Close Ad

The Spoon

Daily news and analysis about the food tech revolution

Michael Wolf

Will Giving Everyone a Blood Sugar Monitor Lead to Better Health Outcomes? Maybe, But Only If We Tell People What to Do With The Info

Last year, I used a continuous glucose monitor (CGM) for the first time, and it completely changed how I eat.

After a couple of weeks using an over-the-counter Stelo CGM, I learned that sugary snacks shoot my blood sugar into the stratosphere, salads and veggies keep it at a manageable level, and light exercise, even a short walk after a meal, helps bring it down almost instantly.

This information was so revelatory that I began to wonder whether putting CGMs and the information they provide into the hands of a broader set of people could help us better manage societal health over time. After all, if it changed the way I eat, could it do the same for millions of others?

One way to explore that question is to talk about the technology with smart people. Last week at CES, I did just that when I moderated a session titled “From Brainwaves to Blood Sugar: How Next-Gen Tech Shapes Diets” during The Spoon’s Food Tech Conference at CES. I purposely programmed the session with a mix of panelists to bring medical, startup, investor, and researcher perspectives to the conversation.

After sharing my experience with a Stelo CGM on stage, I asked Howard Zisser about the importance of over-the-counter CGMs. Zisser, a physician and longtime pioneer in diabetes technology who worked on some of the earliest CGM systems in the early 2000s, when the technology was designed almost exclusively for people with diabetes, said the value of these newer CGMs lies in the data and what you can do with it.

“Instead of one or two readings a day, you suddenly have 300, 500, 600 readings a day,” Zisser said. “You start to see trends. What happens when you fast, when you exercise, during a menstrual cycle. IYou get a rich data set that’s your data personally.”

When I talked about how jarring my own experience was in seeing my blood sugar spike, Zisser argued that the shock is part of the value. Unlike biomarkers such as cortisol, which are difficult to influence in real time, glucose is actionable.

“You see it, and you can change your behavior,” he said. “Next time, you make a different choice.”

He likened glucose feedback to learning how to drive with a speedometer. Without it, he said, you’re guessing. With it, you can learn how your actions translate into outcomes.

But not everyone will wear a sensor on their arm. Noosheen Hashemi, founder and CEO of January AI, argued that while hardware CGMs are powerful, they are not scalable to the hundreds of millions of people with undiagnosed prediabetes or metabolic dysfunction. She said technology like that developed by her company leverages machine learning models trained on years of CGM data to predict glucose responses without requiring buying a hardware sensor.

“Our claim to fame is creating the world’s first continuous glucose monitor with AI,” Hashemi said, explaining that the system can generate directionally accurate predictions using inputs such as age, weight, activity level, sleep, and food intake.

But for all the data and actionable insight these tools can provide, they do not guarantee lasting change. Sherry Frey, VP of Total Wellness at NielsenIQ, shared research showing that even after receiving a diagnosis and initially adjusting their diets, behavior often reverts within months.

“We actually see about nine months in that a lot of behavior reverted,” Frey said. “When people were maybe less engaged and a little tired of of having to eat differently.”

That drop-off highlights both the opportunity and the challenge for health technology. Sustained engagement requires more than numbers on a screen. It requires context, interpretation, and motivation.

Frey also noted that adoption of wearables and health-tracking technologies is expanding beyond affluent early adopters. One of the fastest-growing user groups, according to NielsenIQ data, includes consumers on SNAP benefits, many of whom are using these tools for chronic disease management rather than fitness optimization.

“The addressable market is much larger than people with diabetes,” Frey said.

As we discussed what makes behavior change stick, I asked whether giving consumers more data, as the Nest thermostat did starting a decade ago, would prompt lasting change. Peter Bodenheimer, U.S. venture partner at PeakBridge VC, said yes, but only if the insights are actionable.

“Insights that tell you, ‘if I do this, then something good or bad happens,’ tend to be the things that people respond to and maintain.”

The panel also acknowledged the downside of constant feedback. More data can mean more confusion, anxiety, and misinformation. Hashemi shared an example of a user who believed their glucose should never rise above 110, a misunderstanding fueled by social media rather than clinical reality.

“Metabolic fitness is how you go from fasted to fed efficiently,” she said. “It’s a preposterous idea to keep your blood sugar the same all the time. So yes, there’s a lot of misinformation.”

Zisser reinforced that interpretation depends heavily on individual context, goals, and physiology. The same glucose spike can mean very different things for a professional athlete, a person with diabetes, or someone trying to lose weight.

We also discussed other technologies that can help us understand what’s happening inside our bodies, such as neural implants and other next-generation sensors. Hashemi pointed to implantable sensors capable of reading multiple analytes for years at a time, as well as emerging efforts to continuously measure substances like lactate, ketones, alcohol, and eventually insulin.

“Yeah, there’s definitely implantables,” said Hashemi. “There’s one that reads 20 different analytes, including glucose. It lives, you have to inject it under your skin. It can live 900 days. And it’s still in animals. It’s not in humans yet. But these things are coming.”

As the number of measurable signals grows, so do concerns about privacy, trust, and data ownership. Frey noted that while many consumers want their health data integrated in one place, roughly half remain uncomfortable with embedded sensors and worry about how their information might be used by insurers, governments, or corporations.

Others felt that the benefits of these technologies may ultimately outweigh more abstract fears. When people see tangible improvements in sleep, energy, or focus, trust can follow.

“No government, no doctor can make somebody healthy,” Hashemi said. “The only person that can do that is yourself.”

As we wound down the session, we talked about personalized nutrition, a topic that has long been a point of heated discussion in the world of food and health. The panelists agreed that while personalized nutrition may never be perfectly precise, the combination of biological data, AI, and human context is moving the industry closer to that goal.

“The gold lives in the combination of data,” Hashemi said, suggesting that consumer-generated health data will increasingly merge with clinical care, especially as value-based healthcare models expand.

In the end, the promise of next-generation health tech may be less about perfect prediction and more about empowerment. One idea that Zisser suggested was possibly getting these types of technologies into the hands of young students as we are teaching them how to eat.

“When my dad taught me how to drive, he didn’t put me in a car without a speedometer, right? It’s like, have feedback, I have information. And so to give people that access to that, and not that they would need it all the time, but so they can learn how their choices impacts their glucose.”

Not a bad idea. I can only imagine what my long-term health outlook might be different if I’d had insight into the impact of certain foods on blood sugar when I was much younger.

If you want to hear my conversation with these smart people, just click play below.

CES 2026: From Brainwaves to Blood Sugar: How Next-Gen Tech Shapes Diets

Hold The Humanoids: Why a Couple Robot Experts & a TV Chef Think The Humanoid Takeover of Food May Never Materialize

Ten years from now, CES 2026 may be remembered as the year robots took over the show floor. Humanoids folded clothing, boxed items, played games, and talked like product marketing managers.

