Green Giants – Episode 99

[03:06] Wes Ashworth

Really cool journey. I think that personal connection really does always matter, and it shapes how someone chooses what problems are worth spending years of their life on. As you mentioned there, too, you studied Atmosphere and Energy Engineering at Stanford. What were you convinced the big climate problems were back then, and how has that view changed since?

[03:25] Amrit Robbins

The biggest elements of the overall carbon emission pie have always been transportation and buildings. Those are the ones that get all the attention, and of course then power generation. Of course those ones get most of the attention, as they should. As they very well should. I was really excited about those. Those are the ones that are most tangible. We spend all day in buildings. We drive around in our cars. We consume power all the time. They’re very kind of real and tangible and applicable to our lives.

But there’s all these other categories too. I had an opportunity, and I kind of stumbled into one of these other categories of key drivers of emissions, which is refrigerant leaks. Coming out of Stanford, I went and worked at an energy services company where our largest customer was a major national grocery chain. The goal with this company was basically to help them retrofit their existing buildings to make them more energy efficient, save a bunch of money on energy, on maintenance, on operations, and then finance the project out of those savings that are generated.

When you’re looking at a big portfolio of retail grocery stores, the largest driver of energy consumption by far was refrigeration, which is like 60%. In the first meeting, they told us, you could ignore that, there’s nothing you can do for that, focus on the other 40%. That was like a huge aha moment for me as well, realizing, wow, there’s such a huge opportunity and folks just aren’t paying attention to it.

I didn’t have any answers then, I just had questions. That kind of started pushing me down this path of focusing on food retail and cold storage and the cold chain and then the refrigeration asset class. Again, kind of by accident, stumbled into this other secondary problem, which actually is now really a primary problem that I’m focused on, which is refrigerant leaks.

It turns out that when you look globally at all anthropomorphic, all human-caused climate emissions, or CO2 equivalent emissions, three to four percent of those are driven by refrigerant leaks. Your air conditioner, your refrigeration system, your car leaking refrigerant into the atmosphere, that is a larger impact than the entire aviation industry by a factor of two. It’s more than the entire shipping industry by a factor of two. It’s more than the entire cement industry, just leaking refrigerant gas.

To put this in context, you leak one pound of an HFC refrigerant, that has the same impact as emitting 2,000 to 4,000 pounds of carbon dioxide. You know, you leak two to four pounds of this stuff, that has the same impact as driving a car for an entire year.

Then you start thinking about the customers that we serve in retail grocery. A typical grocery store is leaking 800 to 1,000 pounds of this stuff per year. Basically, we ran the numbers, very high level, very kind of squishy math here, but if we were to completely solve the refrigerant leak problem in retail grocery in the United States, that would have an approximate impact equivalent to taking 30 million cars off the road. It is just a staggering, staggering opportunity here that’s kind of hidden in plain sight. It’s very unsexy, refrigerant leaking in the back of a grocery store, and nobody thinks about that stuff, but there’s so much opportunity here.

My final thought is basically, it’s been estimated that if you can eliminate these F-gas, what they’re called in Europe, at least they call them F-gases, if you can eliminate these F-gas emissions worldwide, you could avoid half a degree Celsius of global warming. This is incredibly meaningful.

[06:29] Wes Ashworth

Incredibly meaningful, certainly probably an understatement. It’s funny, you know, it’s not a sexy topic. Even when we first kind of connected, I was like, all right. But then the more I learned, I was like, oh my gosh, you really hear and see the enormity of this issue, the impact that you can have. Then you’re really making this huge difference there where a lot of people are just completely unaware, completely ignoring it. As you said, oh here’s that, like, just ignore that, we can’t do anything about it, to really doing something about it, is absolutely incredible.

I appreciate you going through that and sharing some of those stats as well, too. Now we’ve kind of set the stage and understand your journey a little bit. I want to make the scale of the problem tangible. You started to there, because this is one of those areas where the climate significance is far bigger than public awareness. I want to really drill into that. Solving refrigerant leakage could have an impact comparable to taking just a huge share of the cars off the road, as you just shared. Anything else there that people should know? Unpack that a little bit more for us. What else is just the overall impact of what this could be?

[07:31] Amrit Robbins

Project Drawdown goes through and ranks every single opportunity that humankind has to reduce climate change in terms of magnitude. How big of an impact could it have to, for example, switch all gas to solar, or et cetera, et cetera, every single thing you can think of. The number one item on that entire list until a couple of years ago was basically eliminating refrigerant leaks.

What’s happened now on the updated list is they split that into two categories. One of which is reducing leaks and managing leaks, and the other one is switching to new natural refrigerants. Now they’re a little bit lower, but they’re still both top 10. Previously, this was literally the number one thing that humans could do to reduce climate change from a magnitude standpoint and from a cost-effectiveness standpoint, is just stop leaking this stuff.

