Resources

• ClimateAi - An AI-driven adaptation platform helping businesses manage climate risk across supply chains
• Climate Corporation - A digital agriculture platform using data science to optimize farming decisions
• Berkeley Energy Institute Blog - Insights on climate, energy markets, and regulation
• MCJ Newsletter - Weekly digest from My Climate Journey on climate tech innovation
• Book: "Food Citizenship: Food System Advocates in an Era of Distrust" by Ray A. Goldberg - One of the foundational reads recommended by Himanshu
• Food and Agriculture Organization of the UN - Source of crop volatility data referenced
• NOAA and ECMWF - Public weather and satellite data sources referenced

Host and Guest

Host: Mike Toffel, Professor, Harvard Business School (LinkedIn)

Guest: Himanshu Gupta, CEO and Co-founder, ClimateAi (LinkedIn)

Transcript

Editor's Note: The following was prepared by a machine algorithm, and may not perfectly reflect the audio file of the interview.

Mike Toffel:

Himanshu, thank you so much for joining us here on Climate Rising.

Himanshu Gupta:

Thanks for having me here, Mike.

Mike Toffel:

So, Himanshu, we brought you here in the midst of our series of episodes on climate resilience. And we're going to learn more about how you're tackling that issue with AI. I'm very excited about this conversation. But let me begin by just asking you to give us a brief introduction to yourself, a little bit about where you came from and how you got to becoming the CEO of Climate AI.

Himanshu Gupta:

Thanks, Mike. So, my name is Himanshu Gupta. I'm the CEO and co-founder of Climate AI. So, I come from a small countryside town or village in the north of India. And that's where I grew up. Then moved to the east side of India for my undergrad from one of those IIT systems. So, I'm very much an engineer by heart, and by practice still. Post that, started my career working in a French energy company, building smart grid products for French as well as Indian grids. And from there, I got a call from the government of India or the planning division of the prime minister's office to help shape the 12th five-year plan for India. We used to have five-year plans like China has. And in the process, also came in contact with many multiple business leaders globally and policy leaders. Then I came to Stanford for my MBA and Masters in Climate Science program, but I still am an engineer by heart, as I mentioned earlier. And started Climate AI right out of dorm rooms at Stanford University with my co-founder, Max, who was my dorm mate.

Mike Toffel:

Wow, interesting. This was during your MBA program.

Himanshu Gupta:

Yes, exactly.

Mike Toffel:

So, how did you and Max get focused on this issue of climate change and forecasting and so on? So, what was the genesis of your focus on this market?

Himanshu Gupta:

Yes, so there's a personal element to it and there is a professional element to it. And I'll talk about how life comes in full circle. So, I talked about how I grew up in this countryside town or village in India. And back then, I'm talking about late 80s and early 90s. There was no concept of a municipal water supply in those villages. And so, we were very much dependent on monsoons, both the timely arrival of monsoons as well as the exact amount of rainfall to fill up the aquifers. And every year, if there was a delay or more variability in the monsoons, the taps will run dry. And I still remember walking half a mile to a mile with my family to fetch water from a nearby river. Imagine doing that in 46-to-47-degree Celsius heat and which is close to 120 to 130 Fahrenheit and so, that memory still is a good memory; it's not a sad memory because for us it used to be a picnic at the riverside. I’m not sure if it’s the same for my mom or my mom and dad. So, the second is the professional life. When I was working in the government of India Farmers committing suicides in the western side part of India, these are the cotton farmers, who used to be a big social issue, economic issue, as well as political issue for the government. These farmers were committing suicides because the same drought that I talked about earlier would also lead to losses in cotton. And these farmers will see this as a big insult or like it's just a big taboo in India to be living under massive debt. And they knew they wouldn't be able to ever repay the debt; they’d commit suicides because of that. I also saw many fashion brands approaching us for a solution to this problem. And to me, that was that light bulb moment and mystery, like why are these fashion brands approaching us for this problem? Yes, of course, there was this reputation issue for these fashion brands because their end suppliers were committing suicides.

