Welcome to the Community Call Podcast.
I am Melissa Travers, Director of Community here at BevNET & NOSH, here with my co-hosts Monica Watress and Mike Schneider.
If you're enjoying the show, please follow and review us on Apple Podcasts or your listening platform of choice.
I would like to start off this show by celebrating our newest Slack community member, woohoo!
Ken Chirouki, President of Sun Noodle.
So he just joined us on Slack.
I feel like that's a big, that's a big get.
Yeah, I'm impressed with us.
Yeah, thank you.
He also was kind enough to send some samples to the office.
Some of which made their way into my refrigerator.
I'm so glad.
He sent four different varieties, miso, shoyu, spicy sesame and kai dama.
I looked that up.
It translates to extra ramen.
So that fourth skew, kai dama, it's just noodles.
It's just lots of noodles.
Don't you love that there's a way to just talk about extra noodles?
In a really fancy way.
Yeah, right?
Right?
So I caught up with him recently and they're majority food service, but they're spreading into retail.
So they're in Whole Foods Market, Foxtrot, Wegmans.
And actually they were at a restaurant that you and I went to, Monica, when you were last visiting the BevNET Newton headquarters, Little Big Diner.
What'd you think of the noodles there?
They were very good noodles.
I concur.
And here's one last little fun fact that I heard from Kenjiro.
Did you know that they make 200 varieties of noodles and before the pandemic, they made 400?
Can you imagine?
They downsized to 200?
Yeah, because they do so much food service and they work with so many chefs who are looking for something specific.
I cannot even imagine keeping track of all that.
If anybody else would like to send us samples, our address here is 65 Chapel Street in Newton, Massachusetts, 02458.
So if you send us samples, we'll probably talk about them and we'll definitely eat them.
Monica, you just recently had an Instagram post that I liked a lot.
I believe it was about Chutney Punch's gorgeous packaging.
As a matter of fact, I did.
So I'm gonna describe it for the listeners.
It's, I mean, I don't even know how would you describe it.
It's colorful, it's regal, it's sophisticated.
It's Indian inspired and just uses, incorporates a lot of colors and imagery.
You nailed it.
I think those are all the good things and you're definitely evoking the image of something that's very, very premium.
It's in the shape of a Pringles container, but it's like a Pringles container fit for a king in heaven.
Seeing your post made me think about containers that you want to use over and over and over again.
Like I think I buy Talenty ice cream just so I can use it.
Those are fun.
They're the best, yeah.
And the tops grew on really well, so they're good for like dressings and stuff.
Then mini Bon Mamon jam containers, I actually got rid of one.
And a friend of mine told me that she puts moisturizing lotion in them for traveling.
And then I was kicking myself for throwing it out.
So I guess I'll just have to go find another one at like a breakfast buffet.
Growing up in the 90s, my family and I ate a lot of leftovers out of Cool Whip containers.
And I can't believe it's not butter tubs.
Oh my gosh.
We also ate a lot of Miracle Whip.
Did you really?
Did you reuse the Miracle Whip containers?
I don't think so.
It's not a miracle.
There's no miracle in there.
It's a miracle anyone eats it.
Right.
Yeah, I mean, mayonnaise is the way to go, right?
Aioli's the hot new kid on the block.
Yep, we were just talking about Haven's Kitchen Aioli last week, would you say you're a Hellman's or a Duke's person?
If you had to choose.
Hellman's.
I think I'm a Duke's.
I think I'm a Duke's girl, yeah.
It's a little sweeter.
We should ask Ken Chiro how to say more aioli.
We need an aioli that's just more aioli.
Yeah, that's a great point.
Maybe there's a good word for that too.
I bet there is.
Right?
Mm-hmm.
Monica, how about you?
What's your, are you a mayonnaise person at all?
I'm not.
Yeah.
But I do like the plant-based mayo from Fabolish.
It's made with aquafaba, which is the leftover water from chickpeas.
Yep.
It's a good mayo.
It's amazing that you can make a mayonnaise-like product from chickpea water.
It's a really versatile ingredient.
I mean, it's like a by-product if you think about it.
And you can whip it up into a lot of different things.
Yeah, yeah, it's amazing.
It's like a, it's a miracle.
Now, that's a miracle.
That's a miracle.
See, it's too late.
They already took the name.
Too bad for you, Fabolish.
But the product is amazing.
I was also thinking about the kinds of containers you wouldn't want to reuse.
And the first thing that came into my head was meat.
But then the second thing I realized is that my grandmother used to save those like styrofoam sort of platforms that meat used to come on.
You know what I mean?
And they have like the plastic wrap on top.
Don't you think she saved those little platforms?