I mean, it’s just such a powerful greenhouse gas, and so much of it is getting leaked. Especially when you’re thinking about where these leaks are coming from, there’s millions and millions of air conditioners out there and they leak tiny amounts, which adds up in totality to massive amounts. But then there’s also these 100,000 or 500,000 facilities worldwide, in that order of magnitude range, where they’re leaking a lot per facility. That’s kind of the low-hanging fruit here. Let’s focus on this relatively small number of facilities that are having an outsized impact on refrigerant leaks worldwide, and as a result an outsized impact on climate change worldwide, overall CO2 equivalent emissions.

[08:50] Wes Ashworth

So glad we’re doing this. Comparisons like that just immediately change how people size the opportunity because they put an invisible issue into familiar terms, and just seeing obviously the magnitude of that is remarkable. We’ll continue to get into that as we go here.

Let me ask you this, and maybe it’s obvious to some, but why do you think this problem has stayed invisible for so long, despite the scale of the emissions involved? You said it was listed as number one, now they kind of break it into two issues, it’s still top 10. Why has it stayed invisible for so long?

[09:17] Amrit Robbins

I think it’s because it is so unsexy. What’s interesting here is that once you start to unpack it from a climate change standpoint, it’s very sexy. Then for our customers, they care about their own quality of life, which is abysmal right now because they’re kind of living 30 or 40 years in the past without using data, without using intelligence to inform their operations. They’re kind of just reacting to an avalanche of emergency alarms all the time.

They also care about things like eliminating compliance risk, which can completely upend your entire organization’s strategic plan for a year. Then they care about things like reducing costs and achieving and meeting budgets for the first time in a few years because they go over every single year. Then you start thinking about it from an investor standpoint. That’s where injecting the AI component and the technology component and the repeatability and scalability component gets really exciting and sexy.

Long story short, this has been very unsexy for a long time, but it’s sexy to climate folks, to customers and end users, and to investors and technologists if you just put the puzzle pieces together in the right way. That’s what we’re really working to do at Axiom.

[10:19] Wes Ashworth

I could see a lot of added value as well too, way beyond the climate discussion. Obviously, our group and listeners, we care deeply about that topic, but some of these business owners and things like that, they’re indifferent about it, or may not have a strong opinion either way. What is it, what’s in it for them? Why should they care when it comes to reducing this, a solution that involves this, outside of just climate concerns? What are the other advantages or benefits for them?

[10:44] Amrit Robbins

Well, I think the biggest one is basically, you think about the, so let’s say you’re a facilities manager overseeing 200 grocery stores or 100 grocery stores in a specific region. Your daily life doesn’t feel so great. You come to work in the morning at seven or eight a.m. and you’re immediately kind of engulfed by a stream of emergencies. Let’s say you have 20 hair-on-fire emergencies all happening at once. There is no possible way that you’re going to solve all 20 of those today.

What you have to do is you have to basically, and you have no data, you have no intelligence either. It’s just like a bunch of people calling you and screaming for help, or a bunch of alarms that all come in and they all look like they’re the end of the world, but they don’t actually have very much data or context to them. Or you might have a series of technicians calling you, and some of them are able to yell louder than others or make their problems sound bigger than others. But again, you’re kind of operating on very incomplete information.

The best you can do is figure out what are the two or three things that I can try to triage and at least put a Band-Aid solution on today. Then you go home at the end of the day, not because you’re done or because you solved the problems, just because you’re too tired to keep going and it’s time to go home for dinner. You come back the next day and the list is even longer. It’s just day after day after day of that.You’re in this completely reactive stance where you wake up in the morning and you get hit in the head with a shovel, and then you just do your best all day to recover, and then you go back the next day and do it all over again.

That’s kind of the status quo where our customers are today. Obviously, I’m being kind of hyperbolic. But then there’s also the cost implications of that. It just costs way more money to be in that reactive stance than it does to proactively address these issues.

Then there’s the compliance element too, where there’s all these new regulations coming into effect. These businesses, many of them are big and sophisticated, but many of them are not. They’re trying to navigate this super complicated regulatory landscape. We have federal, and then you have a bunch of patchwork of state-level regulations. They’re like, what do I do to just make this whole problem go away? We’re able to kind of help in all those different areas.

There’s one other thing I wanted to add too, the most important piece, which is operations. When the cooling goes out, you can’t sell groceries and you have angry customers. You have yellow caution tape, you have carts full of spoiling food that you’re wheeling to the back of the store to put in the dumpster. You have a bunch of angry customers who are like, look, I came here to buy six things and now I can’t buy two of them. Now I have to go to a second store, and I’m not feeling so great about that. Maybe I won’t come back next week.

[13:06] Wes Ashworth

Huge, huge impacts there outside of just the climate discussion, so I always like to make that connection. Anything else, in terms of operators, investors, even climate insiders, are there any other common misunderstandings around the refrigeration challenge?

[13:18] Amrit Robbins

That’s a great question. I think that the most interesting part of this entire equation for me is just how challenging the problem is, but also how simple the solution is. The solution is very simply, part one, don’t leak as much of this stuff. How do you do that? You catch the leaks early, and then when you catch a leak, you fix it. Then part two is, when it’s time to replace that equipment, replace it with a different type of refrigerant that is not one of these terrible HFC refrigerants for the planet and something that is a much more healthy refrigerant, like a natural like CO2 or ammonia, for example, or propane. It’s really that simple.

