WEBVTT
00:00:00.160 --> 00:00:04.480
It's a bit reductive to think EV charging reliability is just a hardware problem.
00:00:04.719 --> 00:00:06.879
It's not, at least not primarily.
00:00:07.040 --> 00:00:09.119
The bigger issue is operational fragmentation.
00:00:09.279 --> 00:00:12.960
There's too many systems, too many stakeholders, too many gaps in accountability.
00:00:13.199 --> 00:00:19.839
In this episode, Hillary Taylor, CEO of What's Up, breaks down why uptime isn't just about fixing broken chargers faster.
00:00:19.920 --> 00:00:25.920
It's about better data, better workflows, and giving technicians the right context before a truck ever rolls.
00:00:26.079 --> 00:00:31.920
AI won't fix broken operations by itself, and if we don't solve that, scale really won't matter.
00:00:32.159 --> 00:00:33.359
Let's get into it.
00:01:01.759 --> 00:01:02.320
What's up?
00:01:02.479 --> 00:01:03.439
Clever play there.
00:01:03.759 --> 00:01:04.959
How are you doing, Hillary?
00:01:05.200 --> 00:01:05.680
I'm good.
00:01:05.840 --> 00:01:06.239
I'm good.
00:01:06.319 --> 00:01:06.719
How are you?
00:01:07.120 --> 00:01:07.599
Doing well.
00:01:07.760 --> 00:01:09.280
Thank you for sharing some time.
00:01:09.599 --> 00:01:17.120
We tried to get you on the episode or on the show, uh seems like a little bit, but stars wouldn't align, but finally it's come together.
00:01:17.200 --> 00:01:18.799
So I'm I'm glad it it did.
00:01:18.879 --> 00:01:26.400
And uh, you know, we appreciate the uh the the connection between Field Advantage and what's up, and we're glad to have you on the show.
00:01:26.480 --> 00:01:28.640
And and uh let's just get into it.
00:01:28.799 --> 00:01:41.200
So I'd uh I'd like to start off the conversation with you walking us through your personal origin story, share with the audience what pulled you into the EV charging space uh, I guess over a decade ago, but now it's been a long time.
00:01:41.680 --> 00:01:45.359
You're engaging me, you know, giving me away.
00:01:45.680 --> 00:01:49.680
But yeah, so right out of college, actually, I'm a Georgia Tech grad.
00:01:49.840 --> 00:01:58.319
I graduated with a degree in electrical engineering, and I went straight from being a college student to working in a late-stage startup called Whitricity.
00:01:58.719 --> 00:02:04.239
So that that was my entrance to vehicles and specifically electric vehicle infrastructure.
00:02:04.319 --> 00:02:14.560
And this one was very futuristic in that it was wireless charging of electric vehicles, which I still think is on the horizon in terms of technology deployment for us.
00:02:14.719 --> 00:02:21.039
But from there, I went to Spark Charge, uh, which is mobile charging of electric vehicles.
00:02:21.199 --> 00:02:25.599
And that's where I got the first opportunity to lead an engineering team.
00:02:25.840 --> 00:02:36.159
So I've always been on the technical side of the EV charging game, where, and that's where I built mobile grid list uh DC fast chargers.
00:02:36.319 --> 00:02:41.039
And then from there, I went to megawatt level charging in new.
00:02:41.360 --> 00:03:10.879
But where where it really clicked for me that uh there was a reliability issue that as a hardware provider, you know, I was contributing to an ish, uh a big issue that was global, was a a drive from Boston to Buffalo, which should have taken seven to eight hours, but it ended up taking 17 hours and I was running a fever and I was in a Chevy bolt, which at the time you couldn't charge the Chevy bolt due to uh uh fear of the something happening with the battery.
00:03:11.120 --> 00:03:14.719
Um, so you can only charge it to 80% and it was 23 degrees outside.
00:03:14.960 --> 00:03:24.240
So it was just a nail biter the entire way, and my drive effectively doubled because the chargers we were encountering along the way were not working.