She would wash them off and save them.
And if you left her house, she'd put some cookies on it for you and wrap it back up in plastic.
So they become like beef infused cookies.
Yeah, I mean, like she washed them.
But when I think about the kind of packaging I don't want to reuse, it usually includes raw meat.
What other containers would you never want to reuse?
I once tried to repurpose a mini hand sanitizer bottle to bring salad dressing to work in like a little salad kit type situation.
Wasting your coworkers.
It definitely tasted like Harvest Pumpkin or whatever the flavor of the scent.
I probably wouldn't reuse like a varnish container.
Yeah, a gasoline can.
A gas can, oil.
That makes me want to show up to a potluck with like a gas can.
Motor oil, motor oil.
Yeah, I probably wouldn't reuse olive oil either though.
I mean, unless I was adding more olive oil.
Or making a dressing or something.
Monica, yours makes me think you wouldn't want to use a Yankee candle jar for anything.
Yeah, well, I'm glad we got to the bottom of all this, Mike.
I feel like you could create some fun AI mashups of repurposing containers.
Let's do it.
AI could certainly be used for that.
This episode of Community Call, which is a companion piece to a NOSH article on the same topic, which is how brands are actively using AI to help fuel innovation in business operations.
Brad Avery, Senior Reporter at BevNET & NOSH, joined us on Community Call to talk about this topic, along with Peter Barrick of Rivalz and Melissa Li, Investor at Offline Ventures.
Check out the article on nosh.com, and we hope you enjoy this conversation.
Get the latest in natural food industry news, trends and data with the NOSH Daily Briefing newsletter.
Choose the free light edition or upgrade to Insider Access for exclusive story recaps and insights.
Stay informed and stay ahead by signing up at nosh.com/dailybriefing.
Thank you.
Today on Community Call, we are going to delve into the intersection of artificial intelligence and CPG, exploring how AI is evolving everything from product development to marketing and operations.
This Community Call is a companion piece to an article on nosh.com, so make sure you head over there for more information.
My guests today are Peter Barrick, CEO of the snack brand Rivalz, Melissa Li is an investor at Offline Ventures, and our very own Brad Avery is senior reporter at BevNET & NOSH.
Thank you all so much for joining me.
This is going to be a great discussion.
Why don't we start off with Brad?
Could you set the stage a little bit and help us understand how you see AI technology being used in CPG right now?
Yes, absolutely.
So for the article I wrote, I wanted to really peel back the layers of AI and the discussion around it, because right now we're very much in this new, exciting phase of what this technology is.
Everyone's sort of learning about it for the first time.
And so there really is a lot more than just on the surface.
I think we're seeing with food and beverage brands, you see Coca-Cola doing an immersive AI experience.
You see brands going on their social media and auto-generating images for Instagram or copy from ChatGPT, but there's so many deeper uses that are going to be much more transformative than just that marketing element.
A small example is a small startup brand that I spoke with who said they simply used ChatGPT to an ID and ingredient that they couldn't get an answer for from anyone else.
They needed a preservative that wasn't going to damage the taste of the product or artificial.
No one that they asked could tell them.
They punched into ChatGPT and they got the answer.
That's a very small example.
That's just the lowest level of how this technology is being used and as Peter's going to talk about, it's being used for recipe development.
There's uses for supply chain, warehouse management, tracking trucks on the road.
It's a way to really track and manage the entire back end of a company in ways that have never been easier to do so long as you can access the technology to do it.
Great stuff.
Thanks, Brad.
Well, let's dive below the surface a little bit more.
Peter, please tell us a little bit about your background and about Rivalz.
Rivalz is certainly a brand that's using AI in a way that many other consumer package goods haven't started using yet.
Yeah, absolutely.
So Peter Barrick, CEO of Rivalz, my background, it's bifurcated in between military, I was a former Marine officer, combat veteran, decorated, but then also a startup CPG veteran as well.
Multiple decades in leadership, it's some of the most stressful and uncertain environments out there.
So my core competency is essentially controlling the chaos and moving forward to a clear mission vision, right?
And so for Rivalz, our mission is, we're gonna bring great tasting, affordable nutrition to the markets worldwide, with snacks that taste better, feel better and do better.
Our goal is to roll back the 50 year pandemic of malnutrition, diabetes and obesity.
And so for every bag sold, Rivalz bag sold, our customers have a new lease on life.
They can live more in the moment while maximizing.
Melissa Li, I would love to hear a little bit more about your background and how you became interested in AI.
Yeah, absolutely.
So for background, I'm a seed stage investor, meaning that I invest in disruptive technology companies, a year, a year and a half out from formation.
And so often we are one of the first checks in.
