Congratulations, you’ve made it to the 21st Century! We’ve got smartphones, smart cars, and smart mouths, but no smart batteries. Unlike our other tech, batteries lack any ability to think and they, as we have all experienced, die way too quickly. Solving this powerful problem is Watt-Learn, who just emerged from the Pittsburgh-based accelerator AlphaLab. Watt-Learn is using machine learning to give batteries the ability to think independently. So, when you need to manage power for things like getting the most out of your home’s solar + battery or have sufficient energy during a thunderstorm, batteries can make it happen without help.

To find out more about this innovative tech, we interviewed Watt-Learn’s CEO, Matt Maroon, and CTO, Matineh Eybpoosh.

Listen to the Interview

Read The Full Interview

Mr. Maroon:  Watt-Learn, we are a cloud-based intelligence software company. We are focused on adding a brain to battery systems. So, we are looking at the world and how much renewables are being added to the grid, and right behind the renewables is energy storage; in order to balance out supply and demand, and keep prices where we like them. But, there is a gap in terms of how those batteries operate. And so we’re filling that gap with our artificial intelligence platform.
Matt Maroon, CEO of Watt-Learn.        

Mrs. Eybpoosh: I am Matineh Eybpoosh, CTO of Watt-Learn

Ben: First off, thanks  

Mr. Maroon: Absolutely

Mrs. Eybpoosh: Yeah

Ben: It’s early on a Thursday, I think. It’ a busy week. You’ve got demo day. So, how are you feeling? How’s demo day feeling?

Mrs. Eybpoosh: I’m… he should be…

Mr. Maroon: No, it’s fine.

Ben: Yeah?

Mr. Maroon: Yeah, I mean, we know the story, and the story is compelling. We’ve obviously been practicing it as we’ve been talking to customer and investors. So, today is just an opportunity to do it in front of a few more people in the room then what we typically do. But, the story doesn’t change.  

Ben: That’s wonderful. That’s wonderful. Is there anything on the market that is similar, and if so, how is yours different?

Mr. Maroon: So, there are some companies that are doing what we call measure and respond approach, so they look for a specific signal and then they respond to it. They are very reactive. And the challenge that those softwares have is that they are unable to optimize both the life of the battery and the value generated. And that’s what’s different about us. We are proactive; we use machine learning to forecast market and grid conditions and prices. We forecast the help of the battery and any on-site loads or needs for power. And we use all those proactive forecasts in order to maximize both the life and the value generated. So, that’s the big difference.

Ben: It’s like you’re telling the weather. Can you talk about the smart part of your software?

Mrs. Eybpoosh: Yeah, sure. These batteries, in order to make smart decisions they have to act like a smart human. They have to know, they have to have a foresight in all the areas they care about, that affects their performance. Right? For example, what is going to be the PV generation on-site for this house I am connected to or what is going to be the load of this house? Or, if they are providing some grid services what is going to be the prices electricity or what is going to be the regulation signals for the next day. And, they also have to be aware of what is going to happen in the battery as a result of any decisions that they make. So, that is a foresight that is a smart battery they have to have. That is where machine learning comes in. We use machine learning to give these batteries foresight for all these factors that are very important. And then we give them a decision-making ability, an optimal decision-making ability. So, they get this foresight and we run optimization models through this foresight and make the batteries decide, what is the best thing for me, considering all these factors and considering what I care the most about. What is the best thing to do for this battery? So, basically making the battery think like a smart human.

Ben: So, you’re saying it’s learning because it’s accessing the various factors. Is it taking previous learned information, such as from a database or a previous client or a previous test and using that information to access those variety of factors?   

Mrs. Eybpoosh: Yeah, in most of the areas that we use machine learning we use historical data to learn from. There are some aspects of it that we don’t have historical data, so we use other methods; unsupervised method that is so call in machine learning domain. But, yeah, we learn from history and try to predict the future.