00:03:24.400 --> 00:03:41.120
And that was the click for me that there is something going on here that's really serious with the infrastructure and how I mean it it really put things into perspective for me, then as the driver, how can we ask others to experience something like I just went through?
00:03:41.280 --> 00:03:45.759
And that's where, you know, the seed was planted and what's up was eventually born.
00:03:46.000 --> 00:03:46.159
Yeah.
00:03:46.560 --> 00:03:52.159
So what year would that trip have been if you were in the in that model of vehicle that could only go to 80%?
00:03:52.560 --> 00:03:53.199
What year was that?
00:03:53.599 --> 00:03:57.439
That was like late 2022, probably, or early 2023.
00:03:57.840 --> 00:03:59.919
Um and it was still cold in Boston.
00:04:00.080 --> 00:04:03.120
And then in 2024 was when What's Up was founded.
00:04:03.360 --> 00:04:04.240
Yeah, cool.
00:04:04.800 --> 00:04:12.800
Well, let's uh let's talk about, you know, you've kind of already alluded to it, but you know, what you saw in the field that led you to to building what's up.
00:04:12.879 --> 00:04:18.240
But what, you know, what where does What's Up sit in the ecosystem and what problem is it solving?
00:04:18.480 --> 00:04:18.639
Sure.
00:04:19.040 --> 00:04:19.839
Share that with the audience.
00:04:20.160 --> 00:04:20.399
Sure.
00:04:20.560 --> 00:04:21.040
Yeah.
00:04:21.360 --> 00:04:31.519
So I think fundamentally that the position we take with what's up is that there's a fundamental gap and layer missing in the ecosystem.
00:04:31.839 --> 00:04:35.120
So we're not integrating into an existing layer.
00:04:35.279 --> 00:04:51.519
We are a new layer that sits up top of the ecosystem and solves the reliability issue plaguing literally everybody on all layers, the software, the existing software layer, the existing hardware layer, the technician layer.
00:04:51.759 --> 00:04:54.240
We are the unifier to all of that.
00:04:54.480 --> 00:05:01.279
And with that, with that unification and standardization powered by AI, we will solve it for everybody.
00:05:01.600 --> 00:05:01.839
Yeah.
00:05:02.000 --> 00:05:06.319
So, you know, I understand what WhatsApp does for the audience.
00:05:06.480 --> 00:05:17.839
I I would just, and I don't want to oversimplify this, so you correct me if I'm if I'm wrong, Hillary, but but your your platform is not a replacement to the EMSP platforms that maybe CPOs are operating right now.
00:05:17.920 --> 00:05:20.160
It's it's a it's a supplement to that.
00:05:20.240 --> 00:05:28.959
It is it is uh and and and it's bolstering some of the gaps uh outside of just a standard CMS out-of-the-box CMS platform.
00:05:29.279 --> 00:05:29.519
Right.
00:05:29.920 --> 00:05:41.040
So not a replacement to a CMS, you're bolstering what CMS is are are trying to do in the space now as as maybe a layer below and a layer above, and maybe even a layer in between.
00:05:41.199 --> 00:05:41.519
I don't know.
00:05:41.600 --> 00:05:48.959
I don't I don't want to oversimplify it, but I want to kind of connect the dots for the audience that I don't can understand exactly and and we'll we'll get into it.
00:05:49.439 --> 00:05:56.800
Obviously, we'll we'll we'll break it all down, but uh uh is that a is that a good good breakdown of of the I I I think so.
00:05:57.040 --> 00:06:00.560
I think we are a connector in a fragmented ecosystem.
00:06:00.959 --> 00:06:01.360
There you go.
00:06:01.839 --> 00:06:03.839
We do we do three things, right?
00:06:04.079 --> 00:06:08.959
We bring proactivity to a very reactive ecosystem.
00:06:09.279 --> 00:06:24.319
We take action, and then that action is a continuous improvement mechanism where every future action is always smarter, better, faster, and will lead you to just increase success.
00:06:24.720 --> 00:06:25.279
Yeah.
