Ep 24 - Artificial Intelligence (AI) in Brewing - Ankur Napa

Is AI at the forefront of the next industrial revolution? And how does that affect the brewing industry? 

With me is Ankur Napa who has worked for ABInBev, SABMiller & United Breweries as a brewer and senior data analyst. 

In this episode we go cover: 

• What is Artificial Intelligence (AI)
• What current tools and systems can it replace in brewing? 
• How will this affect brewers/breweries? 
• Real-life examples of AI in brewing and how they impacted the brewery
• Future of AI in Brewing 
• Can it replace brewers themselves? 
• Ethics and data privacy of AI
• How to integrate AI into your brewery
• What does Ankur think AI should be used for in the brewery

Chapters (written by AI)

00:00 Introduction to the Modern Brewer Podcast
00:22 The Evolution of Technology and AI
01:36 AI in Brewing: Potential and Concerns
02:07 Introducing Brew Resourceful
04:29 Guest Introduction: Ankur Napa
05:23 Ankur's Journey in the Brewing Industry
10:19 Understanding AI and Data Science
18:57 AI's Impact on Brewing Processes
22:43 Future of AI in Brewing
28:42 Real-World Applications and Challenges
33:42 The Future of Automated Brewing
34:03 Can AI Replace Brewers?
34:34 The Role of AI in Brewing Decisions
38:02 Challenges and Limitations of AI in Craft Breweries
42:00 Ethics and Data Privacy in AI
47:19 Integrating AI in Breweries: Practical Steps
55:53 Sustainability and AI in Brewing
01:01:06 Conclusion and Final Thoughts

🔗 Chris' LinkedIn

📨 Register interest in Brew Resourceful's new AI-powered brewhouse efficiency monitoring system

🔗 Ankur's LinkedIn

✉️ Ankur's Email: napaankur@gmail.com

🤖 IntelligentX AI brewed beer - sadly has closed, they were the definition of ahead of their time. Check out the promo video made 8 years ago. Proof of AI life before chat GPT https://vimeo.com/172395607

Transcript

0:14

Hello everyone and welcome to the latest episode of the Modern Brewer Podcast with me, your host, Chris Lewington. The world changed when we entered the first industrial revolution, when humans first built machines powered by kinetic movements, steam and water that were capable of doing humans manual work more efficiently than ever before. Then we discovered electricity, propelling these machines into capabilities far beyond their steam and water power predecessors. At this time, we also started advanced communications, which started a globalization of our world. we invented machines capable of processing complex calculations. storing large amounts of data and digitalizing actions. And during this time, one of the most important people of our species, Tim Berners Lee, discovered a way to connect these together, creating a worldwide infrastructure of processing power far beyond the capabilities of the very people who created them. And now we are living in a world where artificial intelligence, or AI, is becoming the latest significant step towards a new world. AI's being hyped beyond belief. Even my PDF viewer boasts a new AI tool. How does AI fit into our brewing world? Can it create beers for us? Recipes? Can it enhance our production efficiency and sustainability? Has it already been doing this for years without the AI label? And I guess for some, there is a concern that ultimately can it replace brewers themselves or worse part time brewing podcast hosts? Before I lose my job to ai, I'd like to use its power to enhance brewhouse efficiency. So Brew Resourceful is actually on the cusp of doing just that. After founding my company Brew Resourceful, I quickly notice that the majority of breweries do not measure brewhouse efficiency for every batch. they don't even store their brewhouse data digitally in a way that can be comprehensively analyzed. So with a small team at Brew Resourceful, we are very close to releasing a brewhouse efficiency monitoring system for craft breweries that will database All of your brewery's brew house data from mashing temperature all the way through to final casting out volume and gravity. Your data will then be visualized in a dashboard that is designed by brewers. And all you need to do as a brewer in this system is upload a picture of your paper brew sheet. I know how much brewers love paper and when you're jumping around the brew house, it's important to have something that's quite mobile, and not every brewery has access to tablets or phones, to hand out to their brewers. So I understand paper is being used, but with this tool, all you have to do is upload a picture of that brew sheet, and everything else is handled by our good friend AI. If you've already got a digital way of recording, this can be integrated as well, and you can forgo the need to upload the paper brew sheet. But you still get access to the powerful dashboard and analytics. So now brewers can circumnavigate the consuming task of creating, storing, and visualizing their brewhouse data. On top of this, I will personally give each subscribing brewery a 30 minute analysis of your data with the ultimate goal of increasing your brewhouse efficiency through data driven decisions. We are working on this right now at the time of recording. And if you are interested, there is a link in the show notes where you can register your interest. Early adopters will receive their first month for free and discounts after the first month has ended. Okay. So back to the show. Now let's learn if AI is simply a marketing tool. Or if it can genuinely change the world we live in, in the same way that steam, water, powered machines, electricity, computers, and the internet did. With me today is Ankur Napa, an experienced brewer and senior data analyst with over a decade of diverse experience in the brewing industry. Ankur has worked with some of the biggest names in brewing, including AB InBev, Saab Miller, United Breweries, and various craft beer brands. During his career, he led major projects, developed 12 unique beers for the Indian market, and introduced innovative digital tools to improve brewing processes. So, welcome to the show, Ankur. Thanks, Chris, No problem, man. Yeah. I'm really excited for this one. Um, I think it's one of those really hot topics in the globe right now. So it's really cool to be able to put that onto the, put that onto the brewing world. But before we start, do you want to introduce yourself to the listeners, who you are and sort of what you, what you've been doing? Yes, absolutely. So I'm into this, journey, into the brewing industry. 