Against that backdrop, I led a conversation on the food tech stage about whether robots may soon take over the kitchen. In a session titled “Robot vs. Chef: Will AI Augment or Replace the Cook?”, I brought together longtime TV chef Tyler Florence with a pair of robot builders: Nicole Maffeo of Gambit Robotics and Ali Kashani of Serve Robotics.

And when I say “pitted,” I mean I let everyone jump into a wide-ranging conversation about the future, one in which most participants were largely in agreement about how robots should be used in home and professional kitchens, though not always.

While tens of thousands of attendees were checking out robots on the show floor and seeing what they could theoretically do, I asked my panelists what robots should actually be doing. From the get-go, they rejected the idea that humans will be replaced by AI or robotics in the kitchen. Chef Tyler Florence framed AI not as a creative force, but as a responsive one, noting that its output is entirely dependent on human input.

“As great as AI is right now,” he said, “it’s really all about the prompts. It’s not going to do anything if it’s just sitting there by itself.”

Rather than replacing chefs, all the panelists agreed that AI and robotics are far better suited to working alongside them, handling the repetitive and unglamorous work that drains time and energy from kitchens.

But what about boring, dangerous, or repetitive tasks? Clearly, not all jobs are fulfilling or even ones that many humans want. And when people do those jobs, there is always the risk of injury.

According to Kashani, repetitive, injury-prone, and hard-to-staff tasks are already being automated.

“If you have that job, like coring an avocado, that’s not a great job,” he said. “It’s actually dangerous. People cut their fingers.” In those cases, Kashani argued, a robot can reduce injuries while freeing humans to focus on creative and guest-facing work.

This idea of using robots that are often focused on a single task and look nothing like a human stood in stark contrast to what we saw on the show floor, where humanoids seemed to be everywhere. When I asked the panelists whether a human-like form factor made sense, all agreed that we would not see humanoids in restaurants or home kitchens anytime soon.

“No one wants a man coming out of their closet to come and cook them dinner and then going back in,” said Kashani.

Maffeo agreed. “We don’t need someone coming out and doing all these things for us,” she said. “Just help us solve these simple pain points that waste so much of our time.”

Maffeo said she believes distributed, specialized robots are both cheaper and more practical than generalized humanoids, at least for the next decade.

Still, there is no doubt that robotics and AI are advancing quickly across the food system. So where does that leave someone like Tyler Florence, who has long made a name for crafting recipes and cooking for people in their own spaces without the help of technology? According to Florence, as robotics becomes more prevalent, the value equation flips, and people begin to crave food crafted entirely by humans. In other words, while machines can do many things well and cheaply, the scarce commodity becomes human judgment, taste, and presence.

“Human-made will become the new luxury item,” Florence said. “Things that feel like this is made by a human being, thought of by a human being, produced by a human being.”

In high-end dining especially, Florence predicted that automation would remain largely invisible, while human interaction becomes a premium experience people are willing to pay for.

But what about the home? Restaurant kitchens and front-of-house operations are businesses where people are accustomed to paying premiums for food prepared by others. The vast majority of meals, however, are eaten at home and made from food in our own pantries and refrigerators. What role will automation and AI play in the home of the future?

According to Kashani, we will increasingly see intelligence from technologies like computer vision, IoT, and automation integrated into everyday appliances to help people plan meals, reduce food waste, and prepare food more easily.

“Every step of that process, we can be assisting people with the help of AI and robots.”

Kashani also pointed to aging-in-place scenarios as an area where automation and AI could be especially helpful. Maffeo agreed and said she believes we will see more technology embedded in pantries and refrigerators to help people better plan meals.

As we closed out the panel, we talked about what the rise of robots and AI in food means for culture, jobs, and society over the long term. I was surprised that, by and large, everyone was cautiously optimistic. Kashani pointed to history as a guide, arguing that productivity gains tend to create new work rather than eliminate it outright. “Every such prediction in the past has been wrong,” he said, noting that employment has historically grown alongside technological change.

I disagreed to a point, suggesting that jobs will be lost, though this was not the place for a deeper conversation about universal basic income.

Florence raised a cultural concern, arguing that food is memory and identity, something passed down through families and communities. “We’re all defined by what our grandparents cooked,” he said. “And that really defines us as people.”

It was a fun and thoughtful conversation, one that explored the implications of what might happen if what we saw on the show floor ultimately becomes the norm. You can watch the full session below.:

CES 2026: Robot vs Chef: Will AI Augment or Replace the Cook?

Why the Most Interesting Knife at CES Launched Without Its Inventor

This week at CES, a new ultrasonic chef’s knife picked up write-ups in The Verge, Mashable, and a handful of other outlets after debuting at Unveiled, the opening press event for the big show in Las Vegas. With all the coverage rolling in, the product’s inventor, Scott Heimendinger, could feel confident that everything was going according to plan after six years of work to bring the knife to market, with one small exception.

He wasn’t there.

Of course, Heimendinger had always planned to be at CES. A presence at Unveiled was a core part of his launch strategy, a plan that crystallized over the long six years it took to bring the product from idea to reality. But life intervened in the form of excruciating pain caused by cervical radiculopathy, a condition in which nerves are impinged by discs and bone growth in the neck. The pain became so acute that when Heimendinger was offered the chance to move his surgery up by two months last December, he took it.

Not that the decision came easily.

Last fall, Heimendinger was on a call with his longtime friend, Rand Fishkin, who was not pleased with how he was handling things.

“I was laid up in bed, and all I could do was take out a laptop, totally just drowned in high-dose pain and nerve meds and stuff, and Rand and I had a little video chat,” Heimendinger told me over Zoom yesterday from Seattle. “And (Rand) basically threatened to speak at my early funeral if I didn’t take better care of myself. Like, actually focus on my health.”

Slowing down didn’t come naturally. After all, you don’t nearly single-handedly launch a new consumer hardware product without being wired to push through discomfort.

“That’s a hard thing for me to do,” Heimendinger said. “I’ve kind of been in power-through mode forever, right? Like my whole life, it’s just like, ‘Oh, what do you do? You power through.’”

Eventually, Heimendinger relented, knowing his friend was right. From there, he began making plans for his small team – a single marketing lead and a part-time PR representative – to handle booth duty at Unveiled without him. He was bummed. CES would be the first time many members of the press would get hands-on with the knife he’d unveiled online in the fall, and he knew how easily a small team could get overwhelmed by the roughly 2,000 journalists cycling through Unveiled during its three-hour run.

When Heimendinger told Fishkin how disappointed he was to miss CES and how much the moment meant to him, Fishkin made an unexpected offer: he and his wife would go in his place.

