Welcome to the Design to Product Podcast!
In this interview episode, we talked to Hall Chen.
Hall is the co-founder and CTO of Gridware, a startup that builds a fascinating device that can prevent wildfires by early detection. We talked about the challenges of building such a device, dealing with supply chain shortages, domestic vs offshoring, how the device works, the different technologies they use, and more.
Today we talked about:
- Gridware’s journey
- Supply chain challenges and how to deal with them
- Domestic vs offshore in manufacturing or supply chain
- How to treat vendors, suppliers and supply chain partners
- How to run a pilot program
- Testing efficiency of a wildfire prevention device
Adar: Our guest today is Hall Chen. Hall is the co-founder and CTO of Gridware, a wildfire prevention device. And they are also a Y combinator alumni or batchmate, this is how I met him. Hello, Hall.
Hall: Hey Adar, it’s great to see a friendly face again. Super stoked to be here and kind of share some of my thoughts.
Adar: Yeah, I’m pretty excited, too. We didn’t talk a lot about last year and where you’ve been heading, so I’m very curious to hear. Maybe you’ll start with just talking about the journey of Gridware.
Hall: Yeah, absolutely. We really kind of started really just with the idea that if you could deploy a network of low power, low cost sensors to sparsely populated areas in the Wildlander urban interface, what would that allow you to do in the context of preventing wildfire ignitions? Being able to detect and localize different types of electric grid failures and faults that can lead to wildfire ignitions or even be able to predict them by looking for the kind of precursor signals. And so this is what we set out to do. And in the past year we’ve gone through several major hardware revisions and are now deployed with utilities and pilot programs to essentially field test our system capabilities. And throughout this entire time, you know, it’s for sure been an interesting challenge working with various different hardware vendors, grappling with these supply chain issues that I’m sure everybody in this industry has been dealing with.
Adar: Yeah, even the larger organizations constantly deal with this trouble. So what are your current challenges and how do you deal with them?
Hall: Yeah, so you know, designing for long term support is absolutely a struggle because at any given time you might only be able to procure certain critical electrical components in quantities of hundreds, like high hundreds, maybe a low thousand, but it’s very difficult to get guaranteed supply of thousands or tens of thousands of components. And so often we find ourselves essentially having to design our hardware in such a way that we’re flexible to minor component replacements here and there. And this is on the electrical side, you know, on the mechanical side. International trade, I think, and international logistics have definitely taken a hit from COVID and just labor shortage challenges. So on that front, you know, we try to keep a lot of our manufacturing, a lot of our mechanical bomb supply to domestic sources as much as possible you know, that way, at least if there are any challenges or issues that come up, we can kind of personally go there to the supplier and help resolve issues. But regardless, the lead time on, say, things like enclosures or custom gaskets are still high, kind of in the order of two to three months or something thereof. And that really ultimately sets the cadence for how quickly we can iterate on our revisions. We definitely want to build as fast as possible and learn lessons and incorporate those lessons into new designs. But if we have to make any physical changes to the form factor at all times
Adar: It takes so much time to get that part
Hall: Exactly, exactly.
Adar: And I think you mentioned domestic versus international, probably minimizing potential issues where you deal with the domestic supplier, there are less potential of trade issues, customss, stuff like that. So what were your considerations between doing it domestically versus outside? Offshore?
Hall: Yeah, so a couple of things that we’ve seen go wrong and also have just anecdotally heard go wrong when we’re trying to do manufacturing or supply chain across international borders on a very short timeline. One is just if the vendor you know, deprioritizes your PO because they have bigger fish in the pipeline, then it’s really hard to kind of convey urgency over email or over phone. Usually an in person visit is much better for this, and international travel is difficult right now. So in many cases, that’s just altogether not possible. Another thing is just getting through the docks and getting things unloaded off of containerships, you know, a lot of the items that we procure cannot be air shipped and come across the Pacific Ocean. And I think the situation is better now, but last year there were giant queues at the loading ports in LA, here in the Bay Area, in Oakland, literally, we had a park that was on a ship already in the Bay Area, but could not be unloaded for an entire month.