The overall scope of the problem is gigantic, the impact is gigantic, but the solutions are just kind of super simple and easy to understand.

[14:01] Wes Ashworth

Which is the brilliance of it. Those are the problems you want to solve, right? Biggest impact with simplest solutions, in a way, which we will get into here in a little bit just in terms of what Axiom does, how that plays in, and how that can be the solution that they’re looking for.

I’ll ask you one more question before we move there. When you’re talking to that sort of retailer or facility operator for the first time and you’re kind of speaking their language, how do you help them see this as more than a compliance problem or maintenance problem? Any other insights or things you share with them and how you approach it?

[14:32] Amrit Robbins

It’s really kind of a bespoke approach where I think just about everyone out there realizes that the status quo is not going that well. There were a lot of people in our industry who moved up through the ranks as technicians and then moved into these leadership roles. Their entire life, their entire career at least, they’ve been steeped in this kind of reactive break-fix mentality because that’s what the industry has been for decades. It’s not their fault. That’s just the way that we have trained them and molded them as an industry for their entire career.

It’s really tough to kind of get through to those folks because a lot of times they’re in a, hey if it isn’t broke, don’t fix it type mode. Also a lot of these folks are towards the end of their career. The idea of doing something new and different and changing things, there better be a really, really good reason for that. Otherwise, I’m out of here in five years, it’s going to be the next guy’s problem. I don’t want to get fired before my pension comes due, that sort of mentality.

For those types of folks, typically we’ll say, okay, that’s great, let me know when you retire and we’ll talk to your successor. But for the other people out there, it really comes down to, it’s not does the pain exist? It exists for every single one of these folks. The question is, is the pain so severe and so unbearable that you are ready to take action, and not next month, but right now?

Because we often get this feedback from customers, which is, I don’t have time to do something new and different. I can’t afford to do that right now. I have too many other reactive emergencies on my hands. We’re banging our head against the wall saying, you don’t have time to not do this. You are just under this avalanche which is growing day by day, and doing more of the same is not working. What do you think is going to be different next year? Was last year better than this year? What do you think is going to happen next year?

That’s how we attempt to get through to folks. But ultimately, when folks come to us and they’re already feeling that pain that is so severe, they’re saying, I’m willing to try something, that’s when we actually have an opportunity to then show them, hey, there’s a different way to do this. Let’s copy all these other industries, which are already using time series data from their machines, are already using work order data, and then applying AI and analytics and US-based experts to do valuable things with that data and figure out how to manage these facilities smarter without working harder.

[16:43] Wes Ashworth

Absolutely. I appreciate that framing and kind of wanted to hear you meeting them where they are. But yeah, you’re right, so many industries are ahead of this curve and are already doing this. But some of those industries, like refrigeration and some others, are really just lagging. But as always, pain is a great motivator. Usually, to provoke true change, you do have to feel that amount of pain when you’re ready. If you can just say, yeah, I don’t know, I don’t like it, but I can deal with it, then usually that’s maintain the status quo.

[17:09] Amrit Robbins

For each one of these folks, their pain point is different too. For some of them it’s, hey, I’m losing a technician who knows these 20 stores like the back of his hand, and he’s got all this knowledge in his head that’s not written down anywhere. He’s going to retire pretty soon. What am I going to do?

Someone else says, hey, I’ve been trying to fill these three technician roles for six months, and I cannot do it. I cannot find the qualified people. I’m finding these very inexperienced folks who just don’t have the experience necessary to succeed right now.

For someone else, it might be, I’ve blown my budget every year for the last three years because we’re so busy putting out fires that it comes down to, do we sell groceries or not? Putting the budget aside week after week, month after month, year after year, and we’re blowing our budgets, and I’m getting in a lot of trouble because I own the P&L.

On and on, there’s so many more examples like that where folks are just stuck in these cycles, I call it like a hamster wheel basically. They’re looking for a solution. They’re looking for an off-ramp. It really depends. It changes from person to person. For a lot of these folks it’s maintenance related. For a lot of them it’s refrigerant leak related. For some of them it’s energy efficiency related as well.

[18:14] Wes Ashworth

Absolutely. Kind of going into that now, the solution, right? The off-ramp when they’re ready, and we’ll turn this to Axiom Cloud itself. Because this is where a hidden system problem becomes something measurable, actionable, and valuable.

For someone hearing about Axiom Cloud for the first time, what exactly do you plug into? What data are you reading? What decisions are you helping customers make better? In a nutshell, how does it work?

[18:39] Amrit Robbins

Axiom is the AI platform for the cold chain. We are focused really heavily today on retail grocery and cold storage, and specifically their refrigeration and their HVAC systems. HVAC is a more recent addition. Historically, it’s been almost all refrigeration.

The whole idea here is that there are already hundreds to thousands of real-time data streams that are being generated by your refrigeration controllers. That data is not being viewed by humans. It never leaves the facility in many cases, and it gets overwritten in just a few days in a lot of cases. There’s this gold mine of data which is just kind of ending up in the trash can.