“And, you know, normally I would just say, like, ‘Oh, that’s so nice of you guys, thanks so much, but no, it’ll be fine,’” Heimendinger said. “But I said, I’m going to try something new and try accepting a little more help when it’s offered. And I said, ‘Actually, if you’re serious, that would be incredible.’”

It made sense. As Heimendinger’s first investor and sole board member of his company, Fishkin was deeply familiar with the product and its backstory. He’s also a seasoned marketer known for his viral videos explaining technology and business trends, while his wife, Geraldine DeRuiter, is a professional author with a strong communications background.

“So they’re well-versed in how to talk about the knife and can do so authentically,” Heimendinger said. “And so I said yes and accepted their help, and they were serious and made good on it.”

In the end, the knife didn’t need its inventor physically behind the table to make an impression. Journalists lined up to try it, coverage followed quickly, and the resulting long-tail coverage Heimendinger had hoped for came off as planned.

For those interested in the knife itself, it uses high-frequency vibration, oscillating 40,000 times per second, to reduce resistance as the blade moves through food. Heimendinger says that the knife can reduce cutting effort by up to 50 percent. When powered off, it functions like a traditional, high-quality chef’s knife.

The C-200 is made with Japanese AUS-10 san mai stainless steel, can be re-sharpened like a conventional blade, and is now available for presale at $499, with deliveries expected in January 2026.

For a product six years in the making, CES didn’t unfold exactly as Heimendinger imagined. But sometimes, even for someone who’s spent a lifetime powering through, the most important step forward is learning when to let someone else take the wheel — or, in this case, the knife.

Shinkei Hopes Bringing Robotics & AI to the Fishing Boat Leads to Fresher Fish and Less Waste

Walk into almost any grocery store and, chances are, what you see in the fish case is not at peak freshness. 

You wouldn’t think it’d be that way, especially in places like Seattle, where I live, because such care is given to getting the fish to market quickly. But according to a new venture-backed startup named Shinkei, the critical factor in determining freshness is not what happens after fish leave the boat, but instead what happens in the moments immediately after they are caught. 

I caught up with Shinkei CEO Saif Khawaja earlier this month to discuss how exactly his company’s tech brings what he claims are Michelin-quality fish at commodity costs. According to Khawaja, conventional handling on the boat fails because it leaves most fish “flopping around,” which triggers stress responses that accelerate quality loss and shorten shelf life.

“Most fish available at a mass market retailer were handled on the boat in a way that releases stress hormone, lactic acid,” said Khawaja. “This stuff makes the meat more acidic, primes bacteria growth, and in turn speeds the shelf life and decay of meat quality.” 

Shinkei’s computer-vision-powered robot is designed to intervene immediately. Fish are placed into a machine, which the company calls Poseidon, while still alive, and it uses computer vision AI to scan the fish to determine the fastest (and least-stressful) path forward for the fish. Once the fish is scanned, the machine performs a fast sequence: a brain spike to euthanize as quickly as possible and a gill cut (to drain blood).

If this system seems, well, rough, it is, but the reality is fish caught are experiencing high stress from the time they’re caught, and the faster the fisherman can move towards euthanization, the more humane (and ultimately fresher and better tasting the fish). Khawaja says each fish is processed in about six seconds, and the company’s goal is to get the fish into the system quickly after landing, ideally within roughly a minute, before quality begins to degrade meaningfully.

Speed on the Boat = Less Waste in the Store

While much of the pitch is focused on better taste, Shinkei’s technology also a food waste angle. According to Khawaja, their solution also helps reduce waste in the store. That’s because, typically, a suffocated fish might enter rigor mortis in about 7 hours, but Shinkei’s process expands that up to 60 hours, which creates a much larger buffer before decomposition starts. Khawaja says it also makes a difference by species, with black cod handled in a traditional way lasting four to five days, where Shinkei-handled fish can stay fresh or up to two weeks. 

Khawaja attributes the compounding effect to two factors: reducing stress (less acidification) and removing blood that would otherwise diffuse through the meat and feed bacterial growth. He says the resulting shelf-life extension gives food distributors more options for logistics, allowing fish to be trucked rather than flown. 

If Shinkei’s technology works as promised, one might expect to see all professional fishermen and processors installing hardware at some point, right? Maybe…not. That’s because the company’s business model is to create a branded direct-to-consumer model for its fish, so instead of selling the hardware outright, Shinkei places machines on partner boats under a zero-cost lease and retains ownership of the machines. They also require an exclusive buying structure that grants Shinkei the right to purchase the catch processed that uses its machinery. 

From there, the company sells the fish into foodservice channels and retail under the brand Seremony, where they’re trying to get “Seremony grade” to catch on.  Khawaja says the company has sold into top-tier restaurants globally, including Michelin 3-star destinations across multiple countries, and recently launched in Wegmans (Manhattan) and FreshDirect (New York).

Today, Khawaja says Shinkei works with eight boats, sourcing species like black cod, rockfish (including vermilion rockfish), and red snapper, plus some ad hoc species (salmon, black sea bass, and others). The boats surf water in the US west coast (Alaska down to California), Texas, and Massachusetts.”

When I asked Khawaja about the underlying technology, he told me they built their AI models in-house, collecting their own data and building a pipeline informed by work like facial recognition research (fish face, that is). The computer vision stack performs a set of inferences: identifying species, detecting key points, and generating cutting paths.

He also talked about two new projects they are working on within the platform. One is Kronos, a weight-estimation model embedded in the machine that sends catch data back to the Shinkei sales team in real time so they can start selling fish before it reaches the dock. Another is Nira, which uses sensors to predict shelf life.

“We integrate sensor data into a model, and we will be able to generate ground truth at any point in the supply chain for what shelf life and quality is for that fish,” said Khawaja. 

The company recently raised $22 million and is currently at Series A. The Series A was co-led by Founders Fund and Interlagos, with new investments from Yamato Holdings, Shrug, CIV, Jaws, and Mantis.

Long-term, I wondered whether the company was open to expanding to a model in which it sells its hardware to fishermen who don’t feed their catches back to the company as part of the Seremoni pipeline.  Khawaja and Shinkei completely shut the door, but for now, they’re “focused on building the brand and basically establishing and making ceremony-grade as a certification.”

Can AI Help Chocolate Survive? NotCo and Swiss Chocolate Maker Barry Callebaut Think So

Is the world of chocolate heading toward the same fate as the dodo bird?

It may be a surprise to some to consider that a centuries-old, worldwide favorite like chocolate is on its way out, but the reality is that most experts agree that climate change, ingredient shortages, rising prices, and other global pressures have put chocolate in jeopardy. Some even predict that it could one day become extinct.

