Close Ad

The Spoon

Daily news and analysis about the food tech revolution

News

Report: Diners Opting for Restaurant-Specific Apps & Kiosks While Deemphasizing 3rd-Party Delivery

A new report published by Tillster shows that quick-service restaurant customers are increasingly opting to use restaurant-specific apps over 3rd party delivery apps.

The report, which analyzed the results of a survey of over 1,000 quick-service and fast-casual diners, showed that the number of customers who used restaurant-owned ordering channels over the past three months has increased 25% compared to last year, and 17% of those surveyed say they plan to use third-party apps and websites less in the coming year.

One reason diners are opting for restaurant-specific apps or websites is that they see them as lower cost. 44% indicated they preferred a restaurant’s app or website because it was less expensive. Another reason is the benefits of native restaurant app loyalty programs; over four in ten of those surveyed pointed to the restaurant loyalty rewards and benefits available through a native restaurant app/website.

Another reason third-party ordering apps are losing their shine is the decreasing number of choices as these platforms scale back the number of restaurants they support. According to Tillster, 45% of those surveyed in 2023 pointed to a “variety of options” as the top reason for preferring third-party apps, a number that dropped to 36% of respondents this year.

The survey also asked diners what they thought of in-store ordering kiosks. The report says an increasing number of diners prefer to order using kiosks, with 57% preferring this option compared to 36% last year.

Why are diners growing more enamored with kiosk ordering? According to the report, a growing preference for kiosk ordering is because many diners see them as quicker, more convenient, and a better way to see ordering options. 34% say ordering with a kiosk is faster (up 10% over last year), and 33% believe it’s more convenient (a 22% bump over the previous year). The biggest reason (45%) cited by diners for preferring kiosks is they say kiosks show them all the options, up 10% over 2023.

While newer approaches, such as remote cashiers or AI-voice bots taking orders, have gained an outsized amount of attention (and, in some cases, provoked outrage), the reality is that the most significant transition taking place today is the rapid adoption of restaurant-specific ordering apps and in-store ordering kiosks. The diner largely sees these solutions as an added convenience compared to solutions that are more on-the-nose regarding technology displacing in-venue workers.

With a Summer Opening In Sight, Flyfish Club Now Letting New Members Join in the Old-School Non-NFT Way

Remember Flyfish Club, the Gary Vaynerchuck-affiliated dining club that made news when it announced it was planning to build the world’s first NFT restaurant?

While you may not, the restaurant industry sure does. After the group raised an impressive $14 million in just weeks by selling tiered memberships via NFT, it seemed everyone started wondering, ‘Could it really be this easy?’ (Editor spoiler: For others, it was not that easy. Copycats proliferated, but none matched the success of Flyfish).

Since the group’s launch over two years ago, Flyfish has been mostly quiet, holding the occasional get-together and trickling out updates. Because of this relative quiet, members were naturally excited last week when Flyfish teased on its Discord that an announcement would be coming on April 5th. And, as it turns out, it was a pretty big announcement.

The group showed off what Gary Vaynerchuk described as a first look at the restaurant, calling the images photos even though some of the images—including the sushi chef—were clearly computer renderings. The renderings looked a lot like almost every Vegas restaurant or lounge I’ve ever been in.

More importantly, though, the announcement provided many more details about membership specifics, some of which the early members found frustrating.

One of those details, which can be seen in this Q&A PDF, is the group has made a fairly significant shift in its approach by letting folks purchase memberships in an old-school, non-Blockchain way. According to Flyfish, prospective members can purchase memberships by paying a one-time entrance fee of $1,500 (no crypto needed) and an annual membership fee of $3,500 ($4,000 if you want membership for yourself and a partner).

Another detail that caused some to bristle was the annual membership fee of $500 for “Blockchain” members (meaning members who entered through buying an NFT). As one Twitter user complained, “The mint was ~$8k in eth at the time, +2yrs of waiting and now we have to pay $500/year to get comp’d $500 in expensive food/drinks. Should be 2-3x that per year to make early backers feel whole.”

My guess is the group decided to become more flexible with membership options because the cost of constructing and operating a restaurant in New York City, as well as paying operating salaries to what is essentially a startup in the Flyfish Club team, is incredibly expensive. It’s also a recognition that while NFT membership is a cool idea that no doubt appeals to the Gary V community, most high-rollers in NYC who usually would pay for a dining club are probably not big NFT enthusiasts. While this will undoubtedly take some of the shine of exclusivity that appealed to the crypto in-crowd when it was first announced, I’m guessing that keeping the restaurant solvent is probably a good long-term thing.

Not all the new details were bummers. One new benefit Flyfish announced is partners are now included. While initially, the blockchain memberships meant only the holder was a member, now they say spouses or partners are also included as members. The group also detailed how membership gave members priority reservation access to other VCR Group (the holding group for Flyfish Club) restaurants.

Of course, none of this actually includes food if you are a non-Blockchain member. After these members pay their dues, no food is included. In other words, it’s pretty much the same as every other dining or country club. It’s mostly the same for blockchain members, though Flyfish will apply their $500 membership as a credit towards meals purchased at the club. Beyond that, though, they’re paying for every meal.