Occasionally I will write pre-seed or series A investments.
But the idea here is like, what is the fundamental technology that is shifting the techno economic curve?
What are some high stakes cases where the technology is making things possible that weren't previously possible?
So AI is a really great example of this.
Obviously there are, you know, what I would call low hanging fruit around copywriting.
You know, 2023 was the era of image generation and consumers definitely had a lot of fun with that.
I think in 2024, we're going to see more high stakes enterprise use cases where AI is bringing in brand new capabilities that weren't even possible, not even by human teams that were working around the clock.
And so I'm really excited to delve more deeply into some of these use cases.
But broadly, I've sort of seen developments in AI on two fronts.
The first front is obviously you have your highly technical AI startups that are innovating at the model level.
But then you have on the other side of the table, consumer facing brands that are deploying AI internally for more efficiency.
And I think, you know, nothing is the same anymore.
You know, the world is talking about AI and for all the right reasons, rightfully so.
Very excited to dig into what this means for CPG founders.
Peter, could you explain the process that you use to integrate AI into Rivalz processes and procedures?
How did you come to integrate it the way that you did?
At Rivalz, since we have a big mission, right?
We look at AI in a macro way.
So how we think about AI, it's a very, very powerful tool, right?
If you use it correctly, you can solve big problems, right?
And so how we applied it was to the cornerstone of our business, cooker extrusion technology.
So the current problem set in the salty snack category is ingredients and how you extrude a salty snack.
The cooker extrusion technology requires cheap carbohydrates, which tend to be unhealthy and provides undernourishment for the consumer.
So how can we bring great taste and affordable nutrition to a product that doesn't exist in the current, in the marketplace?
So every entrepreneur asks the fundamental question, what's the need and want that's not being addressed by current products in the market?
And so we identified what that was, Friedelize just published an article, which we have been understood and identified in 2022.
They call it the time crunch dilemma in snacking for tasty satisfaction.
That's the need and want that's not being met.
We use AI at the extruder, we're the first company in the selfie snack category to do that in order to achieve a design of product that can meet that need.
How do you use artificial intelligence in the product development process?
I assume it's not feeding a question into chat GBT.
So what did we do?
We identified ingredients that we could use to create a next level snack.
So if you're looking at the trends in the salty snack category of legacy brands, they're undeniably delicious, but they're devoid of nutrition.
Then you have functional brands that are noble.
They have better source ingredients and lifestyle certifications.
And then there's Rivalz with eight grams of protein, four grams of fiber, nine net carbs, low glycemic load, seven essential vitamins and minerals.
And it tastes good.
We're at the next evolution.
There's no more excuses.
You can have a great tasting snack.
Innovation and technology provides the capability for that.
That's also nutritious.
And so what did we look at?
When we identified the ingredients we wanted to use, we went ahead and extruded it, right?
We have some of the best extrusion engineers on our team from top companies.
And it was utter failure, right?
Because we need cheap carbohydrates in order to get the cell structure and expansion rate for a delicious product, right?
And so it was utter failure.
And so we asked ourselves, is there a better way?
Right?
And there is.
And so we harnessed the power of AI.
So what do I mean by that?
So when you're looking at cooker extrusion technology, it's a trillion dollar industry worldwide.
We extrude everything from pasta to pet food.
And since the 1930s, cooker extrusion technologies remain relatively unchanged.
You have brilliant extrusion engineers testing one variable using the scientific method, see a given outcome.
Does it hit a nutritional or sensory profile?
And they move on to the next.
The problem with that is that product design typically takes 18 months to 36 months, depending on how difficult the concept is.
When we went to look at after our first failure, how many trials we would have done in order to get the snack that's in the market today, it was over 500,000 experience, right?
Meaningful experience.
With AI, we were able to whittle that down to 71 impactful trials.
And so we're rapidly able to figure out how to get the cell structure and expansion rate to maintain that indulgent texture and flavor while adding best-in-class nutrition to deliver a product that consumers want and need, right?
And it's a very difficult concept.
And then why is it difficult, right?
So you have ingredients that behave differently under stress.
You have hundreds and hundreds of processing variables at the extruder level.
You have dosing levels that affect nutritional and sensory outcomes, and then you have those benchmarks themselves.
It's a highly complex, multi-dimensional problem that is hard to fix.
So we, this first company focused on an extruder.
Our outcome was the Rivalz Stuff Snack.
We developed technology out of that, which I'm not gonna get into.
And we've padded, right?
So we have the ability to do great tasting, affordable nutrition, and our competitors can.
And we have a large mode around us, et cetera.
I have a technical question for you.
And certainly, I won't ask you to divulge any Rivalz secrets.