The impending peril has meant that every global chocolate supplier has started looking for ways to adapt to the future, and increasingly, one of those ways is to do what many companies both in and outside of food are doing: look to AI to accelerate change. Some, like Hershey’s, have developed their own tools such as Atlas, while others are looking to companies with deep experience building AI models focused on food to help transform their business.

In that second category is a new partnership between Barry Callebaut, one of the world’s largest premium chocolate makers, and NotCo, the AI-powered food company that has made a name for itself in recent years with its Giuseppe food AI platform. The deal calls for NotCo AI to embed what it describes as its foundational AI platform directly into Barry Callebaut’s R&D pipeline. The announcement says the collaboration gives Barry Callebaut access to the same engine that has helped NotCo accelerate formulation cycles, solve complex ingredient challenges, and unlock unexpected flavor and functionality breakthroughs for global CPG brands.

For NotCo, the deal marks its most significant category-wide integration yet and reinforces what CEO Matias Muchnick said he and his cofounders believed from the very beginning. NotCo is not simply a maker of plant-based food. It is a next-generation R&D operating system for the food industry.

“This is exactly what we built NotCo for,” Muchnick said at SKS 2025 in July. “The value of our platform comes from a decade of high-fidelity data, from formulations and ingredient chemistry to sensory outputs and manufacturing parameters, all connected so we can solve multi-dimensional problems faster and with no human bias.”

The two companies plan to feed Barry Callebaut’s 100-year knowledge base and ingredient data into NotCo’s AI foundation model and build what they are calling the chocolate industry’s first end-to-end AI innovation hub. The goal is to iterate on new formulations, explore functional ingredients, and optimize for sustainability, cost, and Nutri-Score constraints.

At Future Food Tech in the spring, Muchnick gave a presentation that emphasized their push to become the go-to partner for AI transformation. The company’s Kraft Heinz partnership had already given them some street cred, so it is no surprise that they have seen strong interest from global food brands.

“Every big food company is having board-level conversations: do we have the technology to adapt to new consumers, shortages, and regulations? And consistently the answer is no,” Muchnick said at SKS in July. “That is why they’re coming to us. Everything changed in the last six months.”

In a sense, food AI specialists like NotCo are in a race against time as bigger general-purpose foundation models from OpenAI, Anthropic, and others become easier for different industries to customize with their proprietary data. Most CPGs do not yet have it in their DNA to build AI-forward development cycles, but that is likely to change in the next five years as boards demand transformation while they watch competitors accelerate product development and, in categories like chocolate, identify new alternatives in a market where environmental pressure, inflation, cost volatility, and other external factors force their hand.

“The companies that don’t adopt AI the right way will get the Blockbuster effect,” said Muchnick. “They’ll become obsolete. The future food companies will be AI companies.”

Remilk Launches Recombinant Protein Powered Milk in Israel, Eyes US Launch in 2026.

After spending half a decade (and $150 million in funding) developing recombinant proteins made via precision fermentation, Remilk announced last week the launch of what it’s calling New Milk in partnership with Gad Dairies, one of Israel’s largest dairy distributors. Remilk announced that under the new partnership, they and Gad will begin a nationwide rollout of their new product in their home country of Israel.

Real Milk, which will contain 75% less sugar than regular milk, will be manufactured in Spain via a contract manufacturer. The company indicated it will also distribute its milk to cafes and coffee shops and will follow with retailers later this year. The company is eyeing a US market launch in 2026.

To learn more about ReMiIk’s new product and its US market plans, I decided to catch up with the company’s CEO, Avi Wolff. Our conversation ranged from the technical challenges of rebuilding milk from a single protein to what he learned while working with big CPGs and how the company approaches scaling without building its own mega-plants. The answers have been lightly edited for clarity.

Let’s start with the basics. What is The New Milk, and how are you making it?

“We’re using yeast and precision fermentation to produce beta-lactoglobulin, the primary whey protein in milk. And we’re also responsible for the formulation and the manufacturing of the finished product, while Gad will be handling distribution.

“So we are producing milk in bottles at a commercial scale, and this is the product that we’re launching. We’ve already distributed thousands of gallons of milk to consumers, to cafés and restaurants.”

Remilk started as a B2B ingredients company. Why step into finished products and a consumer-facing brand?

“We started as a B2B company selling the ingredients to CPGs. We piloted some products with General Mills three years ago, and what we realized is that a whole lot of innovation and skill sets are required to turn this subset of protein into a delicious dairy product. So we’ve stepped in and we’ve decided to take ownership of the entire value chain from the production of the protein to the production of the finished product. We’ve spent almost six years after raising $150 million in developing the formulation, and this is why we are now ready to launch.”

Can you tell us more about your joint venture with Gad Dairies?. How does that partnership actually work?

“We knew establishing a pure B2C business model is complex and would take a lot of time. We also recognize that this is not within our area of expertise. In simple terms, we believe that distribution and sales are more of a commodity, allowing us to utilize other partners.

“So we decided to partner and establish a new company. We’ve built a new venture with one of the largest dairy companies in Israel. This new company is co-owned by Remilk and Gad and each party is responsible for different parts of the value chain. Distribution, sales and all the logistics around it is taken care of by Gad and everything else is managed by us.”

“We manufacture the finished product, they take it, they store it, they load it on the trucks, and they’re responsible for selling it and driving it to the point of sales, whether it’s retail or foodservice. And obviously, each party has their own KPIs and objectives.”

Is the reason you are looking at North America next because of Europe’s anti-GMO stance?

“Yes, that’s the primary reason.”

How is your product different than Perfect Day’s or other precision fermentation-derived dairy?

“I think it’s in the small breakthroughs and how you take this protein and make it more functional, and how you create a specific flavor profile. If you mix protein with water and fats, it doesn’t taste like milk, and it will not work in coffee like milk. So you need to enhance functionality, and you need to enhance flavoring and aroma and everything. And this is something that we’ve been doing, I would say, perfectly.”

You mentioned you had worked with big CPGs such as General Mills and you needed to take more control over the process. What did you mean by that?

“We worked with them, helping and supporting them in their development process. But the development process, despite the fact that it was sometimes even two years with ongoing developments, we felt that the product was not good enough.”

“And I think that there are many explanations to this, but I think the primary one is the fact that the skill sets that these CPGs have in their existing R&D centers are just completely different from the ones you need different skill sets than what you need to develop dairy products from a single subset of protein.”

“We have 100 employees in Remilk, 70% of them are scientists, PhDs in biology, chemistry, and they’re developing the products. We’re not just traditional food scientists who used to work at the large dairy companies, because we’re conducting deep research on the protein, on the flavoring, on the milk itself. We analyze the milk in order to understand how to mimic the specific smell and taste.”

What is your timeline for a US launch?

“Definitely for the next six to nine months, our focus is going to be succeeding in Israel. Again, we believe that given the nature of this market, we can reach strong momentum relatively quickly.