00:06:25.839 --> 00:06:26.720
Good, good.
00:06:26.959 --> 00:06:41.600
You know, in your in kind of your catalyst, the experience on the drive that you mentioned, uh you know, it's it it signals something that was happening then, something that was happening before then, and something that really is happening today, and that is public charger reliability.
00:06:42.240 --> 00:06:48.480
And the data around reliability is is really uncomfortable when you look at it from an industry standpoint.
00:06:48.720 --> 00:06:56.399
You know, 78% up time, 22% non-functioning, flatline, just all of these things, these data points.
00:06:56.560 --> 00:07:02.160
And so from your vantage point, what is the what is the root cause of some of this reliability problem?
00:07:02.240 --> 00:07:08.879
I mean, you know, there's more than one, but what you know, kind of high side, what is where where is that reliability problem happening?
00:07:08.959 --> 00:07:10.879
Is it is this a hardware issue?
00:07:11.199 --> 00:07:13.279
Is this a workflow fragmentation?
00:07:13.360 --> 00:07:15.920
It is all of the above plus what's that look like?
00:07:16.160 --> 00:07:16.879
Yeah, yeah.
00:07:16.959 --> 00:07:20.879
And I think it's important to understand it's it's a compounding problem, right?
00:07:21.040 --> 00:07:26.959
So, you know, there are hardware failures, there are going to be software failures when you combine them.
00:07:27.120 --> 00:07:32.560
There are in environments and likelihoods where they will bring out the worst in each other, right?
00:07:32.720 --> 00:07:41.040
But the deep, I think the deep rooted, most painful part of this is the lack of standardized operational strategy and workflow.
00:07:41.199 --> 00:07:50.639
So we need to accept that this is a fragmented, I think there's 90 different hardware brands and 40 different software, you know, brands to pair with that, which is wild, right?
00:07:50.800 --> 00:07:54.319
We need to accept that this is a fragmented ecosystem.
00:07:54.560 --> 00:08:13.040
Now, how can we standardize the operational action workflow that sits on top of that and ensure that with every anomaly and every issue and every service call, there is a paired and strategic operational outcome with that need?
00:08:13.279 --> 00:08:13.519
Yeah.
00:08:14.079 --> 00:08:14.399
Yeah.
00:08:14.879 --> 00:08:18.560
Gosh, you know, the those stats are scary when you when you think about it.
00:08:18.639 --> 00:08:28.160
The the the the quantities of hardware, the quantities of software, multiple payment platforms and payment processors, and so many fail points in all of that.
00:08:28.319 --> 00:08:34.960
And and yet in in that in just that space, just that large space, there there is truly gaps.
00:08:35.120 --> 00:08:37.039
There's a there's an accountability to gap.
00:08:37.120 --> 00:08:43.120
And so when when we as a field service company, field advantage out in the field, we we get we we experience that firsthand.
00:08:43.200 --> 00:08:47.120
But the operator, the customers, the owner operators of these of these sites are dealing with it.
00:08:47.200 --> 00:08:56.080
And so between the the charger OEM and and the CMS and you know the site host that is just there, kind of captive to the process, whether they own the hardware or they don't.
00:08:56.240 --> 00:08:59.679
They're they're a part of the process and the part of the value chain and the field technicians.
00:09:00.000 --> 00:09:04.960
Accountability and and and ownership of the problem gets gets lost.
00:09:05.120 --> 00:09:06.639
It it gets lost in the noise.
00:09:06.720 --> 00:09:08.159
It's and it's not always clear.
00:09:08.240 --> 00:09:15.759
And so, you know, you're building a platform and you have a solution that is is kind of closing those accountability gaps.
00:09:15.919 --> 00:09:22.639
Where from your perspective, where does responsibility break down in in the model that we have today?
00:09:22.960 --> 00:09:23.679
Yeah, yeah.
00:09:23.919 --> 00:09:25.039
Really good question.
00:09:25.279 --> 00:09:28.879
And just going back to yeah, there's so many different stakeholders here.