We’ll be in NYC this summer, so hopefully, we’ll be able to get a sneak peek. If one of our readers gets a chance to visit the Flyfish Club, drop us a line (with actual photos) and let us know what you think.

Restaurant Cashier Zooms In to Work, Press Freaks Out. Better Get Ready for Lots More Change at Checkout

This week, a raft of news stories about an encounter with a cashier manning the checkout remotely from a distant location went viral.

Dozens of shock-take articles proliferated after a Google employee named Brett Goldstein stumbled upon the remote cashier in a New York City restaurant called Sansan Chicken and proceeded to breathlessly tweet about it. From there, everyone from Eater to the New York Post covered the story. According to a publication called 404 Media, the company providing the virtual labor is called Happy Cashier.

The dystopian framing is a bit surprising, especially since this type of technology is a smart way to deploy actual human workers in places where they are needed. The decoupling of the requirement for a physical presence to do work for jobs that don’t require physical labor has been underway for years, something many users experienced during the pandemic. This technology coming to the front line of the quick service and convenience formats makes perfect sense.

Add to that the reality that remote cashiers have been zooming into work for at least the last couple of years, as companies like Bite Ninja have been building out both a platform and training a workforce of remote workers to work the register. Originally targeting drive-thrus, Bite Ninja expanded to inside the restaurant over the past couple of years, and last year, they said they had certified over 10,000 remote workers to use the technology.

My guess is that part of the reason the interaction got so much traction is because the remote cashier was zooming in from the Philippines. While it could have just as easily been a mom working from home in Louisiana or an off-duty trucker side-hustling from Minnesota, the fact that it was someone working outside of the US, doing the job of someone who normally would need to be there in person, made the many in the press jump at the story for various reasons.

Here’s the thing, though: the cashier is the easiest employee to replace with technology in the restaurant, something which has been apparent for years as apps, in-store kiosks, and, more recently, AI-powered bots start to take our orders. That it’s a worker from a different country should not be surprising, as we’ve seen jobs like phone customer service largely move overseas over the past decade. At least these platforms, like Bite Ninja, provide an opportunity for human workers because the much bigger story over the next decade will be that a significant portion (if not a majority) of the customer-service front-line jobs will be lost to AI. You need to look no further than that one of the US’s biggest fast food operators has developed (and given a name to) its own generative AI-powered customer service agent.

And voice AI agents are only the beginning. As we saw at CES a couple of years ago, virtual avatars are already stepping onto the front lines of customer service, and some of them even have names. Meet Cecelia the avatar bartender everyone.

The bottom line is that it’s always interesting and instructive to watch how these technologies are perceived in the field by customers. Solutions like digital kiosks are accepted as a convenient and time-saving way to order food, while remote workers being piped in virtually initially may induce shock and wonder, even though it’s probably the least technically challenging of all the solutions emerging.

Customer reactions are important because they will be evaluated by chains that are evaluating new technologies for deployment. While a smaller regional player like Sansan Chicken may be comfortable as early adopters of a remote cashier at the main checkout counter, don’t expect bigger chains to deploy these widely until they feel their customers will be okay with the change.

Mycelium Protein & Fresh Produce Production System Wins Canadian Edition of Deep Space Food Challenge

Today, the Canadian edition of the Deep Space Food Challenge announced the grand prize winner for its years-long competition to discover new solutions to feed astronauts on long-term space missions.

The winner is a food production system co-developed by Canadian companies Ecoation and Maia Farms. The system, called CANGrow, grows different types of fresh foods, including strawberries and cherry tomatoes, and also creates mycelium-derived protein using Maia Farms’ biomass fermentation technology. According to the Canadian Space Agency, the CANGrow creators will receive $380,000 in prize money.

Over the last three years, a number of semifinalists and finalists in the Canadian edition of the Deep Space Food Challenge presented system concepts for generating protein using algae, insects, or mycelium (fungal) as the primary source. One of the reasons I suspect CANGrow won it all is its ability to produce both plant-based proteins via biomass fermentation AND fresh produce. Astronauts will need a full range of nutrients while up in space for years at a time, and a system that produces both veggies and protein is a pretty compelling concept.

The CANGrow system has five grow chambers, four of which are dedicated to growing plants using LED lighting for photosynthesis and a UV-treated hydroponic system for efficient watering. The fifth chamber is a small bioreactor that produces mycelium-derived protein using a biomass fermentation process developed by Maia Farms.

If having two Canadian startups co-develop the CANGrow system wasn’t Canadian enough, according to Ecoation and Maia Farms, the system also incorporates accelerated composting technology from Lomi, a smart composting system developed by British Columbia-based startup Pela.

If you’re wondering who the winner is for the NASA/US edition of the Deep Space Food Challenge, you’ll have to wait a bit. According to the timeline from NASA and the Deep Space Food Challenge, the winners of the grand prize for the NASA challenge still have yet to undergo simulation training in NASA’s Starlab Ground Lab, which will take place this summer. After the systems are tested via simulated flight testing, the winner will be announced in August of this year.