“While most of the company’s operations will remain in Israel and focus on the Israeli launch, a few key employees and I will begin negotiating a similar deal with a U.S. partnership. A US partnership is definitely something we’re aiming for in 2026.”

We Talked With Nectar About Their Plans to Build an AI for Better Tasting Alt Proteins

A few weeks ago, the philanthropic investment platform Food System Innovations announced that it had received a $2 million grant from the Bezos Earth Fund. FSI’s non-profit group NECTAR has been building a large dataset of consumers’ sensory responses to alt proteins, and the grant will help NECTAR to continue working on, in partnership with Stanford University, an AI model “that connects molecular structure, flavor, texture, and consumer preference.” The goal, according to NECTAR, is to create an open-source tool for CPGs and other food industry players to develop more flavorful—and hopefully better-selling—sustainable proteins.

I’d been following NECTAR for some time and have been closely tracking the impact of AI on food systems, so I thought it would be a good time to connect with NECTAR. I’d talked about the project briefly with Adam Yee, the chief food scientist who helped with the project, while I was in Japan, and this week I caught up with NECTAR managing director Caroline Cotto to get the full download on the project and where it’s all going.

Below is my interview with Caroline.

What are you building with this new Bezos Earth Fund grant?

“One of the things Nectar is doing is we just won a $2 million grant from the Bezos Earth Fund to take our sensory data and build a foundation model that will predict sensory. So we kind of bypass the need for doing these very expensive consumer panels, and then also predict market success from formulation. It’s intended to be sort of a food scientist’s best friend in terms of new product ideation.”

For people who don’t know Nectar, what’s the core mission, and how did this AI project start?

“Basically, Nectar is trying to amass the largest public data set on how sustainable protein products taste to omnivores. That’s what we have set out to do. We’re building that, and we are working heavily with academics to operationalize that data.

Over a year and a half ago, we started talking to the computer science folks at Stanford to say, like, what are things we could do with this novel data set that we’re creating? It happened to be around that time that the phase one Bezos Earth grant was opening up for their AI grand challenge. I connected Adam with the Stanford team, and they did some initial work on LLMs and found that it was able to do some of this support for food scientists. They published a paper together that came out in January for ICML, the largest machine learning conference, and we ended up winning that phase one grant, which then allowed us to apply for the phase two grant that we just found out about in October.”

From a technical standpoint, what kind of AI are you actually building?

“I am not an AI scientist myself here, so we are heavily partnered with Stanford and their computer science team, but it is an LLM base. We’re basically fine-tuning an LLM to be able to do this sensory prediction work, and it’s a multi-modal approach. There’s a similar project that’s been done out of Google DeepMind called Osmo for smell and olfactory, and we’re working with some of the folks that worked on that in order to model taste and sensory more broadly, and then connect that to sales outcomes.”

How does the Bezos Earth Fund AI Grand Challenge work in terms of phases and funding?

“It’s the Bezos Earth Fund AI Grand Challenge for Climate and Nature. It’s $30 million going to these projects. There were 15 phase two winners that each received $2 million and have to deliver over two years.

The phase one was a $50,000 grant to basically work on your idea and prepare a submission for phase two. We spent about six months preparing, trying to connect this Nectar data set with sales data and see which sensory attributes are most predictive of sales success, and also connecting the Nectar sensory data set to molecular-level ingredient data sets. Ideally the chain of prediction would be: can you predict sensory outcome from just putting in an ingredient list, and if so, what about sensory is predictive of sales success? We’re working on the different pieces of that predictive chain.”

What does your sensory testing process look like in practice?

“It’s all in-person blind taste testing. In our most recent study, we tested 122 plant-based meat alternatives across 14 categories. Each product was tried by a minimum of 100 consumers. They come to a restaurant where we’ve closed down the restaurant for the day, but we want to give them that more authentic experience. They try probably six products in a sitting, one at a time, and everything is blind, so they don’t know if they’re eating a plant-based product or an animal-based product and then they fill out a survey as they’re trying the product.”

How big is the data set now, and what’s coming next?

“We do an annual survey called the Taste of the Industry. For 2024, we tested about 45 plant-based meat products. For 2025, we tested 122 plant-based meat products. Outside of that, we have our emerging sector research, which are smaller reports. We’ve done two of those, and both have been on this category we’re calling balanced protein or hybrid products that combine meat. We’ve tested just under 50 products total in that category as well.

We’re testing blends of things like meat plus plant-based meat, meat plus mushrooms, meat plus microprotein, meat plus just savory vegetables in general. For 2026, our Taste of the Industry report is on dairy alternatives. We’re testing 100 dairy alternatives across 10 categories, and that will come out in March.”

When you overlap taste scores with sales data, what have you seen so far?

“The Nectar data set is mostly just focused on sensory. That’s the core of what we do. We are also interested in answering the question ‘do better-tasting products sell more?’ In our last report, we conducted an initial analysis of overlapping sensory data with sales data, finding that better-tasting categories capture a greater market share than worse-tasting categories. Better-tasting products are capturing greater market share than worse-tasting products. In certain categories, that seems to be agnostic of price. Even though the product might be more expensive, if it tastes better, it is capturing a greater market share.

We’re currently working with some data providers to get more granular on this sales data connection, because that analysis was from publicly available sales data. In this AI project, we are trying to connect sensory performance with sales more robustly to see which aspects of sensory are predictive of sales success. It’s hard because there are a ton of confounding variables; we have to figure out how to control for marketing spend, store placement, placement on shelf, that sort of thing. But we have access to the Nielsen consumer panel, this huge data set of grocery store transactions over many years, from households that have agreed to have all of their transactions tracked. We’re able to see what consumers are purchasing over time, and we’re trying to connect the sensory cassette to that.”

You also mentioned bringing ingredient lists and molecular data into the model. How does that fit in?

“We’re trying to say, there are a lot of black boxes in food product development because flavors are a black box. We don’t have a lot of visibility into companies’ actual formulations. We’re trying to determine if we can extract publicly available information from the ingredient list and identify the molecular-level components of those ingredients, and then determine if any correlations can be drawn between them.

It’s all of these factors plus images of the products and trying to see if we can predict that.”

What do you actually hope to deliver at the end of the two-year grant?

“The idea is to deliver an open source tool for the industry to use. The goal would be that you can put in all the constraints you have for sustainability, cost, nutrition, and demographic need, and that it would help you get to an endpoint where you don’t have to do a bunch of bench-top trials and then expensive sensory.”

How do you think about open source, data privacy, and companies actually using this tool?