00:09:29.120 --> 00:09:32.799
And I think having grasping that that is also like a fragmented.
00:09:33.039 --> 00:09:44.799
So we're we're just stacking fragmentation, honestly, on top of each other and acknowledging and realizing that with any fragmentation is a natural gap forms, whether you like it or not.
00:09:44.960 --> 00:09:50.399
And I I think the breakdown is that the problem for people lives in different tools.
00:09:50.639 --> 00:09:54.960
So the the problem for a site operator may live in the CSMS.
00:09:55.039 --> 00:10:01.919
For a technician, it'll live in Salesforce or whatever asset management or field service tracking system, right?
00:10:02.000 --> 00:10:04.799
For a warranty provider, it's living somewhere else, right?
00:10:05.120 --> 00:10:18.080
I think creating this unified and standardized source of truth, no matter the hardware brand, software brand, site operator, technician, a unified source of truth where everyone is communicating in.
00:10:18.320 --> 00:10:29.360
And despite communicating in differently, the way that it's communicated out via our platform, for example, is standardized to everyone, you start to bridge and close those gaps.
00:10:29.600 --> 00:10:29.840
Yeah.
00:10:30.159 --> 00:10:31.600
That's that's good.
00:10:31.679 --> 00:10:42.639
And that the closing of those gaps is is key because there's so many different, not only the fragmentation of the, of the, of the product, but then also how we as a field service company are responding to the needs.
00:10:42.879 --> 00:10:51.759
We get, like you said, the example of whether it's Salesforce as the as the ticket managing platform or Zendez Zoho, what whatever it may be, there's multiple platforms.
00:10:51.919 --> 00:11:05.200
And so we're we're we're coming into the service need and we're reacting to it, but there is sometimes as as a as a field service provider, we we are captive, a part of, you know, part of that bottleneck.
00:11:05.279 --> 00:11:13.679
And I know that in our conversations in the past, as well as just what I've seen from presentations, you've cited kind of vague work order development.
00:11:13.840 --> 00:11:20.399
So basically curating the work on the front end, developing that scope of work and what the SOPs, what the field service procedures.
00:11:20.639 --> 00:11:29.759
And so I've I've heard you cite that vague work order development is and and too much on-site troubleshooting is a part of that butt bottleneck in in service work.
00:11:29.919 --> 00:11:35.600
And so what information needs to be established before the truck even leaves, before the truck even rolls?
00:11:35.759 --> 00:11:38.559
What are kind of some of the fundamentals that needs to be established?
00:11:38.799 --> 00:11:42.559
Yeah, I I think that uh I think fundamentally it's like a mindset shift, right?
00:11:42.799 --> 00:11:49.840
So the technician, and this is how I kind of went into this and how we designed WhatsApp's platform, which is a two-sided platform, right?
00:11:49.919 --> 00:11:56.720
So we cater to the site operator, but also the technician with the idea that the technician is the sleeper, you know, agent.
00:11:56.879 --> 00:11:58.960
They're the most important part of this problem.
00:11:59.120 --> 00:12:05.200
So why are we giving technicians instruction instead of context and instruction?
00:12:05.440 --> 00:12:07.039
So yeah.
00:12:07.200 --> 00:12:17.519
So why aren't we contextualizing the things that we're telling people to do out in the field so that there is a not a better understanding of why they're doing what they're doing?
00:12:17.840 --> 00:12:25.440
Because I I think fundamentally technicians are capable of making really smart decisions because they're the ones, they're the only ones looking at the problem.
00:12:25.840 --> 00:12:34.080
So, with that, I think uh the the platform, right, has been designed to provide all of the history with with respect to this charger.
00:12:34.240 --> 00:12:39.919
Any sort of previous work orders on that charger, the technician should have access to.
00:12:40.000 --> 00:12:47.120
So an entire it's basically like a library or archive of all the work that's been done.
00:12:47.200 --> 00:12:51.919
And if there's repeat errors, the technician should know about that before going on site.
00:12:52.399 --> 00:13:00.240
All the contextual data that was flagged at the time of the work order creation, like why did this get created?