Big Tech Set Its Sights on Reinventing Checkout. Consumers Said ‘Not So Fast’

When it comes to technology and grocery shopping, one primary focus for grocery chains and technology providers in recent years has been the checkout experience.

Amazon and various other technology companies have been developing platforms to enable consumers to skip the checkout counter. These platforms aim to transform the shopping experience into something akin to walking into a giant pantry, loading up your cart, and then walking out without going through a checkout line.

Others (including Amazon) pushed technology into the shopping cart, enabling customers to check out products as they walked through the store, get coupons and ads for special deals, and learn more about items via a built-in touchscreen.

And then there’s online grocery shopping. After two decades of slow adoption by both grocers and shoppers, a pandemic forced every major grocery chain to invest heavily in enabling the easiest of all grocery buying options: letting us shop at home and have our groceries delivered to our door.

Meanwhile, everyday consumers continue to do things the way we’ve always done things. It’s a lazy Sunday, and you’re in no hurry? Get in line and chat it up with the cashier and bagger. Are you hurrying to return to work or arrive home in time for dinner? Jump into the self-checkout line and get out as soon as possible. Too busy to head to the grocery store at all? Order online and have stuff delivered to your home.

In other words, grocery shoppers are not a monolith. Most of us change our behavior depending on the current situation.

But what about Just Walk Out? It’s a radically tech-forward evolution of checkout, but one in which Amazon appears to have widely overestimated just how many people would use it and how easy it would be to implement. As I said in last week’s Food Tech News Show (FTNS), self-checkout fits most shoppers’ needs when they are in a hurry, and there aren’t that many situations where consumers feel they need to skip checkout altogether.

As for self-checkout, it definitely isn’t perfect and could be made a much better experience. As Scott Heimendinger said on the FTNS, self-checkout can sometimes be unnecessarily difficult, almost like plugging in a USB. Amazon and others should probably spend their time using technology to make self-checkout work better.

We love robots - FTNS

Target is doing something about self-checkout, changes which it claims will allow shoppers to get out quicker. According to the company, self-checkout lines with cameras were able to check out twice as fast as self-checkout lines without a camera. Of course, their motivation is mostly somewhat self-motivated, driven by the retailer’s desire to limit theft, so my guess is there’s a good chance they can bungle the rollout if it doesn’t deliver clear benefits and customers are feeling spied on.

All that said, while some shoppers may not like it, the combination of computer vision and self-checkout might be the future, particularly if it makes the self-checkout experience less painful than it currently is. Because of this, Amazon should look at repurposing its Just Walk Out into a self-checkout accelerator, not a platform for making shoppers feel like they are shoplifting. For now, however, they’re emphasizing the rollout of their Dash shopping carts, a solution that is unclear if shoppers are asking for. Others, like Instacart, are also betting big on as well. The company had a blog post touting their progress today, saying they plan to have ‘thousands’ of shopping carts deployed by the end of 2024.

Just Walk Out and other light-touch self-checkout will thrive in the near term in shopping contexts where a consumer needs one or two items and is in a hurry, such as airports and stadiums. One of the smartest implementations I’ve seen with self-checkout is at Costa Coffee at SeaTac airport, where they had a Mashgin AI-powered self-checkout station with a dedicated line for customers who just wanted drip coffee. In other words, a quick and low-touch checkout solution for a product with a high degree of certainty where customers are often in a hurry.

The bottom line is that everyday shoppers will continue to shop the way they’ve become accustomed to, choosing between three primary methods: full-service checkout, self-checkout, and delivery. More advanced technology should primarily focus on improving these existing modes. New technology that allows (or forces) consumers to change their behavior should only be used in scenarios that make sense.

Otherwise, consumers will reject it, and retailers will be forced to retrench, just like we saw last week with Amazon’s pullback of Just Walk Out.

Robomart Unveils Autonomous Retail Collective to Propel Self-Driving Shops

Robomart, a company building a platform for self-driving retail storefronts, has announced a new organization called the Autonomous Retail Collective, a group aimed at accelerating the development of self-driving shops.

The group is a mix of Robomart’s partners, including Whale Dynamic, specializing in autonomous driving systems; Avery Dennison, a supplier of RFID technology; Zeeba, which offers fleet management services; and PIX Moving, a manufacturer of custom autonomous vehicles.

Sure, the organization is self-serving for Robomart, but that doesn’t make it a bad idea. The very concept of autonomous retail storefronts is still nascent, and Robomart has been pretty much the only company banging the drum for this concept since they debuted at CES in 2019. While the company has slowly been announcing partners over the past couple of years, including its recent announcement with PIX Moving (a company building the Robomart second-generation – and first autonomous – retail storefront), something akin to an industry alliance like this collective turns what is a loose grouping of companies into a turnkey solution for companies looking to expand into autonomous retail.

“ARC members are our suppliers and infrastructure partners that are committing to the development and advancement of self-driving shops,” Robomart founder Ali Ahmed told The Spoon. “Instead of just one startup building the category, we established the collective to demonstrate the entire ecosystem of players building around this new retail channel.”