10, 10 years back. So I did my B. Tech in biotechnology and then did my masters in brewing sciences, wine brewing and alcohol technology from India. And, From there, I got into the brewing industry through my campus placement. So Sam Miller took me in, in, in those days. And, I was working there like graduate trainee. They put me into every aspect of brewing from, I can say from brewhouse to brewing quality filtration, seller management, wastewater treatment, utility packaging line every, everywhere, even, They also put me into sales and marketing where we were visiting every point of consumption or, or I can say different sort of, be a sales point for, for three months. So I was, traveling from like shop to shop in India in those days. So that's how my, my journey started. And then, we all know acquisition happened of Sam Miller by AB InBev. So. I moved there and then from there I moved to one Anglo Indian brand called Whitlinger. So it was a craft beer brand. So initially they were brewing in Cardiff, UK, and then they started making those beers in India. So that's how they hired me. And then my job was making similar beer in India, which they were initially doing the contract manufacturing with, A brewer in the UK and then I moved to India's a giant brewer, which is known as a United beverage. The brand is Kingfisher. So there I worked on a stabilization of few beers. I cannot name them. That's right. Plus, I worked on like craft beer section for, for them. And, there, there, my job was basically as a R and D brewer developing a recipe. Or understanding the data, understanding the consumer behavior, what consumer wants, and interpreting the same thing into a recipe. So, this is what I've done. After that, during working in that time, I got interested into data science because I was learning those things by myself. Using various coding things, data visualization tools, various technologies that were available in that time. And, later I ended up getting enrolled into master of data science from a Liverpool Johns university. Thanks to COVID, I did that through distant education and it was very cost effective for me. Yeah. Yeah. It's, it's, it was like that. And, then, from there, I got into a company which was working with AB InBev. So AB InBev has also their headquarter of AI and data analytics in India, Bangalore. So this company was, so ABinBev was outsourcing their work to this company and, that's how they got me in because I had data science experience or capability plus growing knowledge. So I got into this project later, ABinBev took me into their department of data science called commercial planning and analytics. Yes. So during my journey, like I, I work with like various tech teams off of the leading breweries collaborated with. I can say lots of data folks, including data engineers, data scientists, data analysts, to develop various AI driven solutions. And after seeing those results after, achievement of that solution or, or the product in, in the company that, solidified my belief in the transform, transformation of. transformative power of AI. And yes, so since then I'm like dedicated my life or my passion to like integrating AI into peer industry. Well that my, you are the perfect, candidate to be on this, on this podcast, because I've seen what you've been doing. It looks really cool. and it's a pretty amazing experience there, Ankur. so let's just, not everyone is going to be as well versed in AI or artificial intelligence as yourself. So perhaps you can just give us like a good explanation of what AI actually is. very tough question, to convert very typical, technology into simple words, but I'll, I'll try my best. And see AI is a buzzword and everyone does, or everyone thinks, okay, AI means robots. But the reality is AI is a technology which does repetitive task. Which human cannot. So let's, let's make one rule in this podcast, AI is equal to data science I simply divide data science into three parts. So, early days, AI or data science was meant to be like, Rule based thing. Let's say if someone is, giving me a rule. Okay. If pH is 5. 4 for the mash, then it's perfect pH. If it is less than that low extraction, So that's how AI or these rule based codes used to interpret things. Then we have machine learning. Machine learning was the thing where computers started learning from the data. Let's say, If I have a one brewer who is recording mash pH on daily basis, one day he forgot. So what machine learning started doing was it took mean of that. So mean is this is machine learning. Then, there was a thing, which another thing is deep learning. So deep learning is like our brain, computer has, we have neurons, right. Which are like, they think they take decisions, multiple decisions. And then if I have to pick something, so my mind takes several decisions, okay, this, this, pen is lying there and immediately I do these things. Similarly, computer has those neurons like humans and, then they take multiple decisions. And then tell us the things. So if, if I have to convert these things into like daily basis, examples, please, Krish, I'll, I'll be happy to explain them. Yeah. I mean, it's firstly, it's, it's really fascinating to hear you say that. So AI is actually just something, as you said, it's equal to data science, which has been around for a lot longer than this buzzword AI has. that being said, it is a pretty good buzzword. but you're saying now really the major change that we've seen, I would guess over the last few years, I'm not the expert, but in the last few years, it feels to me from the outside looking in that there's been a Significant change in the way that AI has evolved or data science has hmm. Mm that is the deep learning. Is that really the key change that's happened in evolution of, hmm. Yes, yes, yes. deep learning is the chart GPT, I can say. Perfect. There you go. That's it. Putting into more common people terms. Certainly know what chat GPT is. Use it a whole bunch. So yeah, that, that was really the big, the next, advancement, so to speak. I get that. That's really cool. So yeah, I mean, if you could share some examples of what early days, data science slash AI, the machine learning and deep learning really is, that'd be awesome. So do you listen to music or podcasts? So your favorite platform or accessible platform is Spotify, right? Of course, that's where the majority of the listeners, listen to this podcast on. Makes sense. So, so in Spotify, if I'll relate this, rule based thing as, I only want to listen to your podcast. So I'll, I'll search by search by your name, the rule is already decided. It will not give me any other podcasts, only podcasts made by you. You are the artist. That's it. Then machine learning machine learning is, today, I'm in mood, to listen, similar podcasts like yours. So it will tell me deep learning is after understanding my behavior, That this guy comes on Spotify, listens to these kind of podcasts, like in YouTube, right? Suddenly some video pops up and, we're