For anyone in product development and R&D at a food and beverage company, this episode is definitely for you. With us on the podcast today is Manmit Shrimali, co-founder and CEO of Turing Labs AI. Founded in 2019, Turing Labs, which recently concluded $16.5 Million Dollars in Series A funding, has been helping CPG companies get new or improved products to market faster using their unique AI platform.
Listen in to hear how Turing is helping some of the biggest names in CPG create or reformulate faster and more efficiently. Perfect for the post-COVID hybrid R&D Department, Manmit walks us through how the platform uses artificial intelligence to simulate working through different ingredient combinations to determine what works…and what doesn’t. You’ll definitely want to tune in to hear how this revolutionary platform is helping manufacturers take thousands of data points and turn them into formulation gold.
Transcript
Erin: Manmit, welcome to the Food For Thought podcast. Let's dive right in. I want to kick things off by getting to know you a little better. Who are you? And what is it you do within the food and beverage industry, or maybe more specifically, within AI?
Manmit: This whole idea of the food and beverage space is very dear to me. I and my co-founder, Ajith, have decades of experience in this food space. And we have seen firsthand the struggles and the competition that food and beverage companies are up to today. Especially when the cost of entry is pretty low. They're really trying to figure out how do we develop better products and outperform the competition? But the things that they were trying to do in the past were not helping them because of legacy systems and legacy process.
I saw that firsthand and realized that this is an industry that requires a new approach, new systems, new way of thinking to compete in this digital world. And that passion allowed me to form Turing Labs, which is a Y Combinator-backed company, and backed by other major Silicon Valley investors. What we built is a software as a service platform exclusively for food and beverages ground up, to help them innovate better and faster.
Erin: I want to dig in more and talk more about Turning Labs, a tool combines artificial intelligence and product development. That's still a pretty broad description. How are food and beverage companies using this tool for their product development?
Manmit: What has happened is that CPG companies have been among the last in the industry to adopt AI technologies, especially for product development.
AI is very complex. It's been around for decades. But there is a lot of misinterpretation of what exactly AI is. The problem right now is that there is a belief that AI is a tool that can do all things and has magical algorithms and some magic into it that can solve any different kinds of problems for all different industries and for all different kinds of use cases. Unfortunately, that's not true. And that's why major investments in AI have not resulted into massive quantifiable returns.
Our first belief was that how do we first help the food and beverage industry to understand what is AI and how it can be used or can be even used in product development? Because the formulators have been using manual process for more than 100 years, whether they want to develop a new product, or whether they want to renovate a product or a line extension, or create a new category, they relied on the old systems and processes. They are not very much aware of what is AI and how does AI fit in product development.
Companies are using Turing in in two ways. First, is in terms of strategy. How do we think about AI and how AI can actually help in improving and accelerating product development? You may have all this data, you may have all this vision is that you want to go to market faster with better product, you want to, you know, respond and address to the market needs and, you know, market pressure forces. But how does AI help?
What is the foundation of using AI exclusively for R&D and product development? How do we get started about that? Are we even ready for that? We have some data, some digital, some analog, some physical, some in paper forms as well. But is it enough? Do we have something in our infrastructure to even get started with this journey of AI? So that's the first place. They come to Turing to figure out where they are in terms of the AI journey, and do they have what it needs to start? If not, what they need to do.
The second part is about execution, about improving operational efficiency. They use Turing to go from idea to commercialization, end-to-end digitally 10 times faster, replacing the menu trial and error process of doing hundreds and hundreds of R&D cycles of testing the product, testing the idea, changing the ingredients, testing the different formulations on several product attributes, several different kinds of tests, to replace all that with a software that can help you to do that digitally so that you can now develop fail-safe product faster virtually, and only do the last few test as a sanity test and take the product to the market or for commercialization.
Some of the world's largest companies are relying on Turing to really build their digital transformation journey, achieve better return on investment, and also address the changing market needs and consumer needs through a digital tool that leverages AI for renovation or innovation projects.
Erin: I want to go back for a second and talk more about the hesitancy with CPG companies. Why do you think CPG is so late in adopting AI?