What's the format of the software that you're using?
So we've partnered up with one of the best AI firms in the country.
And we have an intimate relationship.
They're also a partner.
So we're able to do things.
We're able to push the envelope in an aviator term.
Where we can adapt what we want to achieve, right?
And so we have the capability to write new algorithms, et cetera, capture the data, link it to outcomes, and our AR partner will adapt the algorithms to our specific needs.
Which is abnormal in the space today, right?
To have a CPG company that's working with a partner that will adapt to our wants and desires, right?
So we're trying to solve food categories, hardest problems to advance humanity forward and not look at the offset of that is efficiencies, right?
Which is the big theme of AI, right?
To make everything more efficient.
We're doing that as a byproduct of solving hard problems, if that makes sense.
And one last question for you, Peter, before we move on.
In order to be able to use cutting edge technology, there's usually a very high price tag.
What's the barrier to entry for technology like this understanding that so many of the brands who would be interesting in being able to cut labor and get to solutions more quickly are likely in the emerging space?
Yeah, so that's a great question, right?
So fellow CPG startup companies can use a range of AI assets, varying degrees of costs, right?
So there's stock AI out there, which costs 25 to $30,000 where you can figure out some consumer trends.
You can figure out what ingredients that you need to use for your recipe in order to achieve a given outcome.
So there's a lot of firms that do that, or just in the marketing space, optimizing and making it more efficient for marketing messaging, PowerPoints, et cetera.
And so Melissa can talk more about that, but there's a wide range of possibilities with AI.
Melissa, I would love to hear from you some of the ways that you are seeing AI be deployed in consumer packaged goods.
And I know that due to the breadth of your experience, there's some food and beverage, but then there are other examples in clothing and cosmetics as well.
So how are you seeing it be deployed?
Yeah, this is a fantastic question.
And so much of what Peter said just now really resonates.
I think I'll start with talking about F&B applications, given that we're on BevNET and NOSH.
But effectively, what generative AI models do is they play huge numbers games against themselves at near-instant speed.
We're now at a level of computational power in the year 2024, where AI can effectively scramble lots of outcomes against lots of other outcomes, which makes it really great for things like recipe development, when you have literally any number of inputs being matched with any other huge level of other inputs, and then you factor in things like cooking time, and what food items work together well on a chemical and molecular level.
Taste is a really difficult thing to get right.
Different demographics of consumers have different taste profiles.
And so AI is really, really powerful here, where it gets smarter over time.
The more data that it ingests, the more prompts that users feed it, it learns that all becomes part of the training corpus.
And so to personalize the model for your own uses, you might be asked to give the model feedback or you might be prompting it in particular directions, but that then trains the model for the benefit of the next user.
And so where there is mass adoption, this becomes super, super valuable for CPG recipe developers at scale.
And I think the exact same goes for allergen management, where when you're looking for the right recipe substitutes, it's difficult to figure out, okay, if I sub in this one ingredient that is kind of the same, but not really, do I have to swap out a number of other ingredients as well to get the taste profile right?
And that's where AI models playing, you know, numbers games against itself really does help.
And I think IBM has an AI chef now that is also doing recipe formulation.
And interestingly, the same goes for chemical formulations across beauty products, for example.
There's a lot of makeup companies now leveraging AI in bits and pieces of their R&D process.
I think perfume is another one where scent notes are a complex matter to get right.
And so AI can be incredibly helpful there.
And then the other case that comes to mind immediately for food and beverage, and this is potentially too upstream for us, but, you know, when we're thinking about like crop management, you know, AI works in tandem with sensors and farm machinery.
And so it's kind of doing two things at once.
The first is that it's taking in massive amounts of data from on the ground over time and assessing productivity, but it's also making active recommendations and figuring out how to optimize so that there's less downtime and machinery.
And, you know, you're better managing your supply chain and the likes of that.
But I feel like that's maybe a little too upstream for food and beverage entrepreneurs.
And then the other thing that I'm seeing, you know, more broadly and not F&B specific is generally how AI has democratized business ownership and business operations.
And this is a piece that I'm really excited by because I love to be an advocate for new innovation and SMB is happening, you know, in all sectors of the economy.
You know, I'm really excited now that, you know, anybody can monetize their existing audience or personal brand by launching products with very lean teams and with very little startup capital.
You know, previous to AI, you had to have some of the best operating experts in the world on your side.
You know, you kind of have to figure out, like, with the money available, you know, how can I hire the very best resources?
And it's actually a lot that AI can do as a thought partner for entrepreneurs from even the very earliest stages of their business, from customer discovery to mood boarding, to mocking up your earliest designs, to actually generating those designs, to helping you locate the right manufacturers, to helping you create brand assets, to be able to distribute that to the market more quickly.