like, Hey, I like this stuff. Like in Netflix, if we have specific history of watching some movies immediately, it recommends you may like this as well. So that is the deep learning or I can say machine learning. And if we comprise whole concept, that is artificial intelligence. But it's data science because everyone is generating data on various platforms and computer is sitting behind that data, making decisions. The ad that I see on my YouTube, probably you will not see. hmm. Mm hmm. And in my YouTube, because I'm more into data analytics. So every time it pops up, Hey, learn data analytics. This is our course. We are selling this. Because YouTube understand this guy only watches this kind of stuff. He would take action on this thing or he'll get trapped into that funnel. Similarly, so these are the real life examples. Any, any example, Chris, from, from your life? yeah, I mean, I would, I mean, the personalized ads, they've been around for a while, right? I guess is that what you're saying by machine learning? That's big data and being able to analyze big data. Mm because if you search something on Google, it knows the trends of the type of person you are. It's more likely to pump certain content your way. hmm. Mm hmm. I mean, that's been around for a good amount of time now with Google ads, Instagram ads, et cetera. but yeah, the deep learning and I have, I don't really know, apart from obviously chat GPT, which is really the flagship and, I have. An AI note taker, I'm going to use whatever I'm, I'm now afraid to use the word AI because I feel like I don't know what I'm talking about, but I'm going to use AI because it's a great buzzword. I have an AI note taker that I use for day to day work and it records and transcripts the meeting, but then you can also ask it questions and be like, what did, Mm hmm. five takeaways from this and it can now. It will give me the list of the five takeaways or what was the tone of the meeting? Which I think is a really fascinating one and it'll be able to It'll be able to analyze the tone hmm. Mm was said and how it was said, So it'd be like that was a it was a positive call but there was overall concerns about x or it was, an educational call which where Person a was delivering hmm. Mm hmm. b and I find that is totally fascinating to me how it can now Exactly what you said. I mean, it's just, it's now moving on from just simple. I say simple, but, from like just relaying information to be able to interpret information and especially things like tone of voice, I find Mm hmm. As you said, I guess it's all data science at the end, but it's the way that it's presented to you is quite fascinating these days, but I guess there'll be the examples that I see my day to day, Yeah, for the brewers, what I can break things into, like, concluding the basics of AIS. If, you are able to predict a missing value in your data using, I can say, average or mean. It's data science. It's machine learning. If you're, if, if there is a pH of, 5. 5, 5. 5, 5. 4, 5. 4. And suddenly there is a pH, which is seven, you'll be taking mode of median. That's it. That is data science. but now you could say, What is the optimal? PH, and now it can tell you which is the optimal based on the rest of the data, correct? That's really where the big change is. Yes. No, no, you are coming in my court Nah, yeah, yeah, it's actually, it's a really nice thing to lead on to, but like how, I mean, you have a lot more experience in this than I do, although I'm working on something with it, but how is AI or, the next, let's say deep learning and the next phase of AI. specifically transforming the brewing industry. Yeah. Mm before, before this, like I can, compare the traditional, methods of the brewing industry with the modern so that brewers can relate to this. initially all brewers like William D. Gossett, who developed a student tea test. He was working for, Guinness brewery. He is my role model and inspiration. I always say he's the first data analyst of the brewing industry. He was statistician as well as the brew master. Yes. Going back to those days in his days, 1800 centuries. Time we used to have mercury or alcohol thermometer and we used to they used to take temperature noted down on paper These days we have RTD sensors still still we do this thermometer, but most of the breweries have RTD sensors That's how they take So yes, that is also artificial intelligence or AI, I can say. Then, earlier, they were like noting down the time on paper. These days we have time and temperature sensors. Similarly, we have, we used to have glass hydrometer, still we use, but these days, few breweries who can, who have good financial background or have capability, they are getting those instruments. the, the personal, my favorite choice or like vast majority in, in India, that is the Anton bar. They give those densitometers, easy dense, you have heard, and then, tilt hydrometer, which is digital hydrometer floats in your, whistle and it gives you real time. So that's how things have changed. And, without, understanding the technology, how it had shifted. We, I'm just breaking down the things, how things have changed. Still without knowing we are using AI and machine learning in our blue houses. Then, as I explained, we used to have papers. Now everyone is like, Give me data in Google sheet or hmm. Sunbrew is still in paper world. That's for sure. But, yeah, yeah, definitely that would be a big, big change, right? From paper to digital. yes, but yes, see, I, I personally love to note down things on paper, but the thing is, if I have to enhance my capabilities, I have to see what is the comparison between this batch versus my, let's say this is my 1500, 1501 group. And I want to compare this with my third group. Literally I have to tear off those pages and. Put it in front of me, but if I have Excel data maintain, I can do that. So that's how still, still we are using these things, but we don't know that we are using AI and data science and our day to day things. So these are the things that I can say that's how brewing industry is transforming and we are using. Future is making beers ultra personalized. Let's say, in one part of Germany, a certain age group likes certain flavors. And they talk about certain types of beer on their social media, listening to their social media, we can, or, or we can give this sentimental analysis to the beer brewer. Let's say those particular age group is talking about, we love beer. If it is too sweet, this is directly relating to the final gravity of beer, Yeah. Oh my God. right? Anka, you're beginning to slowly blow my mind. I've just seen the possibilities of this. So yeah, you could have, it now can communicate down the whole stream, right? So it's