Hall: And so you know, the physical goods were ready here, but we just couldn’t get to them.
Adar: Yeah and why by ship and not by air, by the way?
Hall: These were very heavy components. They were battery cells, and I think there were some shipping constraints as well.
Adar: Yeah, got it. So very interesting. So, that made you make the decision probably for now when you’re still iterating a lot on the products, it’s crucial for you to get the parts faster, being able to cut the lead time as much as possible and not deal with all these logistics problems. Do you plan on changing that down the road when you perhaps have a more stable concept and probably, like, want to set up the supply chain, start thinking costs?
Hall: You know, it’s certainly in our minds. It’s certainly in our minds something that we think about and we review kind of on a regular basis. But you know, I am optimistic about just domestic manufacturing capacity, being able to, A, meet our demand and B, meet our cost requirements. You know, I think we are slowly seeing a shift, particularly in the United Statesto overtime, essentially build a stronger emphasis on being able to produce atoms within US Borders. So, we are very optimistic about that and we intend to really roll with it and take advantage of it, yeah.
Adar: Sounds good, yeah and probably it will be best if you can stick with the same supplier you’ve been working with for prototyping. That has advantages too.
Hall: Yeah. Yeah, 100%. You know, the contract manufacturers and distributors that we’re working with now, we’re very open with them and you know, we tell them, like, right now we’re really starting out kind of with low hundreds of quantities, maybe low thousands. But we’re looking for the right partners to kind of, A, help us grow and B, that we can really grow with us, when we reach commercial scale. And they’re of course, very happy to hear that. And you know, they’re able to provide a lot of value with their expertise.
Adar: Yeah. And that probably helps you with overcoming the fact that you don’t have a lot of negotiation power you mentioned, like suppliers with bigger fish to take care of. So, perhaps like, making sure that you give them a very detailed plan on how you will grow in the future and how many units you will be able to give them to produce. Maybe it will make you look more attractive for them and that transparency is important for the future.
Hall: Yeah, 100%. We treat our vendors and our suppliers like our teammates. Anything that I would do for a colleague, I grew where I’m more than happy to do with a supply chain partner. The same standards of work and collaboration hold internally and externally.
Adar: So important. Treat your supply chain partners, you call them like partners, like teammates. This is so not obvious, but so important. It seems like you also run a pilot program. How do you build it, how did you decide to do it and how does it go?
Hall: Yeah, absolutely happy to speak more to these pilots that we’re doing. So, they’re really great for two things you know, one, on the business development side which I can speak to first. It’s not trivial thing for a utility to fill one of these pilots with new grid technology, especially for a system like ours when the utility puts real labor and real management resources and also, dollars behind fielding our system. This is, you know, a very serious commitment and a very serious statement from them that they want to pursue what we have built and evaluate the capabilities for commercial scale. And so, from a business development perspective this is really a strong signal that we’re building the right thing. That we’re building a technology that these utilities care about, that can provide them with a very strong value, very kind of tantalizing capabilities. On the other hand, you know, we’re largely building a system here where our primary and most immediate mission is to A, stop wildfire ignitions from happening by detecting the grid faults that lead to wildfire ignitions and being able to localize those and B, be able to essentially monitor wildfire ignitions near grid infrastructures in the earliest stages. And so being out there in the real world, in the field deployed for 12 or 18 months in a pilot, it’s just a tremendous opportunity for us to actually be providing value and reducing wildfire ignition risk for our customers and partners here.
Adar: Yeah, it’s interesting to understand, like, how do you measure the effectiveness of wildfire prevention because essentially you prevent wildfire, you don’t know if you did that right? or do you like, take past cases and measure like, how did you reduce them? How do you do that?
