AI Helps Tackle Allergy Risks

Bloomberg

Watch AI Helps Tackle Allergy Risks - Bloomberg on Dailymotion

Transcript

00:00Dylan, you've got a really interesting story. I think like so many startups, it comes from a place

00:04of necessity for the founder. Talk to us a little bit about what you've dealt with.

00:09Yeah, firstly, Tim, thanks very much for having me. Great to be here. And yeah, like you mentioned,

00:14I diagnosed celiac when I was 10 years old and so have a lot of personal experience navigating

00:20dining out of home and ordering online while needing to know what's in my food. And just over

00:25a long period of time, got more and more frustrated with how difficult it was to get that information

00:30and mistakes and inaccuracies and decided to try and do something about it.

00:34So how can AI actually help restaurants do this? Because when you do look at a menu,

00:38when you do talk to a server, I feel like in this day and age, they have a good understanding of at

00:44least some of the most common allergies, gluten, for example, and people who have celiac. So what

00:51does AI allow them to take a step further? How does it do that? Yeah, it's a fair point. I think

00:56you're right. I think a lot of restaurants have got on top of gluten-free, vegan, vegetarian,

01:00the main ones. But there's 173 million Americans who have some form of food allergy or dietary

01:06requirement. And the allergens go far beyond just gluten and vegan. And so what we do is we help

01:13restaurants by ingesting their menu information, their recipe information, and the product information.

01:18And we have trained large language models to break those down to the ingredient level,

01:23tag them with the correct allergen and dietary requirements. And then we're able to power a

01:28personalized menu solution whereby consumers can see exactly what they can and can't eat on the menu,

01:34depending on their personal requirements. You have a background in law. You're a former

01:37corporate attorney. And, you know, I wonder about the liability element here. You know, mistakes happen.

01:43And mistakes get made. AI, LLMs hallucinate. Well, how do you protect around that? And how do you make

01:49sure that even if a food says it doesn't have something, it doesn't become contaminated somewhere

01:52with that ingredient process? Yeah, it's a great question. Firstly, on hallucinations, our technology

01:59never guesses. If there is, you know, it's based on structured ingredient and supplier data. If there's

02:06ever a scenario where it isn't sure, it will tag in the back end for us that it's, there's uncertainty,

02:11and our dietician team will come in over the top and do QA and manually intervene. And as, you know,

02:17they make inputs into the system, the LLM learns it gets smarter and smarter over time. From a legal

02:23liability standpoint, we would argue that not having any documentation on allergens is a much higher

02:29risk because right now you have a member of staff who's likely not trained on all the ingredients and

02:35all the allergens and all the menu items. And they're the line of protection for the restaurant

02:40between the consumer and a potentially life-threatening, a life-threatening incident.

02:45And 54% of all allergic reactions in restaurants occur after the staff have been notified. And so

02:52that tells us that the current system of dealing with this by word of mouth isn't working.

02:56So Dylan, how does it work? Is it a two-sided market where you have to get the restaurant

03:01or the restaurant chain to add your technology, but then also get people who have these allergies to use

03:07it? Yeah. So we partner with the restaurant chains, food service operators. We ingest their menu recipe

03:15product technology from various tech stacks. And then how it works typically is they put a QR code

03:21in venue on physical menus and menu boards and a digital link on their website. This is the most basic

03:27integration. And then when consumers come into the physical environment or digital environment,

03:32they scan the QR, it prompts them to create their dietary profile where they can choose from

03:37over 150 different allergens and dietary requirements. And then instantly it will show

03:42them, here's exactly what you can eat. Here's what you can eat with a modifier and what that modifier is.

03:47And here's what you can't eat and why. So it's completely personalized based on their requirements.

03:51And the consumer discovers this in the restaurant's environment.

03:55Senate Bill 68 in the state of California, this is effective next week. It's going to require

04:01major chains to provide detailed allergen info. Many people argue this is a major step toward

04:06transparency. How has that increased adoption of your product?

04:11Yeah. So just on that, it was signed by Gavin Newsom a month ago. It becomes effective 1 July 26.

04:17And so what it in essence requires is every restaurant chain and food service facility

04:22with 20 plus locations nationwide, where at least one of those is in California to label all of

04:28their physical and digital menus for the major nine food allergens. So this is obviously a major

04:34step change for restaurants. They can do it one of two ways. They can either physically annotate

04:39every one of their menu items with those allergens, or they can use a digital, like a QR code that links

04:46out to a digital allergen menu. And that's obviously what we do. And from speaking to a lot of the

04:52bigger chains recently, as you might imagine, there are strong preferences to use a digital

04:56mechanism. And so we're getting a lot more inbound than we certainly were a few months ago, which

05:01is fantastic. But we continue to work with, like I said, with independence chains, food service

05:07facilities of all types.

Category

Transcript

Be the first to comment

Recommended