like, okay. So people of X. Generation X, country of origin or location are talking about flavors and things that, or even liking certain things that have, a known flavor or known sweetness or not sweet or healthy. And now that can translate all the way down into, into basically dictating what you need to do to even achieve that. Not just, not just that information, which on its, in its own is big information, right? It can now even tell you. Okay, well now the FG of the latest beer is going to be, if you want to appeal to this market, needs to be this. Yes. See. The, this technology is, is a bridge between brewers and the consumer, or I can say business and consumer. And, this bridge is enabling both of us meeting together and making good selling beer. My personal definition of a goods beer is a beer which sells very fast. Even if it has diacetyl, because consumer is liking it. Consumer is, I can say, not a brewer. Consumer is the god, I can say. They, they know what to drink on, on repetitive basis. We, say diacetyl is bad, I've seen people having Irish stout with popcorn. Yeah, it's this is where, well, this will be a very deep subject of what, what really is, who is right, the brewer or the consumer, because that's where craft beer kind of differentiated itself from saying, no, this is actually not what is great. What you've been drinking is actually not great. This is what's good. And so therefore brewers became the people who, to tell The public, what was good and what was not. And that's what you kind of trusted your local brewer to do in a sense. it changes when you're on a bigger scale, of, of of what is appealing to people is now you have. minimum sales targets you have, growing production, growing costs. So now you need to sell more and that's kind of how craft beer in a way slips into that mainstream consumer led, even if the consumer, I mean, I don't know. What is that? Is that necessarily a bad thing? Absolutely not. but if people generally don't, yeah, it's funny. It's a whole different podcast that it's like, what is, what is craft beer? Who is right? The consumer or the brewer. I think it's a really cool, I actually agree with you. And I think a lot of my. Time spent working in brewing in the back end was more. I was actually very vocal about, being consumer led as a, as a, as a recipe designer in a brewery. but, on reflection of that quite very strong opinion I had, on reflection that did only really appeal to me. appeal to breweries with the goal to sell. Some breweries have the goal to make high quality edu I don't want to call them educational, it sounds a bit, I can't say the word what I want to say on, on the podcast, but, yeah, it, it, it's different. It's a different way of approaching the same thing, and I was probably a bit wrong in my vocal, opinion of that, but I, I do actually agree with you that it is consumer led and drinkability and, and repeatability of drinking is, is a very important metric for, what I would define as successful recipe. Yes. So, yes. So whole, whole thing is like, converting consumers, feedback or his likings into beer recipe is a black box for us. Mmm. So I'm not saying this whole, data analytics and AI will replace brewer, but it will help us to understand what I should put in my beer. If he's saying sweet, specific gravity, if I, what I'm assuming as a brewer is okay, age group, right? But in that age group, people are from different authenticity. They have different days. So that's why I said future will be ultra personalized. Mmm. Yeah, that's, yeah, ultra personalised, yeah, because it's feeding from your very, your very yeah, even I was reading one article, there is a company called Intelligent X, they are already doing this. With beer! yeah, with beer, you have to pay them some, some amount, they'll ask you a list of questions to understand your personality and likings. And after 20 or 30 days or a set period, they'll send a case of beer at your home, which will be made only for you. I've, I have like a thousand more questions, but that is a really cool concept, I guess, though, definitely, I've never even heard of that out there, I'm gonna be googling that. if it didn't make noise, I'd be doing it right now, but I'll do it the second this podcast recording finishes. Intelligence X. Do what? I'll put it in the show notes as well. So people don't have to Google it themselves. yeah, that, that's a really fun concept. I'm definitely going to look into that. Yes. So is there anything else that the, I mean, obviously ultra personalization is a really cool, future for AI and brewing, but how else can it impact? see if I already know that because throughout my life, I was making beers for mass. consumption. I did craft brewing, but for less than I can say two, three years. So in India, craft brewing means you're making beers only for a thousand people in any or five kilometers of radius. But while working with giants, I was making beer for whole India. And, if, if, if I have this tool or a solution immediately, I can save lots of money for my business. And on more than that, Let's assume beer is made. It's not selling. We have to recall that product. We have to drain it. Sustainability. We have put steam, electricity, resources, grain, herbs, everything, and it's of no use. But if I have a data driven analysis, which is backing up my intuition, that, that makes lots of sense. Not even in future or not down the line, I can say five years, it will impact for like ages. Yeah, I get that as well. I mean, it definitely been a part of breweries where short dated stock turned into unsaleable stock, nearly always on specials. Hmm. ultimately didn't sell for whatever reason. so yeah, that's an interesting point actually that it would help reduce waste if you, if you were to do that way. what about the brewing process itself? How can AI, what's the future of AI in the brewing process itself? So I can, relate or I can tell a story about this from my personal experience, if you allow. So recently there is a brewer, whose problem was, the, I can say the, I, what is that? secondary fermentation was happening in, in their final beers and they were not knowing it. Also when they were delivering these beers to the point of consumption box. There was a temperature fluctuation. India is hot country. So, you can say average temperature in India we see in summers is, these days it's 30, 35 degrees Celsius. And imagine as Keg is traveling in Bangalore, you can Google what Bangalore traffic is like. those BAs were getting lots of, I can say, bad, treatment during the transportation. And then they were not knowing it. I just put one, sensor. It is a temperature sensor. And it also tells the specific gravity of beer, though there will be very minute changes because of secondary fermentation, but it can detect and it was transmitting that data to us in real time. So it didn't give us the immediate result. It's still in the process, but it's, it's helping us a lot. Let's say we have a paid to guy who has a refrigerator in his van and he's transferring those beers there. And in the middle, something is happening or he's not switching on his AC because of so many things to save cost or something like that. Inject it, it gives me a signal or it gives me a message in my phone. Hey, temperature went up. Or, or, specific gravity is down by 0. 