Axiom is borrowing an approach which has already been successfully applied in almost every other adjacent industry. Very simply, it’s, look, let’s get the data out of those machines. Let’s clean it, label it, normalize it, put it in the cloud, make it accessible on any device. Then from there, let’s use AI, analytics, and US-based refrigeration experts to find what are the biggest problems and opportunities buried in that data for energy efficiency, predictive maintenance, and early leak detection. Then package these up into a way that even an early career technician has a fighting chance to succeed at resolving the root cause of these things.

Then from there, it’s helping the customer move from insights to action. Typically that means a great deal of handholding and meeting the customer where they’re at, and working within their existing operations, their existing tools, and their existing constraints to make sure that they’re able to resolve some things remotely from their desk, or they allow us to do it on their behalf. Then from there, triage out the things they don’t care about. Then from there, figure out, batch these by stores. Maybe there’s three or four issues at a single store. That doesn’t need three or four service calls. You can do that all in one in a lot of cases.

Then prioritizing based on urgency and value. Which ones of these issues out there are going to cause full store outages in the next two days? Let’s do those first. Then from there, hey, there’s a couple of lower urgency items. One is worth $5,000 a year. The other one’s worth $40,000 a year. Let’s do the $40,000 one first.

All that being said, our goal is to help the customer figure out what is that one service call that I need to dispatch today? If I have just a little bit of time, a little bit of money, what’s that one thing I’m going to do today proactively so that I can avoid this flood of alarms in two days or two weeks from now and save a bunch of money and a headache along the way?

Then from there, we help them secure the win. We help them get that win, and then that starts to build momentum. We’re able to help them move to the next and move to the next and build the trust with them, and then with their network of service providers and technicians.

I guess I’m going to take an opportunity to brag here. In 2025, we’re serving 12 big enterprise customers across around 1,000 accounts today, 1,000 sites. For every site, every customer, holistically, for every what we call anomaly, which is one of these actionable insights that’s packaged up really nicely for a technician to resolve or a maintenance manager, 70% of those, the root cause was resolved within 45 days of us sending it. These are proactive, predictive opportunities. For every 10 of those we sent, seven got resolved root cause. That is multiple orders of magnitude better resolution rate than the reactive alarms that our customers are being flooded by today.

Over the course of a month, these same teams might be getting 10,000 alarms and they’re resolving 1% of them in a timely fashion. We’re giving them, during that same month, 100 predictive anomalies and they’re resolving 70 of them before the flood of alarms associated with each one of those occurs. That’s how we measure success, and that’s kind of our North Star metric.

Helping our customers drive from insight to action, that is our secret sauce. That is where we really differentiate. That’s where our customers keep renewing and expanding with us. That’s how we’re winning new logos, too.

[22:16] Wes Ashworth

It’s incredible, great explanation. By all means, I was going to ask you, if you didn’t share, just some of those stories and the impact and things like that as well, too. Absolutely remarkable. I think we’re going to see this more and more across all of our lives and across the economy, being able to interpret this data, really have those actionable insights, and then seeing the value on the other end of it.But how important that is to this discussion and the climate discussion as well, too.

Let me ask you this. When you bring a new site on, is there a place that customer value shows up first? Is it energy savings, maintenance efficiency, food safety, refrigerant leak reduction, something else? Is it kind of all of those, depending? Where does it typically show up first?

[22:54] Amrit Robbins

What’s really interesting here is that refrigeration is the elephant in the room for so many different parts of the overall business model. It drives over half the energy, 90% of maintenance service calls. It’s the whole ball game for EPA and state-level regulation. Refrigeration all by itself drives over 50% of Scope 1 emissions, in many cases over 50% of Scope 2 emissions, in other words, energy cost emissions.

Then putting all that aside, the typical US grocery store spends more money every single month to power and maintain its refrigeration than it takes home in profit.

I’m setting the scene here to kind of answer your question more directly. If we are talking to, let’s say, the Director of Maintenance, that person really cares about maintenance P&L. They care about technicians and time to resolution and first-time fix rate and callbacks, et cetera, et cetera. Those are the types of metrics that we really focus on, and we have a module for that, the predictive maintenance module. We’re able to really focus on that one set of problems that that stakeholder cares about.

If it’s an energy stakeholder, then we focus on energy efficiency opportunities, of which there are unbelievable amounts of low-hanging fruit across the industry today. The short version of that is this is not necessarily, 80% of what we do is not rocket science when it comes to energy, just to dive into that for a second. It’s just simply getting a best practice in place and then making sure it’s not reversed two weeks later. Then constantly monitoring and playing whack-a-mole and keeping that in place every time to drive results. That’s 80% of it, honestly. It’s the simple plug and chug, eyes on everything all the time, fixing the same problem over and over again, adjusting that set point back to where it needs to be multiple times.

Then the other 20%, that’s where the really cool incremental AI-driven stuff happens that humans wouldn’t necessarily have the ability or expertise or capability or bandwidth to do.

But we won’t even get into that today, but there’s some really cool, exciting stuff there. Then if we’re talking to the compliance person, they care about, hey how do I reduce regulatory risk exposure? You know, I’m going to get fired if X, Y, and Z happens. How do we avoid that? Or I’m not going to get a bonus if none of those things happen. How do I avoid that?