“Data privacy is a big thing in this space. We don’t have any interest in companies sharing their proprietary formulations with us. The goal is that they would be able to utilize this tool, download it to their personal servers, and put in their private information and use it to make better products. If we’re rapidly increasing the speed at which these products come to market and they are actually successful, that would be a success for us.

There are other efforts in this space, from NotCo to IFT. Where does Nectar fit?

“I think everybody is trying to do similar things, but with slightly different inputs and different approaches. We are open to collaborating and learning from people. Our end goal is a mission-driven approach here, not to make a ton of money, so it depends on whether or not those partners are aligned with that goal.

IFT has trained its model on all of the IFT papers that have been published over the many years of its organization being around. We’re training our model on our proprietary dataset around sensory data, so there’s some nuance between things. They’re really focused on developing formulations, but there is a limitation to what you can do with that tool. It’ll tell you, ‘here’s how to make a plant-based bacon, add bacon flavoring,’ but there are 10 huge suppliers that provide bacon flavoring, and it doesn’t provide a ton of granularity on at what concentration and from what supplier.”

What’s the bigger climate mission you’re trying to advance with this work?

“Nectar’s specific directive is, how do we make these products favorable and delicious? We know that we need to reduce meat consumption in order to stay within the two degrees of climate warming, and we’re not going to get there by just telling people, ‘eat less steak.’ We have to use that whole lever and make the products really delicious so that people will be incentivized to buy them more and reduce consumption of factory-farmed meat.”

Answers have been lightly edited for grammar and clarity.

MongoDB Founder Eliot Horowitz on Building a WordPress for Robotics (And Why He’s Skeptical of Humanoids in the Kitchen)

Nearly two decades ago, Eliot Horowitz launched a company with a couple of cofounders that would be pivotal in the early days of big data, helping to create more scalable solutions, lower costs and also reducing the need for deep technical expertise needed in setting up big and messy databases. The company, which would eventually become MongoDB, also gave greater power to the developer community who leveraged the open source model to help iterate and make the product better.

Today, Horowitz hopes to apply many of the ideas that helped make him successful in big data to the world of robotics. With newest company, Viam, he is building a platform he believes will enable software engineers to more easily create and iterate software stacks for robotics and automation systems, a space where the tools to build great software are, in Horowitz’s words, “not great”.

I had heard about Viam because the company’s technology is underlies that of Gambit Robotics, a company building a vision guidance system for the kitchen. In fact, Gambit is also led by Horowitz (as well as former Google AI lead Nicole Maffeo), as a sort of a startup within a startup. In fact, Horowitz and Viam are building other vertically focused solutions on top of Viam for commercial kitchens, boat sanding, fishing and more.

I decided to catch up with Horowitz to talk to him about his move into robotics and how he sees the world of robotics development changing over the next decade. You can read an abbreviated interview below and listen to the entire conversation on The Spoon Podcast. Answers below have edited slightly for clarity.

Why did you decide to get into robotics?

And towards the end of 2020, I was frustrated with how hard it was to bring robotics projects and other hardware projects to fruition and how to make them real. And if you talk to other software engineers, their experiences were not great.

One of the topics you’ve discussed is the critical technology gap between software and the physical world of engineering. How does your new company address that?

What we really saw in this space was that a lot of interesting research had been done over the last 30 years in hardware. Batteries, motors, and robot arms have all gotten much better.

At the same time, the way people build software applications now is completely different from what it was 20 years ago. And a lot of what we saw in the hardware space is things that hadn’t been changed in a while. And it’s different, because with a robotics project, it’s both a big software project and a big hardware project. And when you have tools in the software space that don’t mesh with the hardware tools, and don’t mesh with the robotics and automation side of it, everything gets messy and just doesn’t work.

And so we said, ‘Great, what we’re going to do is we’re going to build a platform that lets engineers actually bring these projects to life.’

In the early days of the Internet, creating a website was not very easy until platforms like WordPress came along. Are you essentially making a WordPress for robotics?

It’s a really great analogy. We joke that if you dropped AWS into the hands of a Perl developer in 1998, they wouldn’t know what to do with it. We’re trying to make that same leapfrog happen in robotics. Wwe want to go from where building a robotics project is an enormous undertaking, where you have hundreds of engineers, to where you can actually build a legitimate production hardware project with real robotics and AI with an actual startup-sized budget.

What did you learn from building one of the world’s most popular database platforms in MongoDB that you are applying to your new startup?

What made Mongo great was we built an incredibly great community that we listened to almost to a fault for a long time around making sure that they were able to build exactly what they wanted and get things done

There are not enough engineers building in this space. And if we can enable more and more engineers to build in this space and really build a platform, lots of people will work together and build great things, and that’s when things get really interesting. It’s our job to enable them to do it and to show them what’s possible so that you can get a million people coming up with really cool ideas and actually bringing them to market. That’s what made Mongo great, and that’s the same thing we want to do here.

Oftentimes the time and the capital needed to bring a robot to market is just years and years and hundreds of millions of dollars. Do you believe you will shorten those timelines and reduce the capital costs by significant degree?

By orders of magnitude. What I see a lot in the robotic startups that haven’t worked is their inability to get to a somewhat interesting proof of concept (POC) that happens pretty quickly. All the engineering you have to do to make it real ends up taking years and years and years. The other big thing we see is the sheer amount of iteration time. Going from one version to the next version takes way too long.

When you can actually iterate fast, when you get a new version, and then another version, when you can test really quickly, or even iterate with users live, that’s when magic happens. And how do you bring that to the robotic space? That’s what we’re really focused on is, how do you bring that iteration time down? How do you get the proof of concept to production time dramatically down?

There are approximately 30 million software engineers worldwide, and most of them are intimidated by working with hardware. Hardware engineers typically have not had great experiences working with software engineers. It just hasn’t gone great over the last 30 years. We’re trying to completely bridge that gap so you can get hardware engineers and software engineers working together in a much more tightly knit sort of cycle and that changes everything.

What do you think about the work on humanoid robotics for the kitchen?

I’m skeptical of that. But a robot arm embedded into the back wall of my kitchen that could manage everything on the stove for me, that could stir things and add some ingredients at the right time or lower the temperature on a burner, that would be really useful. That doesn’t require any breakthroughs. We have the technology and we have the software that we could do today. So there’s a lot of middle steps (to a humanoid) that I think are very practical, very interesting, and can actually move the needle forward very quickly.

You can listen to the full interview by clicking play below or on Apple Podcasts or Spotify.

Cultivated Meat Turbulence Leads to IP Churn Through Deal Making and Open Source Initiatives

Cultivated meat companies spent much of the last decade promising to help fend off the climate crisis while also helping to wean the world off animal agriculture. However, as the industry transitioned from bench scale to pilot facilities and eventually to scaled manufacturing, costs increased and timelines lengthened. This shift happened just as the venture capital world began to pull back on big bets in new areas outside of AI.