002. Fermentation started because those beers are unpasteurized. So, yes, this, this is the future. Other thing is, Time. Temperature is enemy of beer. If I'll put another sensor on keg, how many days it's sitting outside that point of consumption because we can geotag, right? In my phone app, I can see in Google map in same vicinity or in same bar or restaurant where this keg was lying for how many time or days or how many hours. That is the future and, then, if you use MASH filter or people who have already used Miura MASH filter. In my experience, when I was working with Miura MASH filter, it used to give me pop ups. Hey, your speedrun gravity is this. Do you want me to put next, sparging water volume this? Or not right being a brewer. I was afraid because at that time I was not introduced to AI. Neura gave us those things and I used to calculate those things on my Excel sheet or with, with help of calculator. But this will be the future immediately. You just have to click. Okay. These days people are doing clicking the okay, but in near future, it will happen automatically. So it will be, I mean, yeah, I can completely see based on what we've been talking about today, how, yeah, that deep learning aspect of AI would be able to basically make those. Brewing decisions that brewers make, it does lead me on to a, an interesting question, Ankur, what is there in the future? Do you think it's possible? And do you think it's probable that AI could replace brewers? five years. I never replaced any brewer in my past five years of career, though. I'm working brilliant piece of technologies or solution. You already know I'm working on. It enhances your capability because. A brewery hires a brewer to make good selling beer. Again, good selling beer is underlined, which, right, good selling beer. And brewer's job is not to maintain data, do machine learning and deep learning. His job is to make good selling beer, take decision, hey, my diacetyl rest. It's coming, and I have to switch it on or off or, or I have to increase the temperature by in, in next eight hours or in next one hour, because that will save lots of electricity, right? Or it will also impact the quality of the beer. As of now, the brewer is in black box. He like after eight hours, I'll do this in a window of eight hours or 10 hours. But if AI is there, it's helping him, okay. AI will be telling, okay, as per my understanding, the way you are getting YouTube suggestions or Spotify suggestions, your solution will tell, or AI will tell, it is coming on, 47th hour. And it will be telling, do you want me to switch off, the refrigeration or increase the temperature of your fermenter by that time? It'll click. Okay. When 47th hour will come, it'll happen. Meanwhile, you'll be having chemical analysis, everything. So do you think, AI can replace brewers? Chris I don't really know. It sounds like yes, to be perfectly honest, but I really, I think it wouldn't be the first time that we've over fantasized about what technology can do. And ultimately the human element has been very hard to, you to replace, I think, the first time we've heard that machines are replacing brewers, it's interesting. One of some of the things I'm most proud of in our team's, our team achievements in previous breweries has been things that I'm pretty confident AI would be able to do in a matter of seconds, which is quite terrifying. from, because I'm like, okay, that was like four people's work, for a long time. And one of them was actually something you just talked about was just automated ferment fermentation. So. We managed to dial in processes and understand the key parameters that were affecting the length of fermentation. So it meant that we were scheduling our chillers. We had an automated brew house. So it helps a lot and a semi automated setup. It meant we could schedule the chillers, To chill after we already know the daffodil rested or basically everything automated. So when it hit primary fermentation, when it finished, we would change the temperatures when the beer was needs to be warmed up for dry hop or dropped for yeast harvest and the cone cooling needs to come on. That was all automated. We didn't that wasn't operators or brewers putting in. Yes, we're ready to chill. Everything happened on the hour. based off that and we had a lot of success with that reduced our overall tank capacity Got complete repeatability from the brands. It was awesome But I think about that now now when you were saying that I was thinking about how quickly Ai would be able to do that Now, if you had the access to the same information that we had, and that was pretty, yeah, that's pretty daunting. However, that was also in a state of the art brewery where everything was based off sensors and automation, all of the inputs. and all of the data was already there. Now that doesn't exist in the majority of craft breweries. so until the affordability of automation comes down, I can't see how, it would ultimately replace brewers. if, let's say in 10 years time, can I imagine a brewery or a concept of a brewery where the entire process is automated and you have exactly, do as you said, exactly that sort of like. feeding in from sales and marketing coming in, the data from sales and marketing saying generate this, the X amount of this type of beer, because it's going to go appeal to X market, then yes, you could automate that in a brewery, but you could probably have a good go at automating that now, with very little. human intervention and actually you've seen it in the big brewers you've worked at they have very few brewers producing significant volumes of beer. So I can't see that. It's not impossible to replace brewers in in the right environment, but I I personally don't want to I don't want to feel naive About it. I I do feel in the small craft brewery until automation becomes more readily available and affordable for people Those manual kits with people operating them is always going to be the way that You craft beer, quote unquote is, is made is my total uneducated opinion anchor. Hmm. if you remember I mentioned, earlier that. William D. Gossett in 18th