If you look at the cotton supply chain, India, Texas, Brazil, Morocco are the four or five major cotton producing crops of the world. And these brands were worried that if more farmers leave cotton farming altogether in India, what will happen to the long-term reliability of their cotton supply chains? So, they wanted to work with us, which is the government of India, on a solution for those farmers.

So, what that means is managing their supply chains, managing their operations, deciding their long-term strategy. So, all of that must change with respect to the new volatility we are seeing because of climate change, the new volatility of weather.

And so, I knew it's a big problem. Now the latest Bloomberg report came up and showed how it's a $3 trillion problem. But then the question was, do we have the solution for it? And that's where I met my co-founder Max. He is a son of a farmer from Ecuador. He grew up on his dad's farms growing pineapples and papayas in Ecuador. And he was also a math nerd and I was a climate nerd. So, that's how both of us were living together discussing the problem. And we thought, let's try and create a solution.

Mike Toffel:

Now we're talking about weather fluctuations. So, people think about climate change sometimes as like a growing slow but sure rise in temperatures. But of course, it's also about fluctuations both about heat, about moisture, which means droughts and floods. As you talked about the monsoon frequency, duration, predictability. There's a whole bunch of weather issues that we're seeing being disrupted. Given what you're seeing and these stories, how do you go from there to think about what some solutions could be?

Himanshu Gupta:

So, I'll talk about what we do; we are an adaptation platform. And there's a reason I use the word adaptation platform rather than just a forecasting platform. So, I will talk about that later, but back to your point, Mike. One of my complaints from the climate policy world over the last 15 years has been to frame this question of climate change as a 2050 issue and talking in terms of averages, like 1.5-degree targets, 2-degree targets, 4 degrees targets. So, what it does is that it for one, takes away the focus from this being a now issue versus a 2050 issue, right? And 2050 is not a magic number we are looking at. The second is the narrative from volatility as you described to long-term averages. And the moment you start talking about volatility, it does not matter whether the increase in volatility we are seeing is it because of natural cycles of climate change or is it because of man-made anthropogenic emissions. Volatility is increasing.

And the moment you craft a narrative around volatility, that's when businesses start to act and focus on it. Because risk is the name of the game for businesses. And what we have seen in our work is across the supply chains, across the geographies, across all the variables you describe like temperature, precipitation, soil moisture or hurricanes or floods or flash floods, we have seen an increase in the volatility. In fact, if you look at the data from the Food Security Initiative of UN, it talked about how from 2016 to 2022, some of the grains saw the highest volatility ever observed in grain pricing over the last 50, 60 years.

And one last point, and this is the quote from one of the largest tomato processors in the world. It's a European tomato processor. I was talking to the CEO of this tomato processor, they are, I think, more than a billion dollars in annual business, just selling tomatoes. And he told me that I've been running this business for the last 30 years. The first three years in terms of production of tomatoes in Europe have been the last three years from now. But if you look at the data on tomato yields, you'll see not much signal whether the yields are increasing or decreasing, but it's the volatility that is causing concerns among the businesses.

Mike Toffel:

Interesting. Yeah, so, okay, so in light of these, this volatility, how do you decide which markets to pursue? Because I imagine there's at least two dimensions. One is like geography and the other is like what crops. We're talking so far about cotton or tomatoes. So, it sounds like agricultural is the right, is the target, which makes a lot of sense because that's of all industries. But even there, agriculture is a huge industry, lots of markets, and of course, geographic dimensions to this, different crops in different places. There is grain, there's vegetable, there's a textile supply chain like cotton. So, when you're staring at this big, massive opportunity, how do you decide where are we going to focus?

Himanshu Gupta:

Absolutely. So, I'll give an example from a startup standpoint, like what is the startup playbook around where to focus. And I'll give an example from a product standpoint as well, like from a go-to-market standpoint and product standpoint, where to focus. So, for the first four years of our company, when we started the company, we just focused on developing and working on these AI models, which is how do we forecast risk of heat waves, wildfires, droughts, from two weeks and beyond?