It really depends on which customer stakeholder that we’re working with. We’re really able to kind of tailor the approach by turning on one or more modules to solve the problems they really care about.

What’s really cool is that we can get over 100 sites a week up and running after we have the IT interconnection. That’s the longest pole in the tent. The longest pole in the tent is getting that customer to approve our IT interconnection. After that, we can get 100 sites a week up and running without any site visits, any new hardware, any new sensors, fully remote deployment, and the customer sees measurable business outcomes in the first month of operations.

[25:29] Wes Ashworth

Absolutely incredible. I love hearing the different perspectives from those different stakeholders as well, too, and what they care about.

I’ll ask you this, maybe you already covered it, but what’s an example of an issue Axiom can catch that even a talented human operator might miss, not because they’re bad at their job, but just the economics of watching every signal, that’s just not possible, just doesn’t work?

[25:48] Amrit Robbins

Really good question. This goes back to kind of like 80% of what we do. A technician, like if you could hire a bunch of technicians, like a lot of technicians, and just have them look at all data from all sites all the time, they could do exactly what we do, 80% of it at least. The problem is no one can afford those people because that would require hiring hundreds or thousands of extra people. On top of that, they don’t exist. We have a massive shortage of qualified technicians today.

That kind of begs the question, how do we fill this gap? How do we support the existing technicians and enable them to do more with less because there simply are not enough of them, and we can’t afford to hire any more even if they did exist?

But then the other 20% is work that’s really exciting. One that I really like to talk about is leak detection. Basically what the AI algorithms are able to do. Our AI is, we’re a totally AI-native company. Everything we do is using AI models in order to, and these AI models are trained on over 2,000 historical site-years of data, showing them what does good, what does normal look like in terms of operation across a massive variety of different sites. Some are big cold storage warehouses, others are very small corner grocery stores, different climate zones, different ages of equipment, configurations, manufacturers, refrigerant types, everything you can think of.

We have a really wide data set that kind of sets the baseline. Then of course there are individual site-level baselines as well, because each site is different. Then on top of that, we have this second data set, which we’re very proud of, which is at this point over 125,000 anomalies. These are events which have occurred, specific events which have occurred in facilities that we’re serving. Each one of them is a problem or an opportunity related to energy efficiency, predictive maintenance, or early leak detection. For each one of them, we have what’s all the context leading up to that event, and then when did that event start, what was the root cause, when did that event end, and then what’s all the context after that event?

We have that hand-labeled and verified by a US-based refrigeration expert 125,000 times. That data set, that super specialized expert data set, has become a key data moat and differentiator for Axiom.

We didn’t realize this, but there are entire billion-dollar companies being built on this premise of like, hey, let’s have experts hand-label data in their field. Let’s get PhDs to do this, to help train AI models. We didn’t realize that. Maybe we should have done that in retrospect, should have become a data labeling company. But ultimately we did it because we had to.

As a result, now we have the ability to do things that humans just can’t. For example, we’re able to look at tens or hundreds of data streams coming out of a grocery store and look for tiny, tiny deviations in the data. If a human even had the time or bandwidth to look at this data, they would never even notice these deviations. They just look like noise.

The crazy part is that the AI is able to now say, hey, I’m noticing these tiny deviations in suction pressure over here. I’m also noticing this valve is open a little bit more than it used to be under similar operating conditions last month. This compressor is running a tiny bit less than it was last month. None of those things are even visible. But when you start looking at those and then taking them in their totality, that’s the signature for a leak.

Suddenly, now we know there’s a refrigerant leak at this site. By the way, that was a totally illustrative example. I’m not giving away any secret sauce by giving the real data streams. But that’s the sort of concept here, connecting the dots between really totally invisible and seemingly disparate data streams over time to find these kind of hidden nuggets of extreme value that are buried in the data, and no human would ever see because it’s not feasible.

[29:15] Wes Ashworth

Absolutely incredible. To your point too, even if you could hire a huge team of technicians and analyze all this data, they would still miss it. But also that’s next to impossible. I can tell you from the recruiting world, that is the most in-demand position in the country, probably. It’s got to be top five. Those people are aging out as well, too. There’s a huge amount of them that are retiring. That problem is going to get even worse.

We need solutions like this. It’s clear and obvious as you go through it, but you really spell that out really well. I guess what has been maybe the hardest part about getting this technology adopted? You mentioned sort of the IT side of the equation. It sounds like once you get that going, the rest is pretty smooth. But is it technical integration with that? Is it organizational inertia, trust, or the reality that refrigeration is simply not glamorous enough to command executive attention? What’s the hardest part about getting it going?

[30:07] Amrit Robbins

Well, you hit on a lot of different barriers that we’ve run into over the years. We’re about, let’s see, what, six years into this journey now. I’ve been in the space now for like 15-ish, and we’ve encountered almost every single one of those barriers you just talked about.