century was doing some data analytics. So he made one algorithm. You can even Google it. Student t test is developed by him or invented by him, which is a statistical method. He did a study. How atmosphere or, or the weather impacts the sale of beer in those days without using High end computers that we have these days. These days it's very cheap and affordable Still 18th century till now brewers were not advocates and I always say if Chachapiti 5, 10 or Chachapiti 100 will start drinking beer Then we can think of because beer is made by humans consumed by humans, right? and My PLC SCADA system or my mash kettle Cannot replace me or my my glassware cannot replace human glass will not tell hey beer is good drink it hypothetically that can also Be a use case if you have a particular glass It has all those algorithms built in and it has your history wherever you go use this, that glass immediately that glass will tell you the spear is good or bad or it when it can give you the reviews or if you are handing off this glass to your friend. yeah. And I guess ultimately with process, which is what, or production, which is what brewing is. Anything that creates data that anything that creates data in theory can be automated fully without human intervention, especially now it can now make decisions. So it can be like, okay, my, yeah, it can make decisions. It can now, whereas before it was, you said that you set enough rules around something you can force. You can make decisions for it, but it cannot actively think. Whereas now with the deep learning, it can, it can make decisions based off independently, basically. Right. Cause it's now learned how to, which is a big step forward yeah, it's a big step forward. So maybe I could see it, but I'm not sure. I'm not sure I want to see it. Do one question, Ankur? I do, I do, well, I have many questions to be honest, but one I'm really interested in, because you read a lot about the ethics and sort of data privacy of, of, of a, of the modern AI, the deep learning side, what are your opinions on that? On the ethics of, of AI. And also, what should brewers be careful of when giving data, especially to chat GPT? Yeah, very interesting. And I can say important question. these days, chat GPT is also working on, data governance and, data ethics. They also telling that they'll not share your data with someone else. Even they gave you the feature that you want them to use your data or not. So you have to be mindful. I'm going to switch on off that thing. Other thing I can say is, always anonymize your data. Let's say if you're putting their X brewery Excel sheet, give me analysis or make histogram of my, OG versus PG or line chart of my, beer final beer pH. Always anonymize your data and, these are the things that you should take care of. Also if you're very, very, I can say protective about your data, hire a guy who can build a chat GPT on your data in your company virtual premises. In that way, this data will not go out and it will remain in your brewery. This is what I was doing in my. Previous jobs when I was working as a data scientist or data analyst or into data science for giants So we build those system in house Hmm. so that we cannot expose, ended up exposing our data to someone. Because it's quite easy now with the latest version of chat GPT or the latest version at the time of this recording, whoever's listening to it, but, maybe it's in the future thinking, God, that's all archaic. but you can now upload files. So you can upload files to chat GPT and it'll analyze, ask it questions or whatever you need to do on it. But, people are gonna be quite careful about what files they use. And as you said, like if that data is not anonymous and you've, allowed the data privacy, then it has every right to use that data. In the model, right? Or in its machine learning. Is that correct? Yes. So it gets, there's a GDPR nightmare yeah. for those people dealing with that data. But even if it's sensitive sales data, I actually was on a call with a really good friend of mine, Christian, on, Friday maybe. And we were talking about this, episode and we were talking about the data privacy aspect of it. Cause I think it was really interesting. We were like, I wonder if you could just put in to, I don't know, into chat GPT and ask it for like current sales of, let's say a supermarket and what's trending well, because I bet someone out there has uploaded a document to chat GPT at some point asking it to sort out its data, but that's really sensitive data. And now you're giving it to, you can, in theory, or giving it to the hands of anyone. So it is quite. is, I bet a lot of people don't realize that side of it. Yes, but see, still you have that intuition. You're not sure. Yeah. Okay. Yeah. You don't know. You don't know. Because, again, see, chat GPT has human like neurons. It will make decisions on basis of its experience and interpretation. We don't know which data set it will take. Probably it'll take a sale of, baby diapers and it'll compare that with beer sales. Right Now, if you have a expert who understand beer data or, or the AI guy, he'll make that model in your, let's say in your laptop. It'll not connect it to internet. This model will take sales data. Let's say there are company who sell this data via sales data for your competitor or. Other people also, I've seen that in UK as well. And you can buy that data from them. They give access on like monthly basis. They have various plans. Use that data, use your sales data, put it into that code or solution. It spits out your sales forecasting. That way it's. More, feasible, but this is about the technology. So chat GPT also comes under AI and machine learning. So sales forecasting is happening on, on, on that. But my, personal interest is somewhere else. So my heart resides in, in the brewery. So once a brewer, always a brewer. Yeah. Yeah. My heart resides there too, as well. which is, brewers aren't always, and nor should they be like the most, technologically savvy, yeah. people, because there's a lot of things to manage in a brewery and yeah, it's just not really why anyone originally got into brewing. I would guess not everyone anyway, maybe Anka you're different, but for the, for the brewers listening to this, they might think, okay, this sounds, AI sounds, the latest development of AI or whatever sounds great. how should I integrate it into my brewery from the, from, let's say the start from almost the ground up as if they don't have much, but they really want to use this, the power of AI now, how should they? Yeah, from this again, I'll use my personal experience again. It's not hard guidelines. Probably my, my truth is this, your truth, maybe something else. So whatever I'm explaining, I'll