Now that's a key term for your audience because we realized that most of the work that has been done in this space is up to 10 days out or 12 days out, which is the weather forecast. But as we talk to many of our customers and prospects, they said that if you tell me there's a heat wave coming next week, there's nothing I can do about it. What if you tell me about the risk a month in advance, two months in advance? I understand the risk forecast cannot be perfect, but if you tell me that, that makes it a bit more actionable for me to act. So, for the first four years, we just focused on developing these models. And then the question became which sector we target, right? Because the AI technology can be applicable to multiple sectors, both in private and public domains. So, one of our board members has this approach called the double diamond approach. So, you start from a go-to-market standpoint, start conducting a lot of interviews across the sectors. So, we talked to many executives in the insurance space, in the finance space, food agriculture, manufacturing, and in the public sector. And then we realized that food and agriculture and beverage industry is the first and the right industry to focus from an initial traction standpoint. And the reason is 40 to 80 % of the output of the industry is driven by weather, depending upon the location of the commodity you're looking at. So, there's a bigger pain there. Some of the companies we talk to, like I'm talking of the likes of major beverage companies and food companies, were spending billions of dollars in procuring those raw materials or agricultural linked raw materials. And the impact of volatility was up to 20, 30 percent.

So if you're spending $2 billion in buying potatoes, 30 % of that is a lot of money, $600 million, which these leaders told us that they are either losing or in terms of additional cost or loss of potential revenue because they didn't have enough supplies available. So, we realize there's a big pain point. This pain point converts into a big financial loss, loss of financial opportunity. So, it's worth focusing on. So, that's from a go-to-market standpoint.

Himanshu Gupta:

And the double diamond approach is you talk to multiple stakeholders widely, then you narrow down on, you know, these are the two sectors we think there's a big pain point. And then you go deeper, like expanding the diamond a bit. And then you say, like, OK, I'm going to focus on this now. go and so diverge, converge, diverge, converge. That's the double diamond approach.

So, we focused on food, agriculture and beverages. That's how we started. But within that, from a product standpoint, again, our technology could be applicable to procurement teams, to logistics teams, someone who is thinking about how I transport crops from farm gate to the shipping yard or to my factory. And given many of these crops have short shelf life, optimizing your logistics with respect to transport, to weather volatility also becomes very important. So, there's procurement, there's logistics, then there is shipping, then there is demand planning as well, right? So, there are four different teams. We realized that let's start with procurement teams from a product standpoint. Who is the persona we sell to? It's the chief procurement officer who's spending, I said, billions of dollars in buying these agriculture commodity products. And if we can help reduce volatility in procurement, in buying decisions for this persona, then of course our product will scale fast.

So, we started with that, and now it's been seven and half years since we started the company and three and a half years since we commercialized. Now we, of course, have a maturity in food, agriculture and beverage. We have expanded beyond that sector and are selling into manufacturing, governments like DOD in the US. We have expanded beyond procurement personas into logistics personas, demand marketing personas. And so that's how we call it, the climate intelligence model for the entire supply chain.

Mike Toffel:

And are you also selling to the food producer side? What you've just mentioned sounds mostly like is the food demand side.

Himanshu Gupta:

Yes. So, we work across the value chain. So, all the way from seed industry. So, we work with 30 % of the crop seed industry. Then into the agrochemical side. So, I'm talking about the likes of companies selling fertilizers and pesticides. So, I think it's public information. I can talk about BASF, UPL. And then on the food processing side, so Driscoll’s, Dole, Ocean Sprays of the World, and then to beverage side as well. So, PepsiCo and Suntory and AB InBev and so on and so forth. So, that completes the entire value chain.

Mike Toffel:

Okay, great. So, let's dig in a little bit more, and you started this, but I want to hear more about the decisions. So, your technology is helping to forecast these weather anomalies. And I imagine in the near term, it might affect like when do we harvest? If there's a big storm coming and our fruits are fragile, maybe we pick them up a little sooner than we otherwise would have to avoid the calamity of them all, like getting a hailstorm or a hurricane. And that's on the short side. On the long run, it might say, okay, so given trends, maybe we should be rethinking which crops or which seeds or which varieties were growing because what worked in the past may not work in the future given the forecast, the longer-term forecast.

So, am I right, first of all, in bookmarking the decisions as everything from like, let's pick in a week, not two weeks or three weeks, versus let's change what we're planting and then what are some of the key decisions in between? And this is more, I guess, on the supply side and then we can talk about the demand side.