Manmit: So there are three things. First is the culture. They're still based on few approaches embedded into the processes which were founded 100 years ago. And because these companies are so giant, and there are different stages, it's much difficult for them to pivot to more agile and digital stream. If you're a software company, it's much easier for you to develop agile system because it's virtual software. CPG is among the few industries where you have to develop a product by different stages. And it's not just about efficacy, there are so many other factors that you have to consider apart from developing a formula or developing a product, let's say, is it manufacturer level, for example? Would it meet the consumer requirements? Would it hit the goals in terms of our sales? Would it have the shelf life? Would it appeal the retailer to give the shelf space to them? It's a very complex multi-stage problem. They had a very difficult time trying to figure out, "We want to be agile, but how do we transform this multi-stage approach, you know, into a multi-agile approach?"
Second, their processes, which are not really suited for typical AI or typical ML kind of statistical software, because those kinds of software assume that you have all the data in place, it's pretty much clean, and you have much more data at every stage, and there's so complexities out there, like financial services, for example. The whole complexity of developing a product brings unique challenges. And that's why generic approaches or generally AI systems never needed to the CPG world or food and beverages world.
The second is the data part. CPGs have been a little bit late in digitizing their data. In fact, you'll be surprised even today we encounter quite a few companies whose data is still on the paper notebooks, or they still do a lot of testing or capture the test results on paper pencil, for example. And some of the test results that they do, they don't even record all the results. They may only record the results where the formula was winning. But there are quite a few areas where the formula didn't perform, did not perform well. And they just literally discarded that information and there was no storage of such information. So that was another bottleneck for them to, you know, jump on the journey.
The third thing is they were very reactive. There is significantly more competition. The cost of entry is so low that anybody can develop a beverage or sauce and start selling it on Amazon. For at least the last 10 years, there has been this tremendous pressure to rethink the entire way to survive in the market and continuously create a new category and explore new categories.
These are some of the biggest bottlenecks for them in the past to jump on this journey. And one of the big reasons we are seeing now is companies want to go on this journey, but they don't know how to do it. They don't have the right people, they have never done this in the past, or they don't get enough data scientists to do the right kind of things because it's very complex that it requires a lot of domain knowledge.
This is one of the few industries that relies heavily on some of the people who have decades and decades of experience creating the product. This industry is still heavily reliant on those few veterans or few folks in this industry. And it's really difficult to replace them. So it becomes very difficult for use of digital tools when you are heavily dependent on the creativity of these amazing experts in the organization.
Erin: It sounds like Turing takes a lot of pieces of data from different sources, and helps manufacturers put it in one place so they can make fast and efficient changes to their formulations. Am I on target with that, or is it more to it than that?
Manmit: There's much more to that. When we built this entire interface, the workflow, the algorithms for food and beverages, in that journey, one of the things we learned is about the data. And customers like Turing because we have changed the way they think about the data and the way they think about AI, and we have significantly accelerated the time to realize the value and return on their investments.
As an example. There's a foundation belief that let's just collect all the data that we have, digitize it, let's use AI, and then see what's going to happen. We approach that very differently. The way we approach is that what are the decisions you're trying to form? Where do you want to be in terms of a digital product development of virtual product development? Which are the categories or the brands that are most critical to you? And then do the reverse engineering. Okay, for this space, where you need to get which data and what quality of the data do you currently have, and what the missing information or missing gap you need to fill to embark on this journey.
Usually, customers don't go through the journey where they're trying to get all different data and different things from the organization. Rather, they go a very targeted search that in order to achieve a specific innovation or reformulation goal, for example, “What do I need at minimum to get started?” Because Turing allows them to assess the quality, the quantity of the data that's required to make those decisions virtually.
Turing can get data from different sources and different stages. The typical stage-gate process of developing a new product, data analytical, data sensory or brief on the consumer side, manufacturing side, packaging side, or process side. We collect all this data, but the approach is much more targeted, much more refined. We only use the data that's really meaningful rather than boiling the entire ocean. This really helps the manufacturers and the food and beverage companies in terms of leveraging what they have, collecting what's really required to fill the information gap or the knowledge gap, and then start building institution knowledge for a specific category or specific brand.