Ben: So, you take the data that has already been recorded for energy fluctuations…

Mrs. Eybpoosh: For example, yes. What has been the prices before, what has been the battery operation before, or what has been the degradation before, and what is it going to be in the future?

Ben: So, where exactly or specific is this data coming from? Is it from public or is it from private? Or where is it coming from?

Mrs. Eybpoosh: Most of the data we use is either publically available through the website of these deregulated markets in North America or can be purchased through third parties. For example weather data, there are so many third parties that provide weather data. Or some of them are customer data. So, some of the customers that they are interacting with, they provide historical data, they’ve already collected this data, and they provide us, and we use them.

Ben: Well that is great, we brought in weather.

Mrs. Eybpoosh: Weather is always important.

Ben: Weather is always important.

So, you can see, let’s say, that an area has a lot of power outages, so you would use that information and pre-program your battery to think about these things so that it knows in this season, ‘hey, there is going to be a thunderstorm…

Mrs. Eybpoosh: Yeah, that is a scenario, that is a possible scenario. If, we know things are going to go bad and the battery can do something about it, and doing that something about it contributes to achieving the objective of the battery, whatever it is. Whether it is increasing the revenue of the battery or increasing the health of the batter or decreasing the carbon emissions for the battery. Whatever it is that your objective is for the battery. So, we take those projected and predictive conditions and make the battery act accordingly.

Ben: And then on the other end of it, you’ve had the data coming  in, how are you auditing its performance? Are there specific metrics that you’re gathering that are programmed in that it is recording, so it says xyz what is it doing?

Mrs. Eybpoosh: Yeah, no forecast is perfect. Right we are going to wrong in our forecast and the battery is going to have to; so there are these policy functions that no matter what we say, they take into the account, the current real time conditions of the battery and make a decisions accordingly. And we have the report of those. We know what the batter has done. First of all, our algorithms learn from those historical performances and improve So, what’s called active learning. And at the same time, we check those and we are like, ‘okay, in this kind of situation it seems like we were very off and we make sure we turn our algorithms accordingly.  

Ben: Oh, well that’s pretty cool.

I’d like to get very specifics about your product. Can you explain what a Proactive Lifetime-Aware Automation System is? Or play, plays?   

Mr. Maroon: plus!

Ben: Plus!

Mrs. Eybpoosh: Plus!

Ben: Okay, I messed that up already. Do you mind explaining what that is?

Mrs. Eybpoosh: Yeah, do you want to go?

Mr. Maroon: Yeah, sure. So, the PLAAS platform is our primary product that we sell to customers. It’s a subscription based model.  So, if you are a customer and you own and operate a battery system, whether you’re a homeowner or a utility company that has a massive battery system; what you would do is sign up for an annual subscription we take all the data that Matineh discussed, right. And run it through all the forecasting engine and the optimization engine. And what we are able to do is provide you with day ahead instructions on what to do with your battery tomorrow. So, give you the optimal way to charge and discharge your battery to maximize whatever objection you’re trying to reach. Whether that’s the lifetime of the battery, the amount of revenue you’re generating, or cost savings, or whatever the case maybe.So, the PLAAS platform is the core software product that we sell, sell a subscription to, we don’t, right.  

Ben: Now, let’s talk about application. You said home, and you said city. For the consumer both the at home and the city, what is this battery doing and what are they doing that they would use this battery for?

Mr. Maroon: So, let’s take the residential for example. It does depend geographically where you are at for what you are asking the battery to do. But, the typical homeowner in California that has a battery in their house likely also has solar panel on the roof and so what they are able to do with the battery is capture any extra solar that they generate during the day stored in the battery and then discharge it from the battery to household loads later on at night. So, they are trying to get themselves independent from pulling electricity off the grid. So, in residential application, that’s mostly what the battery is used for. It’s used for backup power when the grid goes down. But, in general, it’s used to offset how much electricity you pull from the grid itself.

Ben: And for municipalities, the same?