Himanshu Gupta:

You encapsulated it really, well, Mike. In fact, you could be the honorary founder of Climate AI. So, as I mentioned, we are an adaptation platform. So, there are three parts of the platform. One is forecasting risk. So, let's say if you are chief procurement officer of Centauri, buying hops for producing beer in the beer value chain, right?

What is the risk of heat wave wildfire droughts on the key procurement areas? Number one, that's the risk focus. Second is converting that into insights. Would that heat wave coming in France going to be impacting my hops production or hops supply chain? That's the insight part.

And the third is, we call it the adaptation playbook, which is the sets of decisions that these region makers can make to build resilience from an operational standpoint for this season, but also build resilience from a long-term standpoint. So, the decision that again, the head of procurement for Centauri would make is number one, if I'm seeing a heat wave come in this part of Idaho in the US, which will lead to reduction in yields by XYZ percent or volumes by XYZ percent, should I allocate more production to the Southern Hemisphere?

Number two if our models are forecasting reduction in volume from these regions, then should I tap into the spot markets before the market realizes, and the price of hops go up considerably? That's the second decision, right? How do you think about the right hedging strategy for your business?

And the third is now if you're working directly with the farmers, which almost all our customers worked directly with the farmers because these are specialized commodities which they want to control the quality.

And in this case, our platform can also send alerts on like there's a heat wave coming up in this region which will have an impact on the harvesting cycle of this crop. Ask your farmers to harvest a week earlier or two weeks earlier. So, that's the third set of decisions that food companies or beverage companies, the procurement managers can work with the farmers to basically optimize that production volume.

But eventually, as they start seeing increased volatility in production or quality in many supplier regions, then they're also thinking about the long-term viability of their procurement strategy, right? And should we be moving to a new location or new supplier to reduce the volatility? Should I be introducing a new variety in this region to have climate resilience post-production? Should I be recommending the right water management strategy in those regions if those regions are very much going to be impacted by drought and so on and so forth? So, we call this the adaptation playbook. And that's what gives us the most satisfaction for all of my colleagues working passionately in the field of climate adaptation resilience is that we have enabled adaptation at the ground level with the farmers too.

Mike Toffel:

So, what's the technology driving these recommendation engines that you've developed? So, I imagine you must be pulling from at least some publicly available data sets that governments and perhaps their satellite, weather, et cetera, maybe there's private data as well. And then you're processing that data. I'm going to guess through machine learning algorithms and perhaps other even more sophisticated techniques. So, you give us a little bit of a technology 101 on what's under the engine here.

Himanshu Gupta:

So, there are two parts of our technology.

One is the key risk focus, as I described, what is the risk of a heat wave wildfires droughts from two weeks plus at this location? And there, we have a machine learning algorithm that basically taps into the long-term memory of Earth's climate. So, whatever is happening, I'm assuming, Mike, you are based out of Boston right now? OK. So, if you want to predict what is the probability of a heat wave in Boston next month, then oceans will have a more significant influence on that. So, whatever is happening in oceans today in terms of the bottom layer, the top layer of the ocean, the salt content in the oceans, and so on and so forth will drive the medium to long-term weather of Boston. When talking about weather timeframes, that's the local atmosphere that basically drives the short-term weather of any location. So, we realized that many of these publicly available models, like from NOAA and European Center, have not optimized their risk forecast from medium to long-term standpoint. So, what that means is they don't integrate the oceanic data as effectively as it should be.

So, our machine learning model basically identifies patterns. And this could be, there is this delta between the top layer and the bottom layer of the ocean in terms of temperature that is developing somewhere in the Indian Ocean. Or there is this high-pressure gradient over Atlantic Ocean, the North Atlantic somewhere developing, or the salt content is actually, we're seeing some deltas there. That gives us some signal. And one of the best signals is, of course, there are thousands of such patterns. We have all heard about Al Nino and La Nina. So, those are the pressure and temperature gradients over the Southern Pacific Ocean. But there are hundreds and literally thousands of them, which people have not studied. So, that's what our machine learning algorithm is able to pick up and then add these signals on top of the forecast from these government centers. So, that makes our forecast more reliable in terms of accuracy. But also, we then reduce it to one kilometer by one kilometer resolution to make it actionable as well.