For its part, Robomart plans to roll out its PIX-powered vehicles starting in 2025. The company plans to integrate the new autonomous platforms with the current Robomart fleets, which are retrofitted sprinter vans manned by a driver.

The Food Tech News Show: Apple’s Building Home Robots!

Join us at 1 PM Pacific where we’ll be talking about the food tech stories of the week!

This week, we’ll be joined by Scott Heimendinger, former founder of Sansaire and current builder of the next big idea in home knives at Seattle Ultrasonics.

The live stream can be watched below.

We love robots - FTNS

Here are the stories we’ll be talking about:

Apple is working on robots for the home! – From Bloomberg: “The original concept for the robot was a device that could navigate entirely on its own without human intervention — like the car — and serve as a videoconferencing tool. One pie-in-the-sky idea within Apple was having it be able to handle chores, like cleaning dishes in a sink.

Google has used its latest AI to create a new tool that creates a fusion recipe between two culinary cultures. Lifehacker: “According to Google,  Food Mood can provide creative inspiration for your next meal. Select two countries and this fusion recipe generator will create one for you. You can choose whether you would like to cook a starter, a soup, a main course or a dessert. For example, why not try a unique blend of flavors from Sri Lanka and Uganda or mixing influences from Oman and Belgium? This experiment was created by artists Emmanuel Durgoni and Gaël Hugo, and uses Gemini 1.0 Pro via Vertex AI.”

Whirlpool has let the entire Yummly team go. Appliance giant Whirlpool has laid off its entire Yummly team. According to industry sources, the company recently laid off all the employees for the recipe and cooking app and website. These sources tell the Spoon that it’s unclear what the company plans to do with the property it acquired in 2017.

Amazon Pulling ‘Just Walk Out’ from Amazon Fresh Grocery Stores – According to a story published in The Information, Amazon is planning to pull its Just Walk Out cashierless technology from its large-format grocery store, Amazon Fresh.

One Robot Pizza Chain Operator Breaks Down the Cost Each Part of the Pizza-Making Process – For small operators (and big ones as well) in the pizza business, Andrew Simmons’s posts on Linkedin have become must-read material.

Food Tech Weekly: AI’s Impact on Food Accelerating, Whirlpool Lays Off Yummly Team

This is the online version of The Spoon Newsletter. You can subscribe to The Spoon and get deliver director to your inbox.

Over the past month, we’ve seen more and more signals that AI is having a once-in-a-generation transformative impact on the food business. To note:

Yum, the owner of Pizza Hut and Taco Bell, sees its franchises becoming AI-first organizations where the technology will impact every phase of operation.

Chief Digital and Technology Officer Joe Park “sees a future where AI is in every aspect of Yum’s restaurants, with generative AI—the technology behind ChatGPT—in the pockets of franchisees.

“A lot of that gets automated in the future, where you don’t have to interface directly with the technology,” he said. “You can do it through generative AI.”

Google released a new Food Mood tool that uses generative AI to create a fusion recipe between two types of cuisine. 

“This playful fusion recipe generator creates recipes inspired by multi country cuisine with the help of Google AI.

Choose whether you would like to cook a starter, a soup, a main course, or a dessert, and select the two countries you want to create your unique food fusion from. You can even select a dietary preference from the options provided, and include specific ingredients of your choice.”

Researchers released a study that found that survey participants prefer AI-generated images of food over images of the real thing.

“Study supervisor and co-author Professor Charles Spence (Department of Experimental Psychology, University of Oxford) said: ‘While AI-generated visuals may offer cost-saving opportunities for marketers and the industry by reducing the cost of commissioning food photoshoots, these findings highlight potential risks associated with exacerbating “visual hunger” amongst consumers—the phenomenon where viewing images of food triggers appetite and cravings. This could potentially influence unhealthy eating behaviours or foster unrealistic expectations about food among consumers.’”

Food brands are exploring small language models as a way to supplement food understanding and fluency of the large language models behind generative AI platforms.

“LLMs’ shortcomings in creating credible and trusted results around those specific domains have led to growing interest in what the AI community is calling small language models (SLMs). What are SLMs? Essentially, they are smaller and simpler language models that require less computational power and fewer lines of code, and often, they are specialized in their focus.”

There’s a lot happening at this intersection of food and AI, something that clearly is going to have an impact on the old way of doing things. One of the potential impacts is on employees, including editorial and content creators. The Spoon has heard from some industry sources that some big brands are exploring replacing the bulk of their editorial teams with generative AI. While we haven’t confirmed that this is what was behind the recent decision by Whirlpool to let go of the entire Yummly team, it does make you wonder if that’s something the company is pondering. 

The impact of AI on the food business is something we’ll be exploring deeply at the Smart Kitchen Summit in June (as well as the Food AI Summit in September – more on that soon).

We’ll also be talking about this on Friday on The Spoon’s Weekly Food Tech Show, which you can watch on YouTube, Linkedin, Twitter and Streamyard at 1 PM Pacific April 5th. Join us and ask questions during the live show!