What’s really cool though is that over the years we’ve kind of managed to resolve a lot of them. For example, technical integrations now, we have this library of integrations which makes us compatible with essentially 100% of retail grocery facilities in North America. We’re able to build new integrations very, very quickly because we figured out how to abstract out multiple components of the integration and only build the new part that’s necessary for a new controller type, for example. Then of course, supercharging all that with AI means we can roll out new integrations in literally days.

Then there’s the question of the overall IT security and gaining trust with the IT teams at customer accounts, which remains a massive challenge. But we have iterated on that and worked so closely with some of the Fortune 50 companies out there to get through their processes successfully, even though it takes us, in some cases, six months, nine months, to get approval from the five different InfoSec teams and AI teams and VPN teams and everything else that’s necessary. But yeah, we’ve really solved that in terms of, can we succeed? It’s a resounding yes at this point. Sometimes it takes time still, depending on customer timelines.

Then there are so many other barriers too around, do our stakeholders feel enough pain at this moment to actually try something? I would argue that that is still a key barrier with a lot of our accounts. These are folks who are really not in the right mindset to try anything new because the pain isn’t great enough, or they can see the exit sign right in front of them and they’re just kind of holding on until the end of their careers.

Then I think another key barrier is basically, it’s kind of a blessing and a curse, honestly. It’s a double-edged sword here where our customers move extremely slowly and they’re very, very conservative. They don’t want to try new things, especially when it comes to back of house. Refrigeration is viewed as a cost center. It’s not viewed as a competitive differentiator in retail grocery. When you talk about refrigeration to the CEO, first of all, his or her eyes glaze over. Second of all, if you’re in a retail grocery chain, their guard is going to go up and they’re going to say, wait a second, are you going to ask me for more money right now?

That’s the type of reaction that our stakeholders are used to getting. Trying to get high-level executive buy-in from CFOs and CEOs for strategic refrigeration projects, it is an uphill battle when you compare it to a merchandising project. New end caps that have Taylor Swift on them, that’s going to get the CEO’s attention.

I think that’s another key barrier that we run into, changing the script on, this really truly is a strategic opportunity when it comes to the company-level P&L and company-level customer satisfaction and things like that, because the refrigeration is operating more of the time and you don’t have a terrible ESG record, and you’re not getting fined by the EPA publicly every two years, that sort of stuff.

I think those are some of the big ones. But the flip side of that, the momentum, the inertia, the lack of wanting to do new things, as soon as you do kind of achieve escape velocity with some of these really big customers out there, then there’s a lot of stickiness and there’s a lot of buy-in and then there’s a reluctance to change away from that as well. For us, it’s been a long and hard road with some of our big enterprise accounts. We gained traction with them, we gained velocity, and we’re making a lot of progress and growing very fast as a result of that. We’re supporting these customers at scale and changing this aircraft carrier now for a lot of our customers, I think, would be a very tall order and very painful.

[33:35] Wes Ashworth

I like that. I like how you went through each one of those, and kind of cool just seeing the positive side of that as well, too. These are some challenges, but also it’s positive once they do decide to go there, how it is stickier and you go there. I’m sure it helps too, the more you do it, the more that market recognition happens and they hear like, XYZ grocery chain did it, and kind of like, I want to be like all the other cool kids. If they did it, we should maybe pay attention to it. You probably get a lot more of that too, just that comfortability as it expands and more and more people are seeing the value of it as well, too.

Thinking about just you internally, how do you measure success internally? Is it avoided emissions, leak reduction, energy performance, customer ROI, uptime, or some blend of all of those?

[34:15] Amrit Robbins

Well, I’m looking over here because I have my 2026 goals posted on the wall in my office here. How do we measure success at Axiom? The number one thing is basically the return that we’re providing to our customers. We’re able to measure that because that’s what it comes down to at the end of the day, the business case. For every dollar that our customers spend on Axiom, how many dollars of value are we providing to them?

Our minimum line is $3, and today we’re well above that. We’re hoping to continue to increase that over time. It’s been pretty incredible to see the type of feedback we’re getting from customers, especially when they’re continuing to renew with us and expand with us and try new services and expand to more and more sites.

The compliance element has also been really interesting as a key driver of new conversations because folks realize, holy crap, I have to do something. I better start at least learning about what the problem is and what my options are. But we’ve never closed a sale, never closed a new customer account, based on compliance alone. It still always comes back to the business case. It’s like, okay, now I understand there’s a problem, I need to solve this, but I’ve got a bunch of different options out there to solve this. Axiom is not the only one in our field. We have competitors. How do I select the right solution for my business? That comes back to the business case.

[35:28] Wes Ashworth

I think that mix is so important. The strongest climate-based businesses, renewable businesses I see, are proving that impact and economics can reinforce each other, and it always comes back to that business case, as you said.

Let’s stay with the technology side and talk about AI a bit, because that’s an area where the gap between useful deployment and market hype can be enormous. You were applying AI long before ChatGPT made the term unavoidable. What were you doing then that people today would still recognize as just real, practical AI?

[35:57] Amrit Robbins

We are a vertical AI company today. I’m really proud to say that we were in the right place at the right time because we’ve been branding ourselves as an AI company since 2020, which is before ChatGPT in 2023 caused this huge, huge change in the technology and humanity landscape for the future and market disruptions, et cetera.