be telling this, these are based on my experience. So first thing is understanding your needs is the most important. Because if, if you don't know what you want to do with help of AI, you ended up paying lots of money to charge equity. Let's say, let's say your need is to find missing PH values. You can do that using Excel. Null values are missing value with help of mean or average. But if you want to do something else, how this pH will impact my sales. Yes, you need something solid. So this is one thing. Then starting with very small things, tiny steps at time, which you can achieve easily. let's assume a brewery, which is not, I was, when you were telling me that You're working with breweries which are semi automated or fully automated. I was thinking about those breweries who doesn't have automation at all. Now, even, I've seen those breweries as well, which are, cannot be automated as well. Because of many things. Pipe is lying there, something is lying there, you cannot put sensors over there. And that pipe doesn't have a I can say capability to hold that sensor. For example, I'm just taking hypothetical or fictional brewery, a brewery, which is, I can say a hundred year old, there is no scope of putting any automation. And I want to have trend of my fermentation. The other day we were discussing about that instrument. You can put that digital floating hydrometer into this fermenting vessel. It will give you things without putting lots of cost infrastructure. Right. So this is what you can start with. Even this is not limited to the, I can say zero automation breweries. This is also can be applicable to fully automated. Yes. Every, every brewery. Yes. These days they are maintaining Google sheets, but only for one year. In AI, there is a very simple and basic rule. The more the data is best, the performance of the prediction will be. That's why we discussed earlier, right, a few minutes back, how to deal with your data ethics while dealing with the chat GPT. Similarly, just even if you have papers in your brewery, Let's say if 10 years back you were not using Excel or Google sheet, try to get them into digital format so that, you can learn from your past. So this is very, very strategic. So invest some money or I can say time or resources, whatever you can in managing your data. These days, things are very cheap and easily available in the market. Google gives 15 GB of free space. Right. We put unnecessary things into this G drive, but just put, and, yes, if, if you don't understand, because I being a brewer, I can understand a pain of brewer or a brewery business owner. They don't know what to do, or they'll be like, Oh, this guy's talking about too much of stuff. It's not my cup of tea. You can talk to the experts. They're happy to help you. And, yes. Listen to podcasts, or this podcast regularly, or other podcasts where they talk about AI. Train yourself, or train your team. Just share this podcast with any of your brewery people. So that they can listen or, or they can see, oh, straining is a, another thing and, yes, monitor your, data, whatever you have, just go back, just try to just build a one bar graph of your sales. Of, of your, I can say any, anything, and then I try to do things by yourself. Then only you will see the small achievable effects in your business. This will enable you. Okay. Because. Let's say you have done very small thing in making bar charts of your brewery, like sales bar chart. And, you started immediately taking decision. Oh, this beer is not selling, this beer is selling. Next month we'll be brewing this beer. Definitely this will give you more money into your business. And more money means? The money will be, or this business benefits will be enabling you to spend money on these solutions, which are by the way, very cheap these days. So these are the things that I suggest to fellow brewers or brewery business owners. think the, one of the key ones from what you said there for me is that, like, investing in data collection effectively, I. The power of data science, I'm going to steal your word again, but the power of data science and the power of AI and everything runs on data. So if you can't, if you're not collecting it, then it can't help you Yes. It can, it can give you guidelines, but it can't help you with what you need. and I really like the, like, understand your needs. Like, I would say like, let's say you're a brewery that, Primarily collecting things on paper, Mhm. yeah, probably majority, process things that you measure, time, temperatures, pHs, pressures, etc. A recording on paper, which is fine, like, what, out of that, Are you most concerned with? So let's say you've got fermentations that are finishing three or four days apart. And you're like, that's weird. You know, this is where collect your data, get it into AI. or yeah, get it, collect your data, get it into a spreadsheet, and then maybe gets, AI chat GPT or a data person to analyze it and be like, what is it that's causing it? And that's probably a really good way of getting in. It was identifying like, what is your problem or what is the problem that you want to fix and then starting from there to integrate it rather than just going, okay, we're going to measure everything, record everything, and then see what happens because you can, I think you can quite easily get drowned with data as well as you can find it helpful. so yeah, I really liked the point of, understand your needs and just go for, pick something that you want To change, it's actually going to benefit you and your business and then work from there. And then as that will show success because kind of data driven decisions generally do show good success, then that will be the impetus to sort of expand that out further and further. Yes. One more thing I would like to add over is, I have been into one, Google data analytics training and, there they were telling me before starting any data science or AI project for, for your business, you should have two sides of page paper. So draw a line middle of the paper. One is. Can I do this using AI or without AI? So that helps you to, understand your needs Hmm. So, sorry, I forgot to mention this very important point. Yeah, no, that's awesome. yeah, really helpful. anchor. So we've talked about the future of AI in brewing, but what would you personally One AI to be used for in brewing, like where do you think it will give the most success? And what do you feel personally about it? So once a brewer, always a brewer, right? So, sustainability is. That's my main point because being a brewer, we always spend or use lots of renewable or non renewable energy resources, or I can say water, which is the major part of the beer. So I've seen breweries use 2. 