Talk soon,

Mike

Whirlpool Lays Off Entire Team for Cooking and Recipe App Yummly

Appliance giant Whirlpool has let its entire Yummly team go. According to industry sources, the company recently laid off all the employees for the recipe and cooking app and website. These sources tell the Spoon that it’s unclear what the company plans to do with the property it acquired in 2017.

The news of the layoffs marks a significant de-emphasis on creating a connected cooking experience tailored around custom-designed recipes with step-by-step cooking.

After Whirlpool acquired Yummly, it beefed up the content team and hired content creators to build a recipe catalog with cooking guidance. It also added features such as built-in food image recognition capabilities and put out a Yummly-connected thermometer (which is still available for purchase). The company announced an update with new features as recently as last fall.

To read the full exclusive story on Yummly, head to The Spoon.

Are You Building The Kitchen Of The Future?

The Smart Kitchen Summit, the pioneering executive summit focused on the digital transformation of the consumer meal journey, is excited to announce its return in 2024. In 2024, SKS will return to its birthplace, Seattle, Washington, scheduled for June 4-5th. Use discount code NEWSLETTER to get 15% off tickets today.

Is The US Power Grid Prepared For The Transition To Induction Cooking?

In case you haven’t heard, electricity demand is shooting through the roof.

After more than two decades of flattened usage due to more efficient lightbulbs, appliances, and factories, the growing adoption of EVs and the explosion in new data centers for compute-intensive applications such as AI over the last few years has resulted in skyrocketing demand for electricity, according to a new report in the New York Times. In fact, forecasters estimate that peak demand in the summer will grow by 38,000 megawatts nationwide in the next five years, which is akin to adding another California to an already overburdened grid.

The Times report does a good job highlighting how EVs and higher usage air conditioning in homes are two of the biggest culprits for reversing the trend, but largely omits any discussion of another potential big driver of electricity usage in the future: induction cooking.

And from the looks of it, induction could significantly impact the overall electricity usage of a family home. While it’s more energy efficient in general, a household switching from gas to electric induction cooking will use more electricity. How much? According to some sources, an hour of induction cooking will use between 1.4 kW and 2 kW per day. That compares with about 2.5 kW per day in charging for the typical EV.

Read the full story at The Spoon.

Podcast: The Story of Mill With Matt Rogers

If you follow the world of kitchen and consumer food tech startups, you know there hasn’t been much in the way of venture-funded startups targeting food waste in the home.

That changed last year when Mill lifted the veil on the company and its first product, the Mill Bin, a smart food recycler. The company’s unique approach included a subscription-based home food waste recycler and an accompanying service that would turn the food grounds into chicken feed. 

We decided to catch up with the company’s CEO, Matt Rogers, to hear about the journey to making Mill. During our conversation, we also talk about:

  • The early lessons in building a tech-powered food recycling appliance and service
  • Why Matt decided to target food waste after building a smart home company in Nest
  • The challenges in getting consumers to think about wasting less food
  • How better data can help us change consumer behavior 
  • The future of food waste reduction technology in the consumer kitchen

You can listen to the full episode below or find it on Apple Podcasts or on The Spoon.

Is The Keto Cereal Craze Over?

I have a soft spot for sugar cereals.

Having grown up in the 80s eating big boxes of Captain Crunch, Lucky Charms, and Life (my friends called me Mikey!), I still salivate when I see big, colorful boxes with leprechauns and monsters in the grocery store cereal aisle.

So when keto-friendly, processed sugar-free sugar cereal substitutes started appearing in 2018 and 2019, I was excited. Like any self-respecting adult, I’d moved on to more responsible breakfast offerings, but saw these new keto-free cereals as a guilt-free time travel machine back to the land of the magically delicious.

I wasn’t the only one. The product’s early success accelerated during the pandemic, a time when people were bored at home and ordering lots of food via delivery. This led to an impressive series B in 2022, where the company scooped up $85 million. That funding fueled the company’s expansion into retail, and now you can find Magic Spoon in places like Costco, Target, and Walmart.

With widespread availability, the company should now be beating the old-school, better-for-you cereals like Grape Nuts and granola, right?

Maybe not. According to a tweet by Andrea Hernández of Snaxshot, Magic Spoon cereal has hit the clearance bin at Sprouts, a chain specializing in premium brands. The pic, which Andrea also posted on Linkedin, led to much discussion about whether the better-for-you keto cereal trend is over.

Read the full story at The Spoon. 

PoLoPo Unveils ‘SuperAA’ to Turn Potatoes Into Protein Factories Via Molecular Farming

Last week, Israel-based startup PoLoPo announced it has deployed its molecular farming technology, a system that uses a genetically engineered potato to produce egg proteins, at greenhouse production scale. The company’s protein production system, which it has dubbed the SuperAA platform, grows proteins within a potato’s tuber, which is then harvested and extracted into protein powder.