The result has been a monumental struggle for cultivated meat startups. Major players, such as Upside and Eat Just, scaled back plans to build large-scale manufacturing plants, while several companies shut down or were acquired.

One interesting wrinkle in this corrective period has been the movement of intellectual property in the form of patents, cell lines, and technical knowledge. As startups look for new paths through acquisition, merger, or wind-down, significant cultivated meat IP is changing hands. In one case, key assets have even been open-sourced. The acquisition activity around IP began in earnest last year, although early signs appeared before that.

Last week, Fork & Good announced it had acquired Orbillion, combining two of the more sophisticated platforms in cultivated pork and beef. Fork & Good has been working on cultivated pork since 2018, and Orbillion, founded in 2020, brings cultivated wagyu beef technology into the fold. The deal creates what the companies say is the largest IP portfolio in cultivated red meat.

“We’re not asking food manufacturers to wait five to ten years for supply chain solutions,” Fork & Good CEO Niya Gupta said in the announcement. “We’re giving them the ability to improve their products right now.”

Orbillion’s CEO Patricia Bubner, now COO of the combined entity, framed the deal as strategic consolidation aligned with a more pragmatic era that is margin-focused, customer-driven, and centered on technical execution rather than R&D alone.

A few months earlier, Meatable acquired Uncommon Bio’s cultivated meat platform, bringing over key technology, cell lines, IP assets, and technical staff as Uncommon shifted toward therapeutics. The acquisition strengthened Meatable’s multi-species lineup and its non-GMO platform, further concentrating Europe’s cultivated meat expertise under fewer roofs.

And just weeks before Fork & Good’s move, Gourmey merged with Vital Meat to form PARIMA. The deal brings “Gourmey’s full-stack industrial platform, which includes premium cultivated duck products validated by Michelin-starred chefs and independently verified production costs below €7/kg, with Vital Meat’s poultry cell-line technology developed from nearly 25 years of avian cell research at Groupe Grimaud, a global reference in animal genetics and biotechnology.”

The merged Paris-based entity unites Gourmey’s cultivated duck and foie gras technology with Vital Meat’s chicken platform, consolidating more than 70 patent filings and regulatory dossiers into a single operation targeting the European market.

Taken together, these deals signal a decisive shift: fewer players, deeper portfolios, and stronger technical moats. The companies that survive are those with enough IP, regulatory traction, and cross-species optionality to prove viable unit economics before pursuing scale.

And Then There’s Open Access

While consolidation was expected, another move announced in October was surprising, both in timing and format.

In mid-October, the Good Food Institute announced it had acquired bovine cell lines and serum-free media formulations from shuttered startup SCiFi Foods and partnered with Tufts University to release them for open research access. The move effectively open-sourced core cultivated beef IP, saving future startups and researchers years of development and millions of dollars.

“By making these cell lines and media broadly accessible to the cultivated meat ecosystem, researchers and companies have a new starting line – one that’s now closer to the finish line of bringing new products to market,” said GFI’s VP of science Amanda Hildebrand. “SCiFi’s pioneering work is like a baton in a relay. Given our role in the field, GFI was able to ensure that baton didn’t drop, and through our partnership with Tufts, copies of that same baton will be handed off to scientists and startups around the world, enabling more people to join the race.”

Joshua March, SCiFi’s co-founder, put it more bluntly: “It took us four years and tens of millions to develop these cells. Now future startups will be able to leapfrog us.”

I sat next to March in the spring of 2024, while in San Francisco, during a tasting of SCiFi’s cultivated meat. At the time, he gave no indication of the company’s financial struggles, but just a couple of months later, SCiFi shut down. Credit to Joshua and the team for working with GFI to make this technology available, potentially enabling breakthroughs for researchers who can use the SCiFi cell lines and media formulations as a jump start.

Cultivated meat still has a long road ahead. Some states have taken an antagonistic stance despite USDA approval for three (now four) companies to sell product in the United States. Investors remain cautious due to long scaling cycles and the challenge of convincing consumers that cultivated meat can be both tasty and healthy. Still, the industry is taking necessary and sometimes painful steps to prepare for the next stage. Combined with promising advances in manufacturing technologies, such as those from Prolific Machines, there is reason to believe the final chapter of the cultivated meat story has yet to be written.

Behind The Scenes at Good Housekeeping With Nicole Papantoniou

Last month I was in New York City, so I decided to drop and visit the Good Housekeeping Institute. I went to visit Nicole Papantoniou, the director of the Kitchen Appliances Lab at Good Housekeeping, who had promised to give me a tour of the place.

If you haven’t visited Good Housekeeping Institute, it’s great because – aside from having one of the best possible views of Midtown Manhattan perched from its location on the 29th floor of the Hearst Building – it’s a cool hybrid of a newsroom meets testing lab, with appliances like air fryers, espresso makers, induction tops and ice cream makers piled high on surfaces everywhere.

“We’re using these items basically the way someone would use them in their home,” Nicole told me on the most recent episode of The Spoon Podcast. “Being able to compare things side by side and then understand the ease of use features, we really get a good understanding of how the product works.”

If you grew up hearing about the Good Housekeeping Seal like me, there’s a reason: for the past century, the publication and the institute helped pioneer consumer product testing. From the time Hearst bought in 1911 until the 1960s, it became a household name, and over the next half century, hitting 5 million in circulation by the 60s.

“In the early 1900s is basically when products were coming to the market and the team members were like, there’s no one really regulating it,” said Nicole. “So really trying to explain to consumers what they should be buying, what they can trust. And that’s what the good housekeeping seal is.”

In a way, being in the Good Housekeeping Lab felt like going back in time. From the different dedicated testing area for appliances, fabrics, and other household items to a full-fledged test kitchen, it was such a big departure from the current way in which most products reviews get generated in 2025, where influencers often will try something out or just see it online and give a review of the product.

According to Nicole, the reviews are around a seasonal rolling calendar which mirrors consumer behavior. “We work three months in advance on print and digital,” she said. “Think of summer… people are going to be searching for ice cream makers. And then think of also Q4, Black Friday, the holidays.”

Some categories, like air fryers, never sleep, while others resurge and periodically come back (stand mixers and bread makers). They also spin up new sub-categories as products evolve.

“We had our espresso maker story forever,” said Nicole. “But now there’s a lot of all-in-one espresso machines. You press your button, you get your cappuccino like you would in a office.”

On the podcast, I asked Nicole how she ended up with such a cool gig. According to her, she had gone to journalism school and knew she wanted to work in magazines, but the inspiration to fuse food and journalism all started with an internship.

“My first internship was at Ladies Home Journal. And I remember going into the test kitchen and someone was grilling pineapple and like candying walnuts. And I was like, how do I get that job?”