Hall: Yeah, that’s a great question, you know. To start off, a wildland fire ignition is a well studied topic so, you know, the great thing is we do generally know how they start, how they spread, what conditions allow them to proliferate really quickly, how we should be gauging our risk of out of control wildfire and so, we know all of this. And we know, kind of, the relationship between these wildfire ignitions and the electric grid failures. The type of faults that can happen on the electric grid that can lead to wildfire ignition. And so these are the faults that we target at Gridware, these are the faults that we detect and help localize in order to provide the most value to our customers which are the utilities. And so for example, if a tree branch falls onto the power lines, it sits there for a minute or two expelling moisture, but then eventually it combusts and when it does it can fall to the ground. And if there’s a lot of fuel around, if it’s windy and dry then you have a really fast spreading ignition event. And if we’re able to detect essentially in the wild this event where a tree branch breaks and then falls onto the power lines, then that is essentially a direct demonstration of a potential wildfire ignition event that was caught and actively monitored for, kind of hazardous evolutions. yeah.
Adar: Right. So basically like sort of an AI that identifies patterns that signal that there is going to bea wildfire and alerts on time. That’s basically how it works. Do you have false alerts that happen or do you learn from that as well?
Hall: Yeah, I mean what you described is basically the gist of it for sure. Having really low false positive, really low false negative rates is something that we care deeply about especially because we don’t want to kind of be the people who cried wolf and constantly send out alerts that are not real. But at the same time we don’t want to miss anything that might be very dangerous as well. So to that end, it’s really about striking a fine balance there and we are a little bit more conservative in our stance here. We would rather be overly sensitive. And so right now you know, we still put a lot of energy and effort into essentially verifying that the alerts that we receive are the exact types of alerts that can provide value. But the kind of insight that we derive from these alerts and tabulating these patterns over large time horizons, it kind of leads to this build up in this library of understanding that we’re then able to bake into the decision making engine of our technology.
Adar: Very interesting. What is the point where you are satisfied enough with the product and its performance with your pilot program and you’re saying, okay, now is the time for us to start setting up a supply chain like, go to producing more units, start selling. Like, what’s the point where that’s what I need, I’m ready right now.
Hall: Oh, yeah. I mean, I suppose to be clear, that’s already happening. We’re already kind of dedicating efforts to securing a larger scale supply chain because we do see this on the immediate horizon. For sure, as a startup, you don’t wait for
Hall: Yeah, exactly. You don’t want to wait. And we’re always looking around the corner to figure out what’s coming next to making preparations for that.
Adar: Nice. So, maybe I could ask that in the past tense. What made you decide that right now it’s good enough and you can keep going.
Hall: Yeah. At the end of the day, the value that we’re able to provide with low power, low cost sensors that can detect these faults even when the electric grid is off, this value is unique and this value is not currently, these capabilities are not something that currently exists. And so I would say that we are ready, kind of gone from zero to one. And that is, yeah, I would say, how we draw that line.
Adar: Yeah, that’s super exciting yeah, and right now, maybe we can talk a little bit about your supply chain challenges and things that you anticipate. Like, did you already start \ building a supply chain team? How does it look like from your side?
Hall: Yeah, so right now kind of manufacturing responsibilities definitely fall under the engineering side of the company and you know, I definitely chipping myself, especially with procurement and coordinating logistics with various different vendors and distributors. We don’t have a dedicated manufacturing team or dedicated supply chain team at the moment. Definitely something that we’re going to be building out in the short term, but not something we have immediately.
Adar: Yeah, very cool. And what kind of, I guess, components do you need to make that device work well? Like, what kind of things are the most interesting components that go into your device?
Hall: Yeah, so you can kind of break this down categorically and I would say that the major categories are our power, so these are things like the solar panels, the batteries, and then there’s the compute, so things like microcontrollers or arm, CPUs, memory, things like that. And then there’s sensing, so all the different sensors that allow us to be able to capture the signals that we care about. And of course communications, the ability for devices to communicate back to the club, those are kind of the most critical categories.
Adar: Fascinating. Hall Chen, co-founder and CTO of Gridwear. Thank you for being with us today. It was great. Fascinating.
Hall: Yeah, thank you, Adar.