Then the second part is converting that into actionable insights. As I said, if you are growing pistachios or almonds, both are cousins of each other in terms of crops. But pistachios are very vulnerable to heat waves during winters or warmer winters. Almonds are not. Almonds are very vulnerable to drought during winters, but pistachios are not. And so how would these risk forecasts impact on the quality and production of these crops in this case, or any asset. It's a second part. And we have these biophysics-driven AI models, which convert these things.

Typically, when you say machine learning model, you assume there's a lot of data available. And you throw data at this model and expect the model to give you output. But that's not how it happens in the climate space, right?

So, we look at the growth of a crop So, typically the physical simulations or biophysical simulations, that's how academia used to do it, would actually put input as sunlight, nutrition from the soil, water, and so on and so forth. And the output of the model would be the vegetative growth or the biomass accumulation in that crop. But then the problem is how crops are grown in Idaho is very different from how crops are grown in France. So, these biophysical models are very good from a research standpoint, not from giving practical recommendations for the practitioners.

Then if you look at the machine learning models, you don't have enough data. Like if you look at crops or barley or any of these crops, outside of corn and soybeans, there's not much data available, historical data available in terms of yield or soil type to be able to just have machine learning models do their magic. So, if you still throw data at machine learning model, you'll have a lot of hallucinations. And what it'll tell you is, as the temperatures cross 110 degrees, the crops will also increase by 3x. That's not how it should happen, right? Because there's not enough data. So, machine learning models are likely to be overfit.

So, we came up with this joint approach called biophysics-driven AI. So, what that means is basically the machine learning model is learning from this physical simulation, biophysical simulations that, hey, instead of trying to find patterns between growth of this crops and all these inputs across the growth stage, instead of focusing on the entire growth stage, focus only on this pollination stage. Because that's what the biophysical models are telling you, right, based on their simulations. So, then it basically guides the machine learning model into finding those patterns from that limited data. And that's how we get more signals and output on how a heat wave would impact crops products.

It's like if you're a self-driving car. But then self-driving cars will have to comply with traffic laws and regulations of every city. And every country is different. So, basically, it’s a simulator of traffic laws guiding the car into training and eventually validating with respect to the local environment in a city. It's the same approach that we used in converting this risk into yield or production insights across these crops.

Mike Toffel:

Very interesting. And so how much of this is based on public versus private data?

Himanshu Gupta:

So, the data input data that we use in our models is basically all the way from publicly available weather station data, from satellite sensor data, from NOAA or European Center and many other meteorological offices around the world. So, and then secondly, some of our clients also share their weather station data with us. Many of these food companies are operating their in-house weather stations at the farm level. So, we integrate our platform with that weather station data. And then many times, this is true for, I think, 60 to 70 % of the cases, some of our customers also share the historical production data and yield data with us. So, that makes our model intelligent.

Even if they are not sharing that data, they are basically sharing the intelligence. So, what do I mean by that? So, let's say if you are growing strawberries in California, and there are multiple strawberry producers in California, their agronomists have the best knowledge of at what temperature thresholds or soil moisture thresholds strawberries start to decline in quality or yield. So, our models come up with our thresholds.

But then these agronomists can go in and say like, hey, for California or Watsonville, is where strawberries are grown in California, the temperature threshold should be 74 rather than 72. So, they go and change that. And that basically further informs the model. And that's how basically we have built these climate intelligence model for many crops for more than 42 crops and 64 varieties, which only we have built. And is not much data that exists for these crops outside, not much intelligence. And that's why every player in the value chain we work with, that makes our model more intelligent. So, now we have, in fact, we only know how climate impacts strawberries production, because of our work with so many strawberry producers globally.

Mike Toffel:

Got it. So, this helps provide a bit of a competitive moat because you're using proprietary data that your clients are sharing with you to help you tune your models to help you help them.

So, are there competitors doing the same thing? I was aware of some competitors, some of which have gone bankrupt in the past, but is this a landscape where there's lots of people trying to grab real estate in this space, combining AI and forecasting and adaptation tools primarily in the ag sector. Is this a busy space or is this a space that's still relatively empty?