Mr. Maroon: For a utility company its actual different. So, there a dozen different applications that a battery can do in a grid-connected utility application. But, the simplest way of thinking about it is that the battery is a buffer between supply and demand. Because electricity is one of those things that has to be constantly balanced in real time. The amount of supply electricity always has to equal that of demand. And batteries are flexible resource that sits between the two. It allows you to, almost some wiggle room. So, what the batteries do is provide overall, increase the robustness of the grid; the overall reliability, and they can allow for more renewables to be put on the grid. So, in general batteries add to the decarbonization, if that’s the proper term…

Ben: That sounds like a good term.    

Mr. Maroon: …the decarbonization because they allow for those renewable resource to continue to be added in mass.

Ben: And that was my next questions, which was about sustainability. And you mentioned solar and you talked about, what was the great word..

Mr. Maroon: …decarbonization, if that is a word.

Ben: Sure, today it’s a word. So, I am an owner. I’ve got great solar stuff that is working during the day. What other, both on residential and grid do you think sustainability wise could come out of this product?

Mr. Maroon: Again, we’re providing the intelligence and, kind of, the look ahead in terms of here’s the power that you’re are going to need, what’s the best place to get it from.

One of the cool things about the platform is today we are working on batteries, but it can be used for any dispatchable asset. So, our platform could be used, for instance, when and where to turn on your water heater. Right, when is the proper time, lowest cost, to turn on your water heater, for instance. Or adjust your smart thermostat inside your house. So, any of those different elements that can be controlled would benefit from foresight and from an intelligence software platform, kind of, optimizing the right time to turn things on and turn things off. So, you could envision in the long run, with more renewables, with more batteries, with the proper intelligence on it that we would be able to reduce the overall amount of electricity that homes need to operate. Because you are picking the right time; for instance, why have your water heater running all day long when no one’s at home. If you have the platform that knows and can project, ‘hey, here is when we need hot water,’ you can just have it ready at those times, for instance.

Ben: And this is good for, you know, commercial, residential or commercial huge facilities. So, if you are a developer and you know that the building is doing to have a bunch of rental units, this could say you money?

Mr. Maroon: Yeah, could be used for commercial apartment buildings, could be used for commercial and industrial buildings as well. Anywhere, where you’re using a variable amount of electrical during the day would benefit from intelligence to predict what’s going to happen and thus smooth it out over time.

Ben: Do you install something to help you track the data?

Mr. Maroon: That’s one of the things that is super exciting. We don’t need any additional hardware. There is no additional integration of hardware or complicated systems that have to be installed. So, we’re simply, as Matineh said, pulling data that is either publically available or the user has the data themselves. So, we don’t need anything additional to what the user already has.

Ben: Wow, that is super cool, so you don’t have to install anything. So, how did this project come about?

Mrs. Eybpoosh: So, the initial, actional inventor of the idea is another Ph.D. alumni from Carnegie Mellon University, Julian Lamie. And his research was in renewable and energy storage during his Ph.D. It was during his time as CMU, although not directly related to his Ph.D. but during his time at CMU, he realized working with these companies and energy storage utilities. He realized okay, it seems like what is missing is these batteries, with wide deployment, these batteries are not smart, for the job they’re signed up for. So, he came up with the conclusion that we need artificial intelligence. He designed this platform that would help these batteries to be smart. And that was when myself, Julian, and Matt came together to make the power team to address this challenge.

Ben: I really like that the power team.  

Mrs. Eybpoosh: Yes, we are the power team.

Mr. Maroon: We’re going to get buttons made up.