Molecular farming, which produces animal protein using seed crops, has gained traction in recent years. The technique, which the Good Food Institute named the “fourth pillar” for alt protein, uses genetic engineering to introduce animal DNA directly into the seeds, transforming the resulting crops into protein factories. Once the genetically engineered seeds are planted, traditional farming management techniques can be employed to grow the crops until they are ready for harvest.

The technique has gained momentum in recent years, partly because of the cost savings it promises to introduce. After all, there is no more efficient way to produce calories for human consumption than by sprouting them from the ground. By transforming plants into small bioreactors, molecular farming companies can take advantage of the scalability and cost-effectiveness of leveraging traditional row crops as protein production engines.

Read the full story at The Spoon. 

Watch as This Robot Pizza Chain Operator Breaks Down the Cost Each Part of the Pizza-Making Process

For small operators (and big ones as well) in the pizza business, Andrew Simmons’s posts on Linkedin have become must-read material.

That’s because Simmons, who I wrote about last year as he experimented with utilizing pizza automation technology in his San Diego area restaurant, has open-sourced his learnings as he continues experimenting with various forms of technology. And boy, is he experimenting!

And it’s not just automation (though that’s a big part). He’s constantly tinkering with every part of his restaurant tech stack as he expands beyond his original restaurant and looks to create a nationwide chain of tech-powered pizza restaurants. Add in the fact that he’s utilizing a crowdfunding model in which he sells subscriptions and a share of future pizza profits, and Simmons has created a live in-process testing lab for how to build a next-gen pizza chain that everyone can learn from.

One example of his highly detailed learnings that I found fascinating is his post today detailing the cost-per-pizza after allocating the costs of the different pizza-making automation he’s deployed in one of his restaurants. The video, seen below, shows how much each part of the process — dough making, doughball prep, dough-pressing, toppings allocation — costs and how he arrives at a 2024 price-per-pie of $1.91.

You can see the full story (and watch the video) about Andrew Simmons’s new cost breakdown of his food robotics stack at The Spoon. 

Watch The Figure 01 Robot Feed A Human, Sort The Dishes, And Stammer Just Like Us

While much of the startup funding for food-centric robots has been for task-specific fast-automation from the likes of Picnic Robot and Chef Robotics, some of the more intriguing – and creepy – action is happening with humanoid robots.

The latest entry into the “watch a humanoid robot handle kitchen tasks” files is from Figure, which just showed off the latest capabilities of the Figure 01 robot by showing how it can identify food and sort through kitchen tasks.

What really stands out to me is the weirdly human voice of the robot, which includes very human-like pauses and slight stammers. As an example, in one exchange, a human interviewer asks Figure 01 to explain why it handed over an apple. Figure 01 responds with a quick “On it” and then goes on to explain, complete with an “uh” pause that makes you almost think there’s an actor behind the curtain spitting out the lines.

You can see the full story (and watch the video) about Figure 01’s cooking prowess at The Spoon. 

Amazon Pulling ‘Just Walk Out’ from Amazon Fresh Grocery Stores

According to a story published in The Information, Amazon is planning to pull its Just Walk Out cashierless technology from its large-format grocery store, Amazon Fresh.

As part of the move, the company will begin to deploy its Dash smart shopping carts. Like Just Walk Out, the Dash carts have embedded computer vision, allowing customers to scan products as they put them in the cart.

“We’ve also heard from customers that while they enjoyed the benefit of skipping the checkout line with Just Walk Out, they also wanted the ability to easily find nearby products and deals, view their receipt as they shop, and know how much money they saved while shopping throughout the store,” Amazon spokesperson Jessica Martin told Chain Store Age. “To deliver even more convenience to our customers, we’re rolling out Amazon Dash Cart, our smart-shopping carts, which allows customers all these benefits including skipping the checkout line.”

That Amazon pulled it from Fresh stores (of which there are 44 locations, nearly half in California) isn’t the end of Just Walk Out. The company plans to continue using the technology in its small-format Amazon Go stores and stadiums (such as Lumen Field).

Read the full story at The Spoon. 

Check Out This Session at Smart Kitchen Summit!

We’re putting together two action-packed days at our upcoming Smart Kitchen Summit, where we’ll discuss how technologies like AI and electrification and emerging trends like the invisible kitchen will change the consumer meal journey. 

One session we’re really excited about is How AI Changes the Game For The Consumer Kitchen, a visionary talk from the founder of Samsung Food, Nick Holzherr. Nick will talk about lessons learned as an early pioneer using AI for consumer recipe recommendation, how Samsung is leveraging AI for its new food app, and where he sees all this heading in the future. 

You can hear Nick’s talk and connect with him, as well as our other great speakers, at the Smart Kitchen Summit on June 4-5th in Seattle. 

Tickets for SKS can be purchased now. Use discount code NEWSLETTER to get 15% off the price of tickets at checkout. If you are interested in sponsoring SKS, you can find out more at the SKS website.

Our Next Food AI Co-Lab Event is on April 18th!

Last month, we kicked off our Food AI Co-Lab with our first ever event!

As I wrote earlier, the Food AI Co-Lab is a collaboration that aims to be a meeting space and learning center for leaders who are building the future of food through artificial intelligence. We will explore different topics, engage with our community, and provide information such as industry surveys about what people are doing at the intersection of food and AI.