We’re very excited to kind of be early adopters of AI from that sense. Now really we’re positioning ourselves as the AI leaders for our customers when it comes to the refrigeration and HVAC asset classes. It’s really cool to see executives reach out to us and ask us for advice or market landscapes and feedback as a result of that thought leadership and that leadership position that we’ve built over the last five or six years.

Ultimately, again, we’re a totally AI-native company. Before the large language models and Gen AI were released in 2023 and became available for use, we were using previous-generation AI, which is deep learning, deep neural nets, essentially like machine learning plus plus plus, I guess you could say, which AI still is, but it’s gone even further.

We started building up that massive data set, that labeled data set, which we’re still growing and still improving over time. We use it to spin up new AI models on a weekly basis and test them, have them compete against one another. Now we have ensembles of AI models, some old-school machine learning, some new Gen AI, and they’re all competing for the best outcome, and we select the best. In other cases we have ensembles of models which are working together, where there are agents essentially trying to answer different parts of the question or trying to solve the same problem in different ways and then selecting among themselves which is the best answer.

We’ve really leaned incredibly heavily into the new AI technologies that are out there, but we’re still leveraging the best of what used to be there before Gen AI as well. But yeah, we are truly an AI-native company.

What I think has really differentiated Axiom over the years as a key takeaway is that we’re taking the AI technology stack and all those best practices and we’re combining it with the white-glove approach that our customers really need to translate the data and insights into actions and measurable business outcomes. Again, that’s Axiom’s key differentiator.

Even if we did have a team of analysts behind the scenes and there was no AI happening whatsoever, but we still had that white-glove element of supporting our customers every step of the way and helping them drive results, arguably we’d still be focusing on the most important problem, which is how do you help customers solve problems? That’s not the case, by the way. That was a hypothetical. We are definitely a very AI company. We don’t have teams of thousands of analysts manually reviewing data. That was just an example.

[38:36] Wes Ashworth

Absolutely. It’s cool. I think the historical perspective is valuable, using this prior to the dawn of ChatGPT and all these other things that people are now, everybody’s paying attention to. Industrial AI has been generating real outcomes long before the hype that was out there.

You obviously, thinking about AI, you seem enthusiastic about AI. AI is a part of your everyday world and what you’re doing. I love the evolution of it, you saying new models even weekly, how quickly it evolves. I guess where do large models help your business today, and where maybe are they still not the right tool?

[39:11] Amrit Robbins

We’ve heavily adopted Claude recently. That’s been kind of the most recent change in our organization over the past three or so months. It’s so deeply ingrained in every part of the business at this point, we’re using it for everything. It is just astounding what we can accomplish. We have a small team and we’re able to punch well above our weight class for a lot of reasons. But this is now becoming a really key driver of enhanced productivity. We’re able to do things we just couldn’t do before, and we’re able to accomplish so much more now. It’s astounding.

But then on top of that, when you’re thinking about how we deliver results to customers, the AI is an incredibly important element of what we do, but it’s only half the picture. The other half is still that we have US-based refrigeration experts who are there every step of the way. We have project managers and operations managers who are there every step of the way to make sure that we’re helping our customers drive that timely anomaly resolution rate, which again is our North Star metric. Because it doesn’t matter how good the AI is, if you’re not helping the customer drive from insights to actions, then there’s no value and there’s no company.

[40:12] Wes Ashworth

Great points there. Important distinction too in terms of when that real human expert is still needed. We’ve said that a lot of times, AI won’t replace you, but somebody that knows how to use AI probably will replace you. It’s good to embrace it. The term co-pilot, all those kind of things, you still need that expert that’s kind of in the driver’s seat and delivering for the customer and those sort of things as well.

[40:33] Amrit Robbins

Blindly trusting AI is a fool’s errand at this point. We cannot blindly trust AI. Even the best models out there, with all the best safeguards in place, you still need to have those experts in the loop every time, without question, full stop.

[40:59] Wes Ashworth

Absolutely, I’ll leave that right there, you’re spot on.

Has the current AI boom helped open doors with customers at all, or has it sometimes created confusion because everyone thinks AI is magic now? What’s that been like?

[40:59] Amrit Robbins

We haven’t seen really significant effects directly when it comes to customer conversations or how we support customers or what kind of questions they’re asking us or objections they put forth. But we definitely anticipate that you see all these trends in other industries, and we definitely anticipate that those are going to start to filter into our space in the coming quarters and the coming years.

We’re doing our best to prepare for those, how to answer those questions, how to help our customers navigate this changing landscape, and how to better serve our customers as their expectations change. But we haven’t really seen it yet, honestly. We still have customers who have controllers on dial-up, just to put this in perspective.

[41:36] Wes Ashworth

Right. It’s coming, still a ways out before they catch-up. It’ll happen.

I appreciate you going through that. Thinking about AI, you’re in it, you’re living it every day. You’re seeing the evolution. You’re seeing these new models constantly upgrading, changing, and modifying. What are you most excited about for the future of it? Where do you think it’s going to go next? Anything on the horizon that you’re excited about in terms of AI and being able to help this issue, climate, energy, with all these sort of things?