3 liters of water to 6 to 10 liters of water per liter of beer. So that is my major concern. So in the coming 10 years. I want to reduce down this average or, or this gap from 2. 3 to 10 should be very specific. If I can reduce this to one liter, that will be like, this is what my aim and goal is. And yes, in, in brewing, we use, three elements. Water, hot energy, cold energy. So I would love to, or even when I'm working on these things extensively, how we can save heat energy, cold energy, and water. Water scarcity is everywhere. Even in India, you won't believe, in India, we are buying water to do beer. Really? Heh. And that is also impacting the sale of beer because one day water source is from someplace and that day it's from someplace. And then we are putting, just, just understand how it's increasing the volume of water used per liter of, making per liter of beer. I'm putting into one RO system, RO is rejecting 40 percent of the water as a permeate water. So if I'm putting 10 liters, I'm getting six liters, four liters again getting wasted. And it is adding to that sustainability. So that brewery is going bad. And, this is my, these are my thoughts initially when I was thinking with heart and mind of a brewer later on after working into this data analytics and, Data science, I also got interacted with people who are working in financial institutes. They're working there as a data scientist and data analyst. So one of the guy had told me that these days they are doing climate analytics for the companies because for business, you need money, you'll be raising funds for funds, you'll be going to these banks and financial institutes. So they always say, if your sustainability score is like this. Your company is going to stay in the business for these many years. Yes, we can give you money and you can give us back. Right. So, so in that way, my vision is not only related to the brewing process. It's related to the business as well. Right. So it's, it's global cause. That's why this is also the reason that's why we see all breweries or everyone is Working on these things apart from like major concern is saving the environment, this climate analytics and getting funds, being sustainable into the business for a longer time. That is also required. Yes, that is my main area or I can say personal interest. Well, yeah, I mean, that's how we both originally got connected, for that, that similar passion for, what brewing sustainably is. and to some extent the data based decisions to get there. So it's really cool. yeah, I mean, obviously I really, I totally loved everything you said there. I think. The potential is huge for AIs, something I spoke about right at the top of the, of the show of where AI is now and what it can mean for ease of, ease of access to, to visualizing data as we spoke about, which is such a powerful point. hmm. and yeah, making basically data driven decisions to reduce. costs, reduced consumption and reduced impact on the environment. Those three goes hand in hand in hand. So, yeah, I think it was really cool. I think there's loads of, future of it. and especially now, chat GP T and, and that, and that use of it is really good for, finding patterns and data, especially big data, et cetera. So it's really cool. and I think. The future is really bright for it because anytime you can, increase your effective, effectively increase your ability to, to think with, with machines is going to generally get you better data driven decisions. So that's really cool. anchor, is there anything else you'd like to add to the podcast? Yes. one more thing that recently I can assume people have data privacy concern. They don't want to chat, use chat GPT or, they don't know how to code or a few breweries are very old. They don't know how to put automation. These days I'm, I'm working on this. I'm helping brewers. like globally, again, with my digital presence or my physical presence, so I am helping these breweries, brewers or brewery owners with various solution. It can be like helping them generating digital version of their data or putting a solution, very simple solution. Or, or just training their brewery staff or business decision maker, what AI is. So these are the things that I'm doing again, my main goal is sustainability so that they use this thing to being in the business for longer. Yeah, I love that. And it is very rare. You meet a data scientist who works and has a lot, an extensive experience within the brewing industry who can help craft breweries out. So, anchor, I'm going to leave in the show notes, a way of contacting you, whichever way you choose. so anyone who's listened to this, who has, wants some support or has questions on. AI in their brewery and how it can help. and if they need someone to point them in the right direction and anchor is going to be there. So, if that's okay with you anchor, I'm sure it is Yeah. I would love to. awesome, man. Yeah. So I'll get that in the show notes for everyone, along with the other stuff. So that's usually there. And, yeah, I think it's look anchor. I think that episode has been really. Fun and educational for me. and I'm sure the people listening, is rare that you get something so important, like AI, being translated into, into our industry. Cause we're normally, we're not a big industry, really the brewing industry, just the craft one. So it's really great to have someone come on and, and be able to expertly put what AI can do for brewing what it has already done. So thank you so much for that. I really appreciate it. and yeah, thank you so much for coming on the show. Yeah, I approached you. okay. You did. Yeah, that's, that's, that's absolutely correct. Yep. Yeah. I saw a post that you did on LinkedIn and I thought it was awesome. And then I liked it and then, yeah, you got in contact. So, yeah. it's funny how these podcasts, guests end up coming on this podcast. Sometimes it's through like personal connections. Sometimes it's through brewing events and sometimes it's a loose as I like someone's Post on LinkedIn once, but yeah, it's really cool. And thank you so much, man. I really appreciate it coming on. Thank you. Thanks a lot. Okay. So a big thank you everyone to anchor for sharing, his invaluable experience with us today. it's been totally fascinating as I said, and if you enjoyed it. Please make sure that you share with your brewing friends and colleagues. This is the level of support. It's just unbeatable for me. And I really need it at the moment to keep the pod going. so yeah, if you can share that around, I would really, really appreciate it. And hopefully the episode, will add value to their lives, as well. So thank you so much. And I will catch you on the next episode of the Modern Brewer Podcast