From there, she went to culinary school at night while working full time, then moved into brand-side roles. “I ended up at Cuisinart, developing products with them and recipes and helping edit user guides, and then ended up Family Circl and then here at Good Housekeeping.”

She told me that brand experience shaped how she evaluates products today. “When you’re working at a brand, you’re working with so many different departments. An engineer will come up with something really exciting and then you kind of have to hone it in and let them know like, this might not work in like a real consumer’s kitchen.”

I asked her if she had any advice for those looking to get into a similar line of work.

“I think, honestly, getting as much experience as you can with people who are in the field, saying yes to things, taking on different experiences,” she said. “At one point I was working for, like, four different jobs at once… but I loved it. Be nice, and say yes and then you’ll find what you’re looking for. Also don’t be afraid to walk away. There’s a lot out there.”

If you want to listen to my full conversation wtih Nicole you can click play below or find it on Apple Podcasts and Spotify.

Are Big Food Companies Really Embracing AI?

While some companies like NotCo have positioned themselves as the OpenAI of the food world, the truth is that the AI transformation of the food industry is still in the first inning. That is partly because the food system itself, a mix of legacy CPG giants, agricultural suppliers, ingredient developers, and regulators, moves at a glacial pace.

In my recent conversation with Jasmin Hume, founder and CEO of Shiru, she confirmed that the industry is still in the early stages, in large part because food companies have massive amounts of data and strong confidence in their own research and development.

“Food companies have world-class R&D teams, and those scientists want to see proof before adopting a new tool. It’s a lot of tire-kicking in the first meetings,” said Hume.

This slow pace does not mean AI is not making inroads. It is simply happening beneath the surface. From discovery platforms like Shiru’s to optimization tools in manufacturing and retail analytics, AI is slowly reshaping how food gets made.

But the true question is not if the food system will use AI, but who will own the models that make it useful. The answer is usually tied to who owns the data. Legacy food and ingredient companies have decades, even centuries, of proprietary chemical, biological, and sensory data. This makes them both powerful and hesitant to engage with AI models that might use that data to build their own foundation models.

Big food brands “are not going to very quickly turn over that data,” said Hume. Many are debating whether to build their own in-house systems, using models fine-tuned on proprietary data that never leaves their servers. Others are beginning to explore partnerships with companies like NotCo and Shiru that specialize in the discovery layer.

That need for validation may be the biggest differentiator between food AI and other industries. As Hume put it, “You have to bring it into the lab and make sure that it actually works. Otherwise, the predictions are worthless.”

When Hume and I discussed whether large players like Microsoft or Google would eventually dominate vertical-specific foundation models, she acknowledged that possibility. However, she stressed that today’s large foundation models are not yet equipped to deal with the physical and regulatory realities of food. “There’s a ton of very specific know-how that goes into making those models usable for applications like protein discovery or formulation.”

For Shiru competitor NotCo, this highly specific data and domain knowledge are what the company is banking on to solidify its position as a key player in building a foundation model for food, a term now featured prominently on its website.

“I think what people need to understand is that AI is truly about the data sets and the value of the data sets that you have and the logic of the algorithms,” said Muchnick in an interview I had with him in July at Smart Kitchen Summit. “It’s really hard to get to where we were, and specifically also because we weren’t just an AI company. We are a CPG brand, and we learned a lot from being a CPG brand.”

In conversations with AI experts outside of food, several have said we are starting to see the big foundation models open up to allow companies to train them with vertical or highly specific domain knowledge. One pointed to Anthropic’s Model Context Protocol (MCP), which lets a foundation model connect to external data sets to process answers.

Another example is Thinking Machines’ newly announced fine-tuning API called Tinker, which could make it significantly easier for a food brand to train a model with domain-specific knowledge by removing the heavy infrastructure and engineering overhead typically required for custom AI development.

For Shiru, NotCo, and others developing food and ingredient-focused AI, there is still significant opportunity because the field is still so early.

“We’re just starting to see companies thinking about their own internal instances,” said Hume. “A lot of this is in progress, boardrooms are having these discussions right now.”

One of the biggest holdups for food brands is that data ownership and business-model alignment remain unsolved. Who owns the training data and the resulting outputs is a key question, and without clear answers, many companies will hold their data close, limiting the ability of shared platforms to reach critical mass.

For that reason, Hume believes partnerships and licensing models, not open data exchanges, will drive progress in the near term. Shiru’s model focuses on IP discovery and licensing, which allows the company to build intellectual property value without requiring massive manufacturing investments. “Our IP portfolio has doubled year over year since 2022,” said Hume. “Now the focus is on monetizing that through licensing and sales.”

The topic of food-specific foundation models and the adoption of AI by food brands is a fast-moving one, so you’ll want to make sure to listen to this episode to get caught up. You can listen to our entire conversation below or find it on Apple Podcasts, Spotify, or wherever you get your podcasts.

Your Smart Light Bulb Just Told You Your Blood Sugar’s Spiking. Is This The Smart Home’s Next Frontier?

Nowadays, when a smart light bulb shifts from soft white to red, it usually means someone’s at the door, the dog got out, or another home automation routine has been initiated in your smart home app.

In the near future, that same light bulb might tell you the donut you ate for breakfast is spiking your blood sugar.

In some ways, this connection between health tracking devices and the smart home is already happening. One (not-so-consumer-friendly) option is using open-source software like Home Assistant, which allows Dexcom CGM users to create scripts that trigger automations on smart devices based on a predetermined blood sugar level.

There’s also new hardware like the Sugar Pixel, a Wi-Fi–connected alarm clock that integrates with a range of glucose monitors, including Dexcom, Libre, and Gluroo. Many of these connections aren’t through official APIs and are a bit MacGyver’d together, but according to the user guide, you can get your Stelo from Dexcom and other CGMs to send readings directly to the Sugar Pixel.

Startups are also moving into this space. Ultrahuman, for example, is building wellness-sensing devices, like smart rings and glucose monitors, alongside a home hub focused on health. The Ultrahuman hub already measures air quality, temperature, and light. It’s not hard to imagine them linking that hub to their M1 Live glucose monitor or Ring AIR smart ring, creating a home environment tuned around a person’s health biomarkers.

Apple seems like an obvious candidate to lead here, given it has both a smart home framework (HomeKit) and a health framework (HealthKit). But so far, there’s no sign the company is interested in merging the two. That’s not shocking since Apple’s support for the smart home has always felt half-hearted, but it’s still worth keeping an eye on Cupertino for future moves.

For now, these integrations are the domain of early adopters, people comfortable tinkering with open-source software or willing to trust a Wi-Fi alarm clock from a small startup. Long term, though, as CGMs become more democratized and widely used, I expect we’ll see a much stronger connection between wellness tracking and the smart home.