You know, there's AI for, you know, brand guidelines and figuring out what's the type of messaging that works best with the market.
So you can kind of see how AI comes into business creation and business operations truly end to end, which I think is really exciting.
And it wasn't that long ago that Sam Maltman of OpenAI was talking about how, you know, AI enables the next one man billion dollar company.
If we think about modern day unicorns, you know, even the leanest operating unicorns have like dozens of team members, if not hundreds.
And the big question used to be, do you have enough venture capital to get started and to grow at an exponential pace?
And can you access the very best team of advisors and operators?
And now I think the real differentiator is, do you have the right AI capabilities and skill set?
Are you truly adaptable?
Are you open to experimenting with technology?
Because as a solo founder or internet creator, if you can figure out a way to loop AI into every step of your process, I think it absolutely changes the game.
There've been tools for a while, and I would say that this is somewhat lower stakes just because it's talked about so much, but things like generating storefront pages or generating a website with just a couple of clicks or AI taking care of your copywriting.
But there's also things like logistics and inventory management.
That's a really difficult thing.
And obviously COVID disrupted the supply chain a lot for a lot of businesses across sectors.
We're looking at applied AI and machine learning algorithms where it can take historical sales data and your customer data and then predict future demand so that you're never gonna run into a situation where you overstock or understock, and that's better for cash flow management and a whole host of other benefits as well.
And the last thing I'll say here is that, I think right now there are lots of AI tools and platforms that take care of different pieces of the business cycle.
You know, one company will take care of copywriting and another company will take care of demand forecasting.
But I think there are going to be platforms increasingly that try and bring all of these functions under the hood and offer this as an all-in-one service offering to, you know, enterprises, certainly in business owners, but even for the everyday creator who has no idea where to get started in business and just wants a landing point.
And a company called Pietra comes to mind, that's P-I-E-T-R-A.
I believe Pietra started by helping, you know, famous celebrities and internet creators launch products to their existing fan base, but now it's really helping even micro influences and the everyday consumer launch Shopify products that they want to.
You know, there's AI to supplement the product design process and then create brand assets and marketing materials.
Then you've got shopfront design.
Then you've got a network of the right packaging providers and where to source the products from.
There's shipping infrastructure.
They take care of storage and logistics and there are human experts to help along the way, which I think is also another piece that we can talk about if we want to, you know, having a human in the loop.
But I think, you know, incredibly, incredibly powerful capabilities for the next generation of business owners and just to someone who loves to see innovation thrive, I'm really excited.
So many examples of how organizations are using AI.
A question that I have is software like SIN7 and Deere, for example, are very expensive ways for consumer package, goods, brands to track order management, logistics, that kind of thing.
How within reach is the technology that you're seeing to brands, especially if they're in an emerging space and under $10 million?
Yeah, this is a great question.
I think the important thing here is for brands to like to take a good look at their workflow and to figure out what is the most important piece of this that they want to automate away.
Like what is the deal breaker?
Where is the highest leverage to do so?
I think we're certainly approaching a future where you can probably leverage AI for every piece of your business creation and operation process.
But I think for now, like zooming in on is it that inventory management is going to be the killer for us if we don't figure it out in the next six months?
Is it that like product R&D is incredibly expensive and we just had to let go of some scientists who we couldn't afford to pay because we don't have cash coming in?
You know, is it something else?
And so figuring out that number one priority and investing into that is important.
And I would also say like there are ways to start small and there are ways to experiment and ramp it up as opposed to really committing to, you know, a one year subscription upfront.
There are lots of consumer facing applications and AI that are really great to experiment with and learn from.
And then, you know, as a business owner, if you can get acquainted with how does this work, what do I think the utility is, even on the consumer side, then that's a baby step towards figuring out whether you want to shell out for, you know, the enterprise version or, you know, at the very least, you can map your workflow and figure out where is the utility for my sector and my business specifically.
I also know some smaller brands that are, you know, more tech savvy founders who are, you know, playing around with their own tools.
And they know a bit or two about coding and about the technology.
So they're able to kind of take some of their datasets and, you know, take some of these more accessible tools and really make the most out of them, as far as I'm aware.
Peter, were there any surprise difficulties or benefits that unraveled as you became more and more entrenched in using artificial intelligence for Rivalz?
Well, yeah, there was plenty of discoveries, right?
Learning how to learn, et cetera.
And Melissa brought some great points up that are relevant.
So productivity, right?
So reducing, you know, the entrepreneur of one, reducing the potential, reducing the labor force down to one entrepreneur and AI.