Ben: Cause it is software could it be used in other applications. Could you use this for another consumer based product like handhelds? So, that someone that is building these mobile devices could track how users are…

Mr. Maroon: …interact with their device through the course of the day. Conceptually, yes…

Mrs. Eybpoosh: Depends. Depends. For our cell phones, our usage continues, right? We don’t want to our usage of a cell phone some point because it makes more sense for a particular reason.  So, you have to be able to use this cell phone no matter when, during the day. And you charge it when you need to. So, it depends on the application. There are more consumer based applications that can benefit. For example, I can think of the these batteries that are attached to these boats. People have these boats; they have these batteries. They care about these boats a lot. And they want – these batteries are expensive investment, for example. And they want these batteries to last as long as possible. So, how should they optimize usage of those? Right now, I can’t figure out in my head how for cell phones we can apply this.

Mr. Maroon: Electric vehicles could be a huge market for us though.

Mrs. Eybpoosh: Definitely.

Mr. Maroon: So, one of the dreams that everyone has always had when we finally move to electrified transportation, we will have all these batteries sitting around that will be plugged in; charging or plugged in, sitting around. Could the grid operators use the grid balancing, that I was describing earlier, acting as the buffer? So, this is a dream that everyone has, but it takes massive intelligence to be able to aggregate all of these individual assets and operate them as one unit. And at the same time, if, when I am fortunate to buy a Tesla, and I go outside and my Tesla is charging I want to be able to get in it and drive. I don’t want to have the battery run all the way down because the utility company has been using it all day. So, there is a lot of intelligence and a lot of considerations that go into making, what’s called the vehicle grid infrastructure a reality. That could, very much, be a future path for us as we continue to develop and as more and more electric vehicles hit the road, could be huge market opportunity for us to jump into.

Ben: Is that along with the same thing as battery degradation. Batteries have a lot of problems keeping the energy for a multitude of reasons; a lot of them are what we can make and the environment. Will your software be addressing that or looking into that?   

Mr. Maroon: That’s one of the core tenants of what we are doing. So the practical example is right you get the new iPhone and today brand new out of the box it last eight hours before you have to charge it. But, two years into it, you’re getting three hours, and you’re desperately trying to find a plug at the airport. That’s because every cycle you put on a battery degrades it a little bit further and a little bit further. Our premise and the premise behind the PLAAS platform is that not every cycle generates the same amount of value. So, you want to be smart about the cycle you do. Because every cycle degrades the battery, but not every cycle contributes to the financial bottom line. So, what we are doing with the platform is only picking the right time to cycle to make sure that the value we generate outways the cost of that degradation of the cycle. So, we are using software to manage when you are using the battery in order to manage the natural electrochemical degradation that is going to happen.  

Ben: Today is demo day as of when we are recording this. How do you see your company growing? What are the next steps? Today, you are presenting, what are the things you need to make me have a more efficient Telsa.

Mr. Maroon: So, I will talk on the commercial side, kind of next steps. And we can sort of go through product platform as well. On the commercial end, we’ve very focused right now on finding the right first set of customers. We do have a subset of customers that we have been talking to now for a couple of months. In fact with one of them, we are probably a week away from, probably, having a first project with. So, these three customers actual cover all of the different facets of where our platform is initially targeted. One is residential energy storage systems. We have a customer we are talking to right now that is focused exclusively on large utility scale systems. And then we have an end user – owner – operator battery. They currently are operating have a dozen energy storage systems in the United States. So, we’re working with all three of those as, kind of, our first launch deployments over the course of the next couple of months. That’s obviously critical for us to have successful projects; demonstrate that to the rest of the market before we expand outward from there.

Ben: Right, so you can buy the Tesla.

Mr. Maroon: That’s the plan.

Mrs. Eybpoosh: That’s the goal.

Mr. Maroon:  That’s the goal.