Himanshu Gupta:

So, as you pointed out, we used to have one big competitor that went bankrupt last year. But then if you look at the timeframes, from up to seven days out, to up to six months. There are some insurance companies who have set up the analytics teams who are trying to do like, okay, in the next 10 years, what is going to be the climate change impact on this crop and that crop? Because that aligns well with their insurance products.

So, that’s across the timeframe, then across the sectors. So, for example, like food and beverage and agriculture, because we went deep inside that sector. So, now we work with 52 of the top 200 food, beverage and agriculture companies globally. Our positioning is very mature, but then if you look at real estate, there are so many competitors out there who are doing the climate impacts on real estate, right? And especially the decadal impacts of floods and hurricanes on real estate values. But that's not the sector that we focus on. As for the applications and the personas that we are focusing on, we don't have a big competitor out there.

Mike Toffel:

Well, that's great. So, tell me a little bit about the organization. So, you're based in San Francisco.

Himanshu Gupta:

Yes, we have our headquarters right opposite the Salesforce Spark. But we have a distributed team. 30 to 40 % of the team is based in San Francisco; the rest of the team is distributed across various parts of the US. We have a small team in New York and small team in Canada, and they used to have an office in Mexico as well.

Mike Toffel:

How large is the team right now?

Himanshu Gupta:

So, currently we have around part-time and full-time 52 employees.

Mike Toffel:

And what are the types of skills that you hire for? What are the different functions that people are engaged in?

Himanshu Gupta:

So, remember I mentioned that for the first four years, you were just focused on developing the models. So, I, as a founder, spent 60 to 70 % of my time trying to chase AI scientists or climate scientists globally who had experience in applying AI to physical science data sets. That's the very specialized skill set out there.

You can't be hiring anyone from Meta or any of these AI companies and expecting them to get comfortable with applying those to physical science data sets. For example, one of the lead climate scientists that we have was also the climate mission director for India's mission to Mars and Moon. And so, we said like, we need the top the brilliant minds in AI who have experience working in climate science, but we also need brilliant minds in climate science who have done it at the ground and know how to implement these models in making decisions, right? So, launching spacecraft, launching missions, you need to decide what specific day and time the mission can be launched. And the most significant determinant of that launch window is medium term weather. Those decisions are made at least a month out or two to three weeks out. So, we had a scientist who was expert at that because if you're wrong, you're looking at basically millions of dollars of losses for the missions. So, that's our core team.

We have around eight PhDs who have published more than 600 papers amongst them in climate science and AI. And then we have the go-to-market team and the product team and the engineering team around it. One single quality of in the business team that you will see is this passion for climate adaptation for food security. That's their passion and that's the single determinant.

We don't expect anyone to know much about climate in the business side. We expect people to be very good at what they do. Like if you're at product management, if you're good at sales in selling SaaS software in the industry, if you're good at customer success for AI solutions, that's all we need. We'll teach you climate, but we can't teach you how to be passionate about climate.

Mike Toffel:

What's your experience with the fundraising process? Who are funders whom you sought money from? What's been their response? I often hear that folks are looking not just for funds, but also for advice from their funders. Can you tell us a little bit about your journey along the fundraising process?

Himanshu Gupta:

Good question, Mike. So, one is what I thought I wanted in terms of my funders. The other is what was the reality. So, I came with this mindset that I think I know a lot about climate after having spent almost a decade back in 2018 in the climate space. So, I have a good network globally of climate leaders.

And so, I don't need expertise from my funders in climate. I need expertise from my funders in enterprise SaaS, AI, or food. And in hindsight, I think that was the right strategy. But what happened is that we launched this company in January 2018. But what happened around that time was the Trump administration won. And the first decision that was made was to pull out of Paris Climate Accord. So, when we started pitching to seed investors, their response was, are you a nonprofit? Who's going to pay money for it at the end of the day? And it was difficult times. I realized that I should go back to India. But our professors within Stanford University, who we took classes from, put in some angel money.

We also got some funding from the early investors in Climate Corporation. And so the lead investor, I reached out cold to the investor, and he said, yeah, he made some good money out of it. It was former partner at NEA, then he started his own fund. And then he basically decided to lead the round as well, but that was difficult. Run the clock forward five years.