That's a very interesting point.
I have several comments on that.
But first, how does productivity apply to Rivalz?
So the product that is on the market right now shouldn't be on the market.
It should be ready for development 12 months from now.
Right?
It's in the market.
So we're capturing revenue we shouldn't have captured.
We're not spending on R&D, we should be.
But most importantly, we're talking about productivity.
So with a team of 1.7 on our R&D team, some of the best in class R&D professionals, so one full-time employee, the rest are consultants.
We were able to achieve in six months' time what a full team of seven at a traditional strategic company would do.
And it would take them 18 to 36 months.
We did it in six with 1.7.
So you're talking about the power of AI.
That's where it's at right there.
That's a perfect example of productivity improvements.
Secondly, we had a return on asset base, right?
So we are using a ubiquitous extruder and fully automated line.
That's half the cost of what strategics would have to use.
And that asset base is worldwide.
So when we expand internationally, we're not gonna have to spend millions and millions and millions of dollars on cap apps.
So some incredible, lessons learned on productivity as it relates to food and beverage.
Now, the one thing I would like to point out to the listeners is be careful.
And the reason why I say be careful is that if you're using AI and it is supplementing a lot of your subject matter expertise, you may be blindsided on the outcomes.
And so what we've chosen to do is pair AI up with subject matter experts.
So it's sort of like a co-pilot if you will.
And so the part of entrepreneurship that's special, and that's why we do things we do, is we take multiple variables, multiple opportunities, and we try to find the truth within that nexus of information.
Can we create a new product that consumers want?
And so if you don't have that expertise to understand what AI is computating, predictive analytics, is accurate or not on the given outcome, then I think you could be steered in a different direction.
So it's these novel insights are wonderful, but it takes a skilled operator to digest it and understand, and then to move forward and implement it rapidly.
I'm curious, were there any particular difficulties you found in the formulation process with AI?
And you can be as specific as you want, like the shape that came out of the extruder wasn't quite what you were looking for.
So when you're trying to do hard things, right, there's failure, right?
So we're using, what we tried to do was extrude high protein, high fiber, low carbohydrates, which is completely antithetical to what the extruder is used to do.
And then, you know, so that was the biggest crux of the problem is how can we set the multiple variables up in a given way, get there quickly, without having to use trial and error, have a better chance at success?
So we had that technological advancement of how do you extrude high protein, high fiber, low carbohydrate?
And why is carbohydrates, it's gain changing in the salt and snack category?
Well, if you look at the category, it's pervasive, right?
And with high carbohydrates come with a blood sugar spike, insulin resistance, diabetes, et cetera, right?
So how can we create a snack that has sustained energy, satiating and tastes great to the consumer?
And AI allowed us to quickly figure that out.
Thanks for that, Peter.
Melissa, I know that you have your eye on what's going out there in a way that perhaps none of us here are.
What are your concerns with brands using AI and what are some of the difficulties that you are seeing them incur?
Yeah, this is a really great question.
I think certainly what Peter said around the importance of having a human in the loop, that really resonates with me.
You know, for most of the companies that I've seen, they largely have, you know, domain experts act in more of an advisory capacity.
So where those domain experts were previously doing all of the work from ground up, now AI lays the groundwork and those domain experts come in and sort of do phase two checking.
Is the human in the loop just to make sure that AI is outputting correct results?
And so also for teams, just to understand how and where AI is fitting into the workflow means that they can figure out, okay, where does the human come into the loop?
Where is this somewhat lower stakes?
And, you know, we can afford to have a spelling error, for example.
And so figuring out, you know, and like, you know, my advice here is like for teams to map out their workflow and understand what is the most important part to automate away.
But I would also say that a lot of the existing concerns out in the marketplace around hallucinations and AI producing inaccurate results, to a degree can be reduced through effective prompt engineering.
The idea is that, you know, knowing how to input the right text commands and how to, you know, how to ask the right follow-up questions is important.
And there are really great templates online for how to do prompt engineering right, by the way.
But it's really just like mental models layered atop one another.
So for example, you can ask the AI to adopt a persona or a certain frame of analysis.
And if you want different frames of analysis and you want a human to compare and contrast the datasets that, you know, your AI spits out, you can also do that.
And so you can put in constraints and that's what human intuition is for.
You know, you might ask for certain sources, you might ask for data that comes within a certain time constraint, or you might make sure that the inputs that you're feeding the AI are like your highest quality dataset and you're exceedingly clear about your instructions.
And so is it perfect?
It's not, but I think, you know, there's a great way towards reducing the perceived risk level through having, you know, the right human checks in place and getting really, really great at prompt engineering and figuring out how do I best use this tool and how do I feed it the highest quality data that I have access to.