We had a great time at our first event talking to Dr. Patrick Story, a professor of Philosophy at Cal Poly, San Luis Obispo, and Kevin Brown, the CEO of Innit, about how they see AI changing food. You can watch our conversation here.

And you definitely won’t want to miss our next event, which will feature Chris Young, coauthor of Modernist Cuisine and founder of ChefSteps and Combustion, and Antonio Gagliardi, technology and design lead for Barilla’s BluRhapsody 3D printed pasta project. You can sign up for this exciting conversation on April 18th here.  I also encourage you to join our Linkedin Group where we will be featuring special content from these conversations as well as more of our projects for the Co-Lab. 

We hope to see you there!

Whirlpool Lays Off Entire Team for Cooking and Recipe App Yummly

Appliance giant Whirlpool has let its entire Yummly team go. According to industry sources, the company recently laid off all the employees for the recipe and cooking app and website. These sources tell the Spoon that it’s unclear what the company plans to do with the property it acquired in 2017.

The news of the layoffs marks a significant de-emphasis on creating a connected cooking experience tailored around custom-designed recipes with step-by-step cooking.

“Every day, millions of consumers around the world use Whirlpool Corporation appliances to prepare meals for their families. The Yummly acquisition will allow these consumers to dramatically reduce the stress from meal planning by helping answer the age-old question, ‘What’s for dinner tonight?'” the company said at the time of the acquisition.

After Whirlpool acquired Yummly, it beefed up the content team and hired content creators to build a recipe catalog with cooking guidance. It also added features such as built-in food image recognition capabilities and put out a Yummly-connected thermometer (which is still available for purchase). The company announced an update with new features as recently as last fall.

The move to let the Yummly team go is indicative of appliance brands de-emphasizing apps with human-powered editorial-driven content, especially as some start to investigate how they can leverage generative AI to power new features and content. The Spoon has heard rumblings that other appliance brands are starting to build generative AI-powered content libraries and consumer-facing UI with mixed results. My guess is this trend will only continue, even as appliances begin to revisit their smart appliance strategies after lessons from the first wave of product build-out.

Amazon Pulling ‘Just Walk Out’ from Amazon Fresh Grocery Stores

According to a story published in The Information, Amazon is planning to pull its Just Walk Out cashierless technology from its large-format grocery store, Amazon Fresh.

As part of the move, the company will begin to deploy its Dash smart shopping carts. Like Just Walk Out, the Dash carts have embedded computer vision, allowing customers to scan products as they put them in the cart.

“We’ve also heard from customers that while they enjoyed the benefit of skipping the checkout line with Just Walk Out, they also wanted the ability to easily find nearby products and deals, view their receipt as they shop, and know how much money they saved while shopping throughout the store,” Amazon spokesperson Jessica Martin told Chain Store Age. “To deliver even more convenience to our customers, we’re rolling out Amazon Dash Cart, our smart-shopping carts, which allows customers all these benefits including skipping the checkout line.”

That Amazon pulled it from Fresh stores (of which there are 44 locations, nearly half in California) isn’t the end of Just Walk Out. The company plans to continue using the technology in its small-format Amazon Go stores and stadiums (such as Lumen Field).

The move, which comes as part of a larger effort to redesign its Fresh stores, is yet another sign that the company’s retail and grocery strategy remains somewhat rudderless. The company paused its Fresh expansion in February of last year before deciding to apparently unpause it later in the year. It has also closed its Fresh pickup format early this year.

My guess is that part of the challenge is that the company seems confused about whether it’s a technology platform provider or one serious about pursuing its own grocery store business. The acquisition of Whole Foods signaled the company was serious about grocery, but since then, it has continued to expand its efforts to pioneer AI and computer vision formats while seemingly neglecting to evolve the Whole Foods stores as the grocery-buying public started craving more flexible solutions around things like curbside pickup.

This move seems to signal that Amazon’s grocery business is beginning to embrace the post-Bezos era (he’s largely stepped away from day-to-day and is focused on his rocket company), moving on from being a company that often pushed new and experimental concepts at the expense building businesses with more conventional, straightforward approaches. According to The Information, the company plans to “spruce up” its stores “across the board” as it prepares to expand Amazon Fresh locations later in 2024.

Watch as This Robot Pizza Chain Operator Breaks Down the Cost Each Part of the Pizza-Making Process

For small operators (and big ones as well) in the pizza business, Andrew Simmons’s posts on Linkedin have become must-read material.

That’s because Simmons, who I wrote about last year as he experimented with utilizing pizza automation technology in his San Diego area restaurant, has open-sourced his learnings as he continues experimenting with various forms of technology. And boy, is he experimenting!

And it’s not just automation (though that’s a big part). He’s constantly tinkering with every part of his restaurant tech stack as he expands beyond his original restaurant and looks to create a nationwide chain of tech-powered pizza restaurants. Add in the fact that he’s utilizing a crowdfunding model in which he sells subscriptions and a share of future pizza profits, and Simmons has created a live in-process testing lab for how to build a next-gen pizza chain that everyone can learn from.

