In this episode of "Let's Talk Farm to Fork," we're joined by Sivam Krish from GoMicro, who we'll talk to about how they use smartphone cameras and AI to identify the health of products moving through the supply chain.

Transcript

[00:00:00] Mitchell Denton: Hi, Sivam. How are you?

[00:00:01] Sivam Krish: Very good.

[00:00:02] Mitchell Denton: Before we get into things, though, I just wanna ask you real quick, can you tell us a bit about your background, how you came to work in the foodtech industry, and maybe a fun fact about yourself?

[00:00:14] Sivam Krish: Right. I think it's um, a series of accidents really. I was involved with a high school called the Australian Maths and Science School, where they're trying to figure out how to engage kids in what's called STEM education. Um, the reality is that STEM education is really boring. 40 percent of kids are just bored with this thing, what we call STEM education. One of the things that they are not bored with is their phones and so we had various ways of teaching physics with phones, but what, what really piqued their interest is just looking at things with, um, you know, magnified using a clip on microscope. So that's how we got started. That's the history behind, uh, Go Micro.
Kids, of course, loved it. But it had no future because, uh, we didn't really fit in well into the curriculum. So what I've learned out of that is sometimes there's a problem. Even if you have solutions, it doesn't mean it goes anywhere. So that's how we started, uh, and then we found that, um, you know, the Department of Agriculture here bought, bought 200 of these devices, which are simple clip on microscopes for phones that we were making. And that brought us into agriculture. And we discovered that farmers are sending, um, you know, really bad images of insects and so on, wanting to know what they are. So that's why they bought them when they distributed them for free. So, we got into agriculture. Then, of course, there's a pest. The problem in South Australia, it's a fruit fly free state, a lot of fruit flies running around and I got interested in AI and we found that, ah, with this phone microscope you can actually get clear pictures of fruit flies.
The problem with fruit flies is that they don't, they're difficult to photograph. So we got a ping pong ball and cut a small hole and put the fruit fly so we can shake it around and take pictures. And we found that, you know, instead of requiring thousands of images, we could manage with about 50 or 60 we were getting stellar accuracies.
It took us a year to find out why, because we had changed the lighting condition in a way that's easy for AI to understand. So that was a breakthrough. And that is how Go Micro was born from, um, you know, a series of accidents.
[00:02:39] Mitchell Denton: yeah. Okay. And do you have any, uh, fun facts up your sleeve for us?
[00:02:45] Sivam Krish: Well, the fun fact is that, you know, sometimes you think there's a big problem, um, and you solve it, and then nobody wants it. That's the fun fact that I catch, yeah. So we crack the education problem, we crack the footprint problem, then we discover nobody really wants it.
[00:03:02] Mitchell Denton: No. Yeah. Okay. Well, you gave us a, a little bit of a history lesson there, um, but I was wondering if you could expand a little bit more on how the. Uh, AI side of things. Sorry, just gonna take that again. You, you gave us a little bit of a, uh, history lesson there on Go Micro. I was wondering if you could expand a little bit more on how the AI is being utilized to accurately assess the quality of products moving through the supply chain.
Yeah,
[00:03:32] Sivam Krish: So, we were fond of this because we started developing this for kids. Very simple microscope attachment. So, once you put it on the phone, you can then access it all along the supply chain. Is really the connective part is really what's missing in the in the agnivalue chain that say somebody at some point says this is this quality or this rightness.
There's no way to coordinate to what happens before. So putting it on the phone, um, gives us terrific insights. The second thing, because of this front fly, we have to magnify the image. When you magnify it, you get a lot more relevant information, like your skin is very different from a kid's skin. So you can tell the age of a person by looking at how wrinkled the skins are.
And exactly the same thing with fruits and vegetables. So you get really deep insights into using really cheap technology into the real state of how you produce, uh, just using phones.
[00:04:31] Mitchell Denton: okay. Okay. So then, which categories of food products can Go Micro identify? And are there future categories that you're gonna expand into?