To add on to that point, what's the filter factor?
What's the mitigating factor that solves for human error and artificial intelligence error?
And that is experience.
Subject matter expertise, decades of experience, you can tell, you can one, mitigate a human error because you've seen it before, you failed and learned.
And then on the artificial intelligence side, if the outcomes are misleading, let's just say, based on your wealth of experience, you can quickly mitigate that.
And so that is a potential solve or risk mitigation, both on the human side and artificial intelligence side.
So I think me personally, I think that subject matter expertise and the co-pilot model is necessary.
When I hear talk of kind of a one-man, billion-dollar brand, I see the power of the technology to automate so much of what gets done in the business, but I still don't know that you can really do it with just one person monitoring everything, because I do think that even when AI gets to new extremes of capability, it's still going to need someone to double-check it.
The calculator did not replace an accountant, nor did Excel or any of the other tools that we use to manage finance.
I would still want, at the end of the day, a human being to look over whatever the AI puts out and say, yeah, it got the numbers right.
Because we've been seeing ChatGPT in the market already making pretty egregious mistakes, both in math and elsewhere.
Google telling people that eggs can melt, for example, to use a food example.
So I know that it's still early, that these products are just hitting the market for the consumer-facing usage.
But we're still gonna have to consider, is the machine giving us the right information at the end of the day?
And that is something that I think when you're talking about a billion dollar brand level, you're not gonna wanna just allow it to work unmonitored.
Brad, what are some of your other concerns about using this kind of technology in CPG?
I know that you and I were talking about it a little bit earlier, but the impact on competition, on the job market, I think, anyone who works in R&D or marketing or supply chain optimization, I think everyone's sort of wondering what parts of the jobs that they're already doing might be replaced.
So what are some of the other concerns you have?
You know, I think it is going to go both ways.
I see places where it does ultimately allow you, like in Rivalz case, to reduce headcount, especially early on when you're trying to be lean.
I also see it simply transforming people's jobs as they do it today.
It's another tool.
When we talk about warehouse management or supply chain or doing planograms for grocery, it allows you to visualize and try new things that you may not have thought of or been able to calculate before that the machine is able to then show you.
But you still have a person on the other end.
It's evolving and changing jobs rather than necessarily eliminating them.
In other cases, it may be that, you know, you find a certain job is no longer necessary or that you can do it with fewer people on a team.
You know, have to wonder how that's going to play out in the end and I'd be curious to hear Mel and Peter, what your thoughts are on that issue.
Super happy to jump in here.
I know there's been a lot of talk in the, you know, on social media and the broader public around, does AI replace jobs?
Like, what is the system for jobs that are going to be automated away?
And I think to a great degree, there is still place for human intuition.
I think there is still place for great taste.
I think great taste and intuition comes hand in hand with knowing how to prompt the AI and work the model to get the outcomes that you most want.
And I think to a great degree, a lot of this is learnable.
It's effectively sort of a skills gap.
The folks that can figure out how to leverage AI to get better at their jobs versus the ones that are resistant to change, I think that gap will continue to grow just because the pace of AI is coming along so quickly.
And so I think people should be open to experimentation and learning.
I think we're also advantaged by the fact that, to a great degree, there's not been an information silo around AI compared to previous eras of groundbreaking consumer and enterprise technologies.
Like if we think about the personal computing revolution, for example, that happened organically among hackers and makers in Bay Area and San Francisco and tinkering in university labs.
And for the most part, the general public had no idea what was going on.
It was very, very difficult for them to get in early.
But now with AI, there's such transparency online about what's being venture funded.
There's a lot of open-source models being released online, so any amateur developer can play around.
We know what's happening.
These tools are live and ready for us to experiment with.
And I think what's so wonderful is that the internet has been this great equalizer for every business owner and every consumer to really understand what's happening at a fundamental technology level.
And so it can definitely be a scary prospect of what happens if AI takes away jobs.
But I think that's remedied by, I think as with every previous technological era or paradigm shift, just sort of being open, being adaptable, being open to learning and sort of figuring out where does this fit within my personal workflow and how can I, as a human, do really great work that is augmented by AI rather than fully replaced?
Because this taste and intuition is a deeply human thing.
At the end of the day, we're also talking about producing food and beverages.
And there's something organic there.
I mean, it is not possible to completely just digitize that.
It is, there's still a end product that must be produced.
And although you can automate the factories and the co-packers, you can have robotics that utilize the AI in order to make this streamline.
You can use tracking in order to make sure that products don't get lost anymore.
But there's still a physical product that has to get to a store that a person then has to either go to or have it delivered to them in some manner.
