every failure and every insight led us to this single, massive conclusion: the future doesn't belong to gatekeepers. It belongs to builders. And we're just getting started.

Next week: Why We Chose Aptos Move Over Everything Else. The technical deep dive into building a global state machine for autonomous agents.

Wayne Culbreth, Co-Founder, Epicuri