[00:04:42] Sivam Krish: Right. So the interesting thing about AI is that. You know, AI doesn't really know what it's looking at. It's just identifying things based on training. So anything you can assess by eye can now be assessed by AI. So with Go Micro, we are able to assess things that you can't assess by eye. Like right now in Indonesia, we're doing a project to assess the freshness of tuna.
So by looking at the eye, we can tell when the tuna was caught. Uh, this is something you can't do by eye, by eye. So it's a generic technology, it's incredibly large range of applications, uh, in the everywhere new chain.
[00:05:22] Mitchell Denton: Yeah. Okay. Um, so then what are some of the challenges that the team at GoMicro have had to face when developing your AI technology?
[00:05:34] Sivam Krish: Well, the development was really not the challenge. The challenge is on the marketing part and, and bringing in a technology to an industry that is highly regulated, uh, very slow to end up and seasonal. So the challenge was really more the marketing. Then on the technology, we've had this technology, we detected fruit flies three years ago, uh, with AI.
Um, it's taken us a very long time to get to market and that's really challenging part, particularly because, um, you know, you need images. So we, we found it very difficult to get images of all the grain types, different grain types. We found it virtually impossible. to get images of, um, invasive pests because we can easily detect them, but where do I get images of invasive pests?
So there were challenges in the data side, uh, and challenges in the market side. Okay.
[00:06:26] Mitchell Denton: right. Okay. So then can you share any success stories or case studies from customers who have utilized GoMicro for their product inspections?
[00:06:36] Sivam Krish: Sure. So right now in Australia, there is, uh, PB Agri Food in Toowoomba. Uh, they buy soy from farmers. They've been using it for one and a half years now to assess the quality of soy. So when farmers come in, they can assess immediately and, and essentially decide what, what price to pay, what quality it is.
So that's a long, long project for us.
[00:07:04] Mitchell Denton: What have you found to be the biggest surprise while working in food tech?
[00:07:11] Sivam Krish: Okay. The biggest surprises are. You know, there are all these problems about food waste, all this publicity about food waste. There are all these announcement of problems of pests and so on and so forth. The surprising thing is when you have a solution, the people who talk about it are not interested.
Because agriculture is full of problems that nobody really owns. It's like global warming.
Uh, I guess it's called the problem of the commons. Nobody takes ownership of, of any problem. So, so, so technology solutions, um, alone won't be sufficient to solve some of the problems that we are facing in agriculture.
[00:07:57] Mitchell Denton: Yeah, okay. Okay.
[00:07:59] Sivam Krish: a hard problem to crack.
[00:08:02] Mitchell Denton: No, definitely. So then, taking a step away from the solutions that Go Micro is providing to the food industry, what in your opinion represents one of the main challenges or pain points in the fight against food loss and waste? The end of
[00:08:18] Sivam Krish: so the question is, it's not the pain, they ask the pain. So as you know, there are a lot of controversy in Australia about supermarket prices and consumer dissatisfaction with those pricing. the person who pays the price is a farmer, right? So, we've got to figure out how to reduce that pain, which can be reduced if you can reduce the waste. But there's a power structure in a purchase chain, and the farmer is a price taker. So, what's really variable in the whole agri value chain is quality, right? The cost of transportation is fixed. The numbers you can count, the size you can count, the weight you can count. So the only place where something can be pushed back is based on quality.
Now, I believe in a lot of cases, there's no incentive to make it objective. if you keep it subjective, you can always tell the farmer, your fruits are alright, or they are not good. So, there's no incentive to essentially make quality assessment objective. And I think if you make it objective, you know, you can solve a lot of wastage problems and you can pass that advantage to the consumer.
[00:09:47] Mitchell Denton: Yeah. Absolutely. So then, how do you see AI being further utilized in the food tech industry in the years to come?