Me personally, I see it, I don't see it's, you know, large devastation effects on the labor markets.
What you see with advent of technologies since the beginning of time is new markets open up, new opportunities, right?
Let's just talk about food and beverage, right?
We're still got to make a product, still got to try, you know, distribute it from point A to point B, it's got to get on the shelves and it's consumer facing, right?
So like Rivalz is an example of how, you know, we're rapidly growing and growing, right?
And so AI just is helping us move faster, right?
And creating a large mode between us and strategics in developing technology.
It just helps us be faster, better, cheaper.
We'll still have to employ humans, right?
The workforce.
We're still gonna grow and employ more and more people domestically and internationally, right?
So we're actually a job creator and that's what you'll see in my prediction is as with the advancement of technology, you'll open up more markets, more markets will be discovered.
And you know, to your point Peter as well, one other thing that's going to happen is smaller brands probably are going to need to start using AI as a requirement because, should have mentioned it earlier, but one of the biggest companies in food and beverage that is using AI in an integrative sense is PepsiCo.
And they've been active with both lawmakers on trying to guide regulation about it, but they've also been using it in deep ways when it comes to their supply chain, when it comes to analyzing sales, when it comes to determining which stores are selling the most product and how to then adjust strategy around that.
So the strategics are already going all in on AI.
So it's going to be something that startups are going to probably need to invest in.
Yes, yes and no.
So strategics are going to be using it for cost savings, right?
Efficiencies, how to improve the margin, right?
How to improve sales, et cetera.
When we were talking about core entrepreneurship, what's the need and want in the market that's not being currently met by products?
Strategics are terrible.
It's based on their model, right?
You know, when you look at the status quo, that is a strategic company, right?
They're not, they don't have the mindset of, okay, what's new and novel?
What's a gap in the market that we can exploit?
They would rather do, in my opinion, they would rather do M&A to figure that out.
So what AI is great for, especially as a startup, is that you can quickly access the gap in the market, create a product design, exploit the gap, still market share, right?
And be highly efficient in being a superior company and a superior competitor.
A startup combined with AI is a very powerful thing.
One, because we can move quickly.
We're always challenging the status quo, right?
And especially if you put subject matter experts onto it and harness that powerful tool, you can move so rapidly, so quickly, before a strategic has time to think we're already there.
They do have a strategic competitive advantage in the fact that they can reduce their costs, right?
And increase their revenue, which makes them a powerful company.
But if they're still focusing on current products that aren't solving the major trends that are emerging, especially with the younger population, they still have a fundamental problem and an acute skill.
We are just about out of time.
Melissa, I would love to throw it over to you as a last comment.
What are your final thoughts on the topic?
I think we are headed in for a really, really exciting year or two, like obviously we're seeing the funding boom of 2023 and generative AI companies somewhat slow.
But I think with this comes a lot more discipline with how AI founders deploy capital.
There's more intentionality around building technology that is in service of an existing problem, as opposed to creating technology that is in search of a problem to be serving.
I think we are also entering a brand new era of multimodal models where you can now feed a model, not just text, but video, audio.
And so it means that AI is going to become a super powerful co-pilot where you don't even have to have clearly formulated thoughts of exactly what you're looking for.
It can be based on vibes, it can be a thought partner.
We're also seeing really interesting discipline happen on the enterprise side of things.
Like 2023 was the year of consumer applications.
We saw a takeoff in photo generation and next-gen search for consumers.
And I think ChatGPT was the fastest ever website to reach 100 million monthly users, which I think it did in two months.
But I think enterprise has had all this time now while consumer side innovation is heating up.
Enterprises had time to think about AI strategy and it's much clearer now.
And so I think in 2024, we're going to see AI companies that have much clearer paths to revenue generation, much clearer understanding of true utility.
And I think this is better for the ecosystem overall.
And I think investors, likewise, are going to be more thoughtful about applications and utility and what problems AI is in service of.
And I think overall, this is trending in a really positive direction for founders and consumers alike when the entire AI space becomes one discipline as a whole.
Melissa Li, Peter Barrick and Brad Avery, thank you so much for joining Community Call with this discussion.
It's been a pleasure to chat with you, and I certainly look forward to seeing how this technology evolves.
Thank you so much for joining us.
Thank you for having us.
Thank you.
That concludes another episode of the Community Call Podcast.
If you've enjoyed this show, please give us a review and follow us on Apple Podcasts or your listening platform of choice.
To join Community Call live on Zoom, go to bevnet.com/communitycall to see what's coming up and register for upcoming shows.
And don't forget to join our BevNET, NOSH, and BrewBound Slack Community at slack.bevnet.com.
Thank you.