One example of his highly detailed learnings that I found fascinating is his post today detailing the cost-per-pizza after allocating the costs of the different pizza-making automation he’s deployed in one of his restaurants. The video, seen below, shows how much each part of the process — dough making, doughball prep, dough-pressing, toppings allocation — costs and how he arrives at a 2024 price-per-pie of $1.91.

Simmons details how he’s tinkered with different automation systems over the past year and how they’ve impacted the price. One change he’s tinkered with is switching out the Picnic pizza robot for a Middleby Pizza Bot, which is more expensive but handles more of the pizza-making process and requires less human intervention.

From Simmons’s post:

Last year, the financial model was built using the Picnic Pizza Station. It was more expensive last year than it is today. This year, I’ve incorporated The Middleby Corporation Automation tool into the equation, but either unit could work. Middleby is a little more costly, adding about 60¢ to the per pizza estimate, but it takes the pizza from dough blank to cooked, whereas the Picnic requires some intervention to cook it. Picnic runs about 38¢ per pizza this year.

Simmons points to recent changes in California’s employment laws as one motivator for his becoming an early adopter of these solutions, saying that the changes will lead to more restaurant chains experimenting with automation.

“Thank you to the pioneers in this space that have tried, adopted, succeeded or failed, equipment manufacturers and restaurateurs alike; and to Governor Newsom, for accelerating adoption of automation,” wrote Simmons.

You can (and I suggest you do) follow Simmons’s posts about his journey to build a robotic restaurant chain on Linkedin.

When It Comes to Making Generative AI Food Smart, Small Language Models Are Doing the Heavy Lifting

Since ChatGPT debuted in the fall of 2022, much of the interest in generative AI has centered around large language models. Large language models, or LLMs, are the giant compute-intensive computer models that are powering the chatbots and image generators that seemingly everyone is using and talking about nowadays.

While there’s no doubt that LLMs produce impressive and human-like responses to most prompts, the reality is most general-purpose LLMs suffer when it comes to deep domain knowledge around things like, say, health, nutrition, or culinary. Not that this has stopped folks from using them, with occasionally bad or even laughable results and all when we ask for a personalized nutrition plan or to make a recipe.

LLMs’ shortcomings in creating credible and trusted results around those specific domains have led to growing interest in what the AI community is calling small language models (SLMs). What are SLMs? Essentially, they are smaller and simpler language models that require less computational power and fewer lines of code, and often, they are specialized in their focus.

From The New Stack:

Small language models are essentially more streamlined versions of LLMs, in regards to the size of their neural networks, and simpler architectures. Compared to LLMs, SLMs have fewer parameters and don’t need as much data and time to be trained — think minutes or a few hours of training time, versus many hours to even days to train a LLM. Because of their smaller size, SLMs are therefore generally more efficient and more straightforward to implement on-site, or on smaller devices.

The shorter development/training time, domain-specific focus, and the ability to put on-device are all benefits that could ultimately be important in all sorts of food, nutrition, and agriculture-specific applications.

Imagine, for example, a startup that wants to create an AI-powered personalized nutrition coach. Some key features of such an application would be an understanding of the nutritional building blocks of food, personal dietary preferences and restrictions, and instant on-demand access to the application at all times of the day. A cloud-based LLM would likely fall short here, partly because it would not only not have all the up-to-date information around various food and nutrition building blocks but also tends to be more susceptible to hallucination (as anyone knows who’s prompted an AI chatbot for recipe suggestions).

There are a number of startups in this space creating focused SLMs around food and nutrition, such as Spoon Guru, that are trained around specific nutrition and food data. Others, like Innit, are building their food and nutrition-specific data sets and associated AI engine to be what they are terming their Innit LLM validator models, which essentially puts food and nutrition intelligence guardrails around the LLM to make sure the LLM output is good information and doesn’t suggest, as Innit CEO Kevin Brown has suggested is possible, a recommendation for “Thai noodles with peanut sauce when asking for food options for someone with a nut allergy.”

The combination of LLMs for generation conversational competency with SLMs for domain-specific knowledge around a subject like food is the best of both worlds; it provides the seemingly realistic interaction capability of an LLM trained on vast swaths of data with savant-y nerdish specificity of a language model focused on the specific domain you care about.

Academic computer scientist researchers have created a model for fusing the LLM and SLMs to deliver this peanut butter and chocolate combination that they call BLADE, which “enhances Black-box LArge language models with small Domain-spEcific models. BLADE consists of a black-box LLM and a small domain-specific LM.” 

As we envision a food future of highly specific specialized AIs helping us navigate personal and professional worlds, my guess is that the combination of LLM and SLM will become more common in building helpful services. Having SLM access on-device, such as through a smartwatch or phone, will be critical for speed of action and accessibility of vital information. Most on-device SLM agents will benefit from persistent access to LLMs, but hopefully, they will be designed to interact independently – even with temporarily limited functionality – when their human users disconnect by choice or through limited access to connectivity.