And of course, climate tech became such a hot and a sexy space for investing. To an extent that an investor preempted our Series B. We didn't even need to go out and raise Series B out there. And what helped was, of course business performance, but also we as a climate company were not exactly dependent on regulations, right? So, our value prop to our customers was all about ROI.

Mike Toffel:

Got it. So, let's look forward. So, you have a background in policy, you mentioned already. Here we are again with the administration taking a very different tact than the prior administration when it comes to policy. Now a lot of that policy was about decarbonization, rather than adaptation. So, maybe the policy shift has not been that important for you, but to some extent, the more we take our foot off the decarbonization lever, the more adaptation we're going to have to do. So, is there a relationship between the policy arena and the changes recently in the U.S. and perhaps elsewhere that's affecting your market plans, your go-to-market approach, anything like that? Or is it just a different work?

Himanshu Gupta:

So, for us, we have been more impacted by tariff policies rather than the climate policies or climate regulations. Because most of our clients and customers are companies with global supply chains. What I believe is uncertain markets are worse than bad markets.

So, when there's a lot of uncertainty, you have these innovation teams and the digital technology teams within the Fortune 5000 companies who are not willing to decide because there's a lot of uncertainty. And that has impacted us a bit more versus the regulation side of things. As I mentioned earlier, our value prop was not dependent on regulations for that matter. But having said that, the way I spent the first eight years of my career in mitigation, because I modeled the emissions for India, which went into our submissions for Paris Climate Accord. And then the last eight years have been adaptation, for that matter. I feel like for some reason adaptation got lower into the stick than mitigation but both required the same level of urgency.

So, what I mean by that is like think about COVID pandemic, right? When the pandemic happened, our first response was to go and develop vaccines and figure out a cure there and that's our immediate priority for us. So, we went from having no idea about the virus to basically researching and trialing and producing more than 6 billion vaccines within 18 months. While at the same time, we don't see more events as we saw in the last couple of weeks of Texas floods. So, this has been a bit of a surprise to me why adaptation got a lower priority over the last seven or eight years, more so because as we see in our work, there is short to medium term ROI for businesses to act on it. And it's not dependent on any regulation.

Mike Toffel:

So, as you're seeing, the pace of technological innovation continues to accelerate, is this providing new opportunities? Does this mean the models you developed just three years ago are now hopelessly out of date?

Himanshu Gupta:

Great question, Mike. So, this is one of our classical founders' mistakes, including mine. We started with the technology-first approach or technology-first mindset, which is you work on an idea that, OK, how do we improve the reliability of forecast in the cross-east time frames using back then, deep learning algorithms.

But we realize it does not matter to the end user. The right mindset to have is a problem first mindset. Like sitting down with the user, the user could be, as I mentioned to you, the head of procurement for PepsiCo. The user could be a farmer somewhere in the north of India. And they don't really care what technology you're using. What they need is like, can you solve their problem in a more intuitive way, and in a more affordable way. And so that was the mindset shift we encouraged in our company three or four years ago. So, start with the problem first.

Mike Toffel:

How do you think about the accuracy piece? So, over time, there can be growing evidence of how accurate your forecasts are and how much value you're bringing. You know, was it a false alarm? Did it actually save them millions of dollars and so on? So, I imagine once you have customers, they're doing this accounting. But for new customers, are you basically just bringing stories to them saying with our prior customers, we've enabled them to avoid having to pay $10 million extra on barley because we advise them to buy it early. Is that the approach to talking about gaining credibility around your forecasting?

Himanshu Gupta:

Yes, that's the best approach, which is not specific to us. For any enterprise grade analytics startup, that's how you gain credibility. In fact, we have three of major Japanese food companies. And for your audience, there are some founders who have tried selling to Japanese companies. And they would have realized how much of that sell is based on your established credibility and trust in the market versus whatever fancy value proposition you can show to those companies.