The way they benefit from the tool is that instead of having this data which was siloed across organization, now they have a single place where they can glean the insights from what they know in the past, what are the key things from their past data? What are the areas they know today about their formulation space, and where they need to learn more about that? That gives them a great departure point to really start working towards building new product or innovate virtually. And from there, they virtually get us, you know, a system where it recommends them in terms of where they should think about building the prototypes.
Even before a person creates a physical benchtop samples, they're pretty much sure which samples are going to return on the best return in terms of efficacy, in terms of the cost, sensory outcomes, etc. And from there, they go to the next stage in terms of virtually assessing hundreds and hundreds of prototypes they would have, and get the answers instantly replacing this physical manual world of scheduling the batches of the sample for testing, coordinating the physical test, working with the suppliers, checking whether it's manufacturable or not, all those things are taken care of by the platform. And they can virtually simulate hundreds of different scenarios and pick the one that is most suitable for them in terms of the product goal or the business goal and things like that.
We then take the last step, which is a recommendation where ingredients, which are a core part of any product. If it's a multi-dimensional problem, where you may have millions of combinations, especially when you nowadays look at even the startups have amazing different kinds of proteins which they could actually use in a product. It becomes very complex for a human being to figure out, "Can I go through all different combinations and come up with a best formula that performs better on all different measures and all different product and business metrics?"
That's where Turing does the heavy lifting. Turing searches all these different data points, siloed data points, loads from it, and takes into consideration the goals of the product, the goals of the business, and recommends the top three areas where they should innovate, and the top four to five formulas which they can quickly assess it and take it to the market. It's an end-to-end approach where we are helping them from the idea, from the data side, and how to really translate all those into a meaningful blockbuster product, which is more likely to win in the market than just, you know, throwing darts on a dartboard.
[javascriptSnippet ]
Erin: How long has Turing Labs been around? And how old is the company?
Manmit: We have been around for more than about two years now. We were in a stealth mode for quite some time because we tested our orders, we did heavy, extensive research to build something exclusively for CPG R&D to take care of all different elements that goes into product development. So it's very hard work of two years instead of working with early adopters in order to have some of the world's largest companies leverage Turing right now. And we are backed by Silicon Valley investors.
It's a pretty young company, but we have a big ambition. We want to make sure that 85% products don't fail in the market. We want to improve the odds of winning those success. We want to help the CPG to be the forefront of leading the innovation and things like that. And we built an entire company, the people that we hire, everybody has a significant amount of domain knowledge. And anything that we do is today around product development, so we are becoming kind of a default platform when it comes to innovation or innovation in the food and beverage homecare and personal care products.
Erin: Did you see any major shift with people or companies using the tool once the pandemic started?
Manmit: You know what, Erin, this has been amazing for us actually. I think what happened is that the pandemic was extremely disruptive for product developers who were locked out of the lab initially, right? They could not perform the testing they were used to, they could not do it in a physical world. Even when allowed back in, extremely safe measures made it very difficult for them to run benchtop samples and run pilot plant samples. And consequently, they and the business teams proceeded with, you know, kind of a guesstimate, and accepted suboptimal formulations in order to meet the timelines.
Turing's speed in reduction integration is a perfect solution in this environment. With Turing, they no longer have to rely on the physical world, and they can do about everything virtually. What has happened is that this pandemic has significantly accelerated the use of the digital tools, or has significantly challenged the culture problem, significantly empowered the formulators and product developers to use other tools and systems at their disposal to address the supply chain issues, for example, the cost issues, the competition issues, the need to hit the timelines based on the manufacturer measurement, manufacturability as well as, you know, the shelf timelines. The pandemic has actually given a huge boost in adoption of our technology, and companies are significantly much more open to embrace new approaches and investment tools.
Erin: Are you able to reveal the names of any of the CPG brands that use your tool?
Manmit: I'd love to but the problem is that we are in a situation where we work exclusively in this industry. It becomes very difficult for us to disclose the names. But we right now work with a few of the world's largest CPGs.
I'll be very surprised that these large companies would not know about this, because many of them are already boosting up their journey to digital transformation and achieving amazing, amazing results in terms of developing their innovative products in record time.