[00:09:58] Sivam Krish: Well, I think we got to, we are slowly figuring out who the beneficiary is, right? So, farmers ultimately benefit by farm and food chains will benefit by. Transparency. But right now, the food chain is segregated, is broken into bits. And at each point, there's quality check, right? And at each point, there's subjectivity. And if you remove all this subjectivity, then the food chain will benefit. So there are some players in the chain, for example, who grow their own food and take it to the consumer. Now they have a real interest in, in reducing all the subjectivity, but all the people in between who dominate the industry have, have no benefit in reducing the subjectivity.
So I think we've got to figure out, you know, where does it fit? So consumers would love to know, you know, how long a fruit will last, whether it's ripe or not ripe. So there are possibilities on the consumer side of things, and there are possibilities in the chain. And I think end of the day, um, if you look at what's happening, uh, farmers are beginning to trade directly with consumers or with buyers. So if you can remove the subjectivity in that purchase chain, I think that Sector of direct sales from farmers to end by will bloom. And I think AI will make that possible.
[00:11:25] Mitchell Denton: Yeah, okay. So then, is there a particular group or innovation within the food industry, whether it be AI or another field entirely? That you're excitedly keeping a watchful eye on.
[00:11:37] Sivam Krish: Absolutely. So, you know, there's blockchain obviously, uh, there's been a lot of hype about blockchain. It hasn't really taken off the way it is, but there is now a means of, of creating an immutable record, right? So, I think blockchain is interesting because it can give, carry a record from farm, uh, to all the way to the consumer.
The other interesting technology is digital twins. So we can now model, um, you know, the state of, now it's interesting that. You know, AI is able to model the entire human language, right? When I speak one word, you can predict the next word, but we can't do that with fruits and vegetables, you know, you don't know what's going to happen the next day, nobody knows,
[00:12:23] Mitchell Denton: Yeah,
[00:12:24] Sivam Krish: So it's ridiculous. So I think modern technology. Uh, or sometimes people call it digital twins will allow to build predictive modules of how agri produce, um, you know, fares along the supply chain. So combined with that and blockchain, there is an incredible possibility that, uh, we are after, which is the ability to make, produce into a digital asset from the time it's harvested.
So, so for example, right now we can assess, um, grain quality. On video, which means we can put it into harvesting machines, which means a farmer will have a digital record of his entire truck of grain, not sampling the entire truck, and that can be put on a blockchain, and it becomes a digital asset. Now for fruits and vegetables, you can also have information about temperature, humidity, transport conditions.
And you can make agri produce into digital goods with very good predictive capability. This will take out all the uncertainty, reduce the insurance cost, reduce the waste. And I think you'll have a huge impact in the food industry if it's done. So there are some technologies coming together very nicely, which I think will help to crack this problem.
But the bigger problem is nobody really benefits by cracking it.
[00:13:56] Mitchell Denton: absolutely. So then, what advice would you give to entrepreneurs looking to start a company in the food tech space? Eh,
[00:14:05] Sivam Krish: Don't be stupid. Don't buy
into the thing that, you know, everybody just because everybody is saying there's a problem that if you come up with a solution, they'll want it. Um, that, that is, uh, extreme naivety we suffer from, but no wrong. Right.
[00:14:26] Mitchell Denton: we are coming to a close on the episode, but before we do, I just wanted to ask, what is the major point you want the listeners to take away from this episode?
[00:14:37] Sivam Krish: I think the technology is to solve access quantities there. I think the technology is to solve the food waste problem is there. And I think every sector in the player, uh, will benefit by being open to these technologies and being early adopters of this technology.
[00:14:55] Mitchell Denton: Yeah. Yeah, great. Well, that's all for today's episode of Let's Talk Farm to Fork. Thanks for listening, and thank you, Sivim, for listening, uh, for, I just ruined that close up. I'm gonna try that again. Ahem. Well, that's all for today's episode of Let's Talk Farm to Fork. Thanks for listening, and thank you, Sivim, for joining us.
[00:15:17] Sivam Krish: My pleasure. Thank you for chatting.