Mrs. Eybpoosh: Right now we are focused on two markets: California ISO, independent system operators in California, PJM territories which is mid-Atlantic area. So, we are focused on PJM and California ISO. Our goal is to expand across North America first. Cause these deregulated markets in North America, each of them have different rules. Each of the operate differently. So, a battery doing the same job in the same side of PJM, doesn’t operate the same way as California. So, that is right off the bat we can’t cover every market out there. The goal is to expand our platform so that it covers all these deregulated markets in North America and then, of course, expand internationally from there. Because there is huge energy storage market in Europe, Asia. And also expand other applications from the metered batteries that are providing grid services to behind the metered batteries that are not connected to the grid that are providing on-site services. And also to aggregation of, what we talked about, distributed resources. Right now we are working with residential resources, but it can be any kind of resources like Matt talked about; electrical vehicles or any other distributed resources. So that is, technologically speaking how we are planning to expand.

Ben: Do you have a lot of competition?

Mr. Maroon: Right now we’re really at the forefront of using this machine learning-based approach for managing batteries. So, our goal is to maintain that advantage that we have and to move quickly into the market place. I would say what we are doing would augment anyone how is doing the traditional kind of measure respond approach to managing batteries would benefit from our platform. So, people that may consider themselves competitors to us could actually benefit our software package as well.

Ben: How do you see the energy market changing over the next 10 to 20 years? What will change?

Mr. Maroon: So, I think we are only going to see more and more energy storage added to the grid. So, similar to the way that solar took off over the past 5 years as prices came down, batteries are projected to do the same thing. By 2025 the forecast is a 25 billion dollar marketplace for energy storage. And as we continue to add more and more renewables, I mean, you have to have batteries to manage the overall grid. The market is going to do nothing but grow, and we’re really at the forefront a very very quickly growing market place.  

Ben: Nice. So, let’s just talk a little bit about money. Because everyone wants to know how this will save them money. We talked a  little about residential, we talked about commercial, and details about the grid. Is there something else specific that would help with overall cost savings?

Mrs. Eybpoosh: The whole premise of our platform is that, hey, we know this is all about money, and we know energy storage projects are really expensive and the financials don’t pencil out. Right now if the subsidies are not there and the incentives are not there, it’s hard to make sense of economics with a lot of these battery projects. The cost of hardware, battery hardware, has dropped significantly through the past decade, but it has also slowed down. Because there is only so much you can do to reduce the cost of the hardware. So, the solution is on software to make these batteries smarter, so that they can make smart decisions in the right time. And our platform is based on real case studies that we have done, can increase the lifetime of batteries around 50%. This is because we increase the life of the battery around 30, 35 percent. So, the longer your battery lives, the longer the life of the battery and the more return on investment. Unless money is not your objective that is also possible. We are not just addressing that. If you care about other aspects of your project, for example as I mentioned briefly if you care about carbon emissions of your project that can be your main objective. That may not increase your revenue and financials, but this is also something we can address.

Ben: It’s a big day. Is there anything else you would like to mention to those you are just discovering your product?

Mrs. Eybpoosh: Buy it!

Ben: buy it.

Mr. Maroon:  How’s the salesman of the two of us? Well, done.

I think just in general with any startup a lot of it is about timing. Right. Again, as we just talked about, the timing is right now for energy storage. These projects are happening. It’s not just some esoteric thing that people are talking about at Universities. There are real homeowners buying batteries. There are utility companies and large building owners and operators that are buying batteries to manage their utility bills. We are excited about the opportunity; we know the product works and right now heads down focused on expanding the products. Where it works geographically and really finding those first early adopter customers.

Ben: Well, I want to thank you both for your time today. It has been really cool to learn about. I’m excited and I am going to buy a bunch of batteries. So, thank you both.

Mr. Maroon: Thank you.

Mrs. Eybpoosh: Thank you, thanks for the time. 

---

To learn more about Watt-Learn visit: Watt-Learn.com

The interview, podcast, and video were created for the Owl Me Not project run by Ben Wonderful. The Owl Me Not project seeks to simply explain that which is unknown, whether it’s an idea or forward-thinking business. Owl Me Not has partnered with IHeartPGH to produce interviews and videos on topics related to Pittsburgh, including a series on innovators.  The project is currently in development, but the latest video can be found here. Feel free to reach Ben about the Owl Me Not project via email.