[42:02] Amrit Robbins

I think that there is just such a massive amount of low-hanging fruit opportunities out there still. In our space, there are these refrigeration systems which you can’t possibly have eyes on all the time and manage intelligently using humans. It’s just not feasible. I think you can apply that same exact concept to so many different other asset classes, industrial machinery, data centers, which is where it’s definitely already happening, automobiles, healthcare assets.

There’s just so many other industries where you can start to apply a similar playbook where it’s already being applied in some ways in different shapes and forms. But there’s just so much more opportunity there too, where you start, you think of an asset as something that can be optimized if you apply the best AI models to it. It’s just astounding what’s possible in a lot of cases in terms of reducing how much energy it consumes, reducing how much refrigerant it emits, improving the outcomes of what it’s able to accomplish for its users, et cetera.

[42:55] Wes Ashworth

Huge opportunity, obviously exciting times to come, a lot to look forward to on the horizon as we continue to integrate this more, and it gets better and better and better as we go.

Let’s close by zooming out a bit because the long-term story here is about more than one company. It’s about how climate markets mature under pressure. There’s a strong argument that periods of policy or market disruption can ultimately force stronger companies and stronger solutions to emerge. Why do you believe that?

[43:21] Amrit Robbins

We’re definitely in kind of a profound moment right now for the industry where a lot of the incentives and the regulatory protections and the government support for offshore wind farms and electric cars and solar and all these different technologies are being pulled. The rug is being pulled out from underneath all those industries right now.

It’s really interesting to see how the industry reacts because this is inevitable, right? It’s going to happen eventually. Any industry that is reliant on these types of market-warping policies is not a sustainable, self-sufficient industry until they’ve proven that they can live without those.

Now this is one of those moments. Could this have been done a little bit more gracefully or given folks more time to prepare? Probably. But ultimately this is going to shake out in a way that shows us who are the companies who are truly delivering sustainable value here on a free-market basis. If you can make the argument that what we’re seeing today is a free market, I don’t know, that’s a different conversation.

But yeah, I think that we’re about to see a real kind of culling event where a lot of companies don’t make it, and the ones that do are going to emerge a lot stronger. And then we’re going to be able to allocate resources more effectively to those companies that do have a really clear path for massive impact in the future.

[44:33] Wes Ashworth

Absolutely. It’s a sharp lens for the moment we’re in because resilience is one of the defining tests for a lot of climate and energy companies, so I agree completely. What gives you confidence that climate and clean energy can become a more obvious, bipartisan, and commercially durable story again?

[44:49] Amrit Robbins

I think just kind of simple physics and economics, honestly. We’re at the point now where doing these things, it’s just such an absolute no-brainer. Like, hey, I have this fleet of refrigeration assets. Things are going really poorly and they have been for 10 years. Should I do more of the same, or should I use hugely forward AI technology to dramatically change how I manage these systems? It’s a no-brainer. Plus it pays for itself in the first month. How is this a question?

It is a question still for a lot of folks, which is the head-scratching part. And that’s our job, to help those folks really overcome those objections and show them examples and get them comfortable with this potential future that we envision and we’re already showing to a lot of our other customers.

[45:31] Wes Ashworth

I love that. It’s hard to argue with physics and economics. It’s our job to tell the story, to inform, to educate as well, too. I absolutely agree with that.

Final question. If we’re sitting here five years from now, what would have to be true for you to say that the refrigeration sector finally got the attention it deserved? Maybe it’s longer than five years, but what would that future look like?

[45:49] Amrit Robbins

Well, I’ll tell you, I don’t think it’s going to be five years. I think it’s going to be five months here until we have some really exciting news coming out here in terms of the really, really big players in the space adopting this, jumping in with both feet. The tipping point moment is here, I think. When some of the really big players make their moves and jump into this with both feet, not only will the other folks be forced to follow, but there’s going to be a lot of pressure for them to do it fast.

We’re very excited to be, ideally, in the right place at the right time with what we believe is the best solution in the space to help our customers navigate this transition and begin benefiting from what’s out there today in a way they haven’t historically.

[46:30] Wes Ashworth

Exciting to say that, and even as short as five months away, I didn’t expect you to say that. I won’t hold you to it, but on the short, short kind of near horizon, that’s incredible, just what that opportunity would be and the potential there. So excited to watch that.

[46:34] Amrit Robbins

Don’t quote me on that number.

[46:45] Wes Ashworth

Amrit, this was an absolutely fantastic conversation. I really love episodes like this because they expand how we think about the energy transition and remind us that some of the biggest climate opportunities are not always the most visible or obvious ones. Thank you for joining me on Green Giants and for sharing that perspective on commercial refrigeration, refrigerant emissions, and the practical role AI can play in solving those real industrial problems.

To everyone out there listening, thank you for being with us. If you enjoyed this episode, subscribe, leave a rating and review, and share it with someone working in climate, energy, buildings, food, retail, or industrial operations. With that, we will see you next time.