So, that's one. And the other market dynamics that impact us is, food, agriculture, and beverage is a big yet concentrated market. So, what that means is the current chief procurement officer at Suntory might have been the head of procurement in PepsiCo somewhere. So, the entire industry is interconnected. So, if you are producing value for one of your clients, the word spreads more quickly than you realize. But if you end up messing up the value with one of the clients, again, the word will spread more quickly than you . So, that's one. And now that's the best way of gaining trust with the customer. But second, of course, that's another headwind that we deal with is at the end of the day, these are probabilistic forecasts. We might go wrong.And so in that case, what helps us is, and that's why our best persona that we sell to are the supply chain or procurement personas or demand planning personas who are making decisions, optimizing decisions across multiple regions or across multiple locations. Where we say, okay, in a few locations we might be wrong, but on an average,

you'll end up getting more revenue or saving some cost. Very similar to like if your mutual fund approach in stock markets. where our prediction for one stock might be wrong, but overall the mutual fund will generate more revenue for you, more returns for you than your current approach. So, those personas who have a bit more global portfolio, a bit more diversified portfolio are a best fit for us for that matter.

Mike Toffel:

Yeah. So, we've been talking mostly about your sales to companies. Are you also selling into the government markets, which I know is a very different type of marketplace?

Himanshu Gupta:

Yes, so it was more of a pull. So, I don't want to name the government, but there was a Middle Eastern government that approached us right after the Russia-Ukraine war. If you remember that year, 2022, the Russia-Ukraine war disrupted the wheat markets because Russia and Ukraine were providing 30 % of the global wheat supply. But on top of that, there was a heat wave in India, there was heat wave in France and a drought in Argentina, all happening at the same time. That led to massive panic in the global wheat markets, panic about affordability in a lot of African economies who were importing wheat and panic about availability in a lot of Middle Eastern countries. I mean, they can afford to pay more, but they were not sure if enough wheat could be available. And they import basically 80 to 90 % of their wheat requirement annually. So, we had one minister of food and environment coming to us and on Zoom talking about what can we do to monitor food security situation for our country through two means. One is how do we monitor and forecast production in our country on wheat, and how do we monitor and forecast wheat production in the countries we are importing from to take care of wheat reliability for the next few months. And then, eventually, how do we think about food security in the long run? Which countries should we be importing from? How do we enhance production in our own countries? And that's where food security and climate security became a geopolitical security narrative.

Mike Toffel:

Right, Fascinating. So, let me ask you my final question that I ask all guests, which is for those who have found this conversation interesting enough that they want to learn more, potentially to work in this space or invest in this space, what are some of the resources that you typically suggest folks pursue, whether it be websites or conferences or newsletters or podcasts, to learn more?

Himanshu Gupta:

Absolutely. So, to me, of course, as a founder, I did not know anything about agriculture when I started this company. I knew a lot about climate. I did not know anything about agriculture. So, I learned a lot from my co-founder, but also from our clients. And being very honest about it, I went to them and told them, hey, head of procurement for potato processors, I do not know anything about potatoes, but I know a lot about climate. Can you help me understand your pain points? And then I can work with you to figure out if we can solve them or not. So, that was one. So, I learned a lot on the job. And the reason I say that is, irrespective of how much you learn on websites and conferences, there is no substitute for going on the field in the factories and hearing from the horse’s mouth directly. I call it wisdom rather than knowledge.

The second is I read a lot. So, I remember talking about HBS when I started a journey in food and agriculture. I used to read his book by Dr. Ray Goldberg. He's one of the OGs in food and agriculture and agribusiness. And I met him, and I read his book a lot. I used to basically read this newsletter. It's called MCJ Newsletter, which is very good to keep you abreast of the latest developments in climate. I do also read blogs from some of the professors because articles and news articles give you a very high-level understanding. So, there, for example, the Berkeley Energy Institute Blog. If someone wants to go deeper into the entire dynamics between extreme weather or climate and electricity pricing and demand pricing and how the electricity sector is evolving globally from a regulation standpoint, from a technology standpoint, that blog is very good.

Mike Toffel:

Great, well Himanshu, thank you so much for spending your time sharing your wisdom with us and our listeners. It's been a great conversation; I really appreciate it. And we'll share those resources that you referred to in our show notes on our website, climaterising.org.

Himanshu Gupta:

Thanks, Mike.