00:13:00.399 --> 00:13:02.720
I think the technician should have insight into all of that.
00:13:02.960 --> 00:13:27.440
Of course, we don't want to overload them with inundation of information, but we want to do it in such a way that they can, you know, very quickly obtain a contextual view of this piece of hardware and not just see it in a moment in time and then allow them to also have influence in in what is is done when they get on site instead of just someone telling them what to do.
00:13:27.840 --> 00:13:28.159
Right.
00:13:28.399 --> 00:13:29.200
Absolutely right.
00:13:29.360 --> 00:13:35.679
Because again, going back to the comments around all the stakeholders and all those involved in in processes.
00:13:35.840 --> 00:13:48.720
So you think about like an OEM who has updated a field service procedure for how maybe a particular component is replaced or how a particular component is troubleshooted in the field, and maybe the field service procedure has changed.
00:13:49.039 --> 00:14:04.960
And so that that those predictive insights that you're you're talking about and is is key to really cost containment and influencing overall the the success of the site and and the opex for the for the site host or the owner operator.
00:14:05.600 --> 00:14:08.879
Predictive maintenance has has really been aspirational.
00:14:08.960 --> 00:14:14.000
And so I like how you're you're you're you're putting guidelines around the information set.
00:14:14.159 --> 00:14:23.679
You're saying context, but then also uh historical data, but not just a data dump that is is so much you don't even know where to start because it's so much information.
00:14:23.840 --> 00:14:32.480
And so when when you're looking at predictive work versus reactive work and kind of the economics in there, how do how do you operationalize that?
00:14:32.559 --> 00:14:36.080
Or and how does WhatsApp help its customers operationalize that?
00:14:36.159 --> 00:14:43.279
So so predictive insights are being used and and maybe false starts aren't aren't a part of uh the process anymore.
00:14:43.519 --> 00:14:43.840
Yeah.
00:14:44.080 --> 00:14:47.360
So I I think fundamentally, and this is something we learned over time.
00:14:47.519 --> 00:15:05.120
So WhatsApp started just as a pure predictive maintenance platform, but we realize that predictive analytics and data in this ecosystem don't solve everything and and aren't super useful unless there is a very prescriptive operational action to go along.
00:15:05.279 --> 00:15:08.879
So I think that's the emphasis in and that's what we've learned over time.
00:15:08.960 --> 00:15:21.360
And that's really how the technicians started to come center, front and center in in the way that we approach solving this really big issue is there needs to be a very strategic operational workflow paired.
00:15:21.519 --> 00:15:35.360
And then using the technician feedback loop to ensure that the economics of what we just told them to do makes sense any future time that something similar arises in terms of anomaly or problem set.
00:15:35.759 --> 00:15:36.000
Yeah.
00:15:36.159 --> 00:16:01.840
Yeah, because they're you know, back to my my comments around OEMs maybe changing or updating rather a field service procedure for how something is if that if that's not ingested um in into a platform, I mean that that that knowledge is living with the OEM, but maybe it hasn't transferred out across all platforms, or maybe maybe they've shared that with the the owner of the of the hardware, or maybe it's available in some place, but how does the technician access that?
00:16:02.000 --> 00:16:13.759
And so it's it's getting into that predictive maintenance and and and and kind of building building that that that fence around this information, or at least a gateway for that information to come inside the fence.
00:16:13.840 --> 00:16:14.240
I don't know.
00:16:14.320 --> 00:16:23.360
But I know I know your platform leans into machine learning, uh machine learning driven insights, because to your earlier comments about learning and relearning.
00:16:23.440 --> 00:16:30.559
And so it's it's it's it's compounded problems, so you're compounding the knowledge base uh to exp you know to to prov provide that solution.
00:16:30.720 --> 00:16:35.440
What what signals are you extracting from the CMSs?
00:16:35.840 --> 00:16:39.279
Or maybe what are you seeing in the CMSs that are missing that are creating that gap?
00:16:39.440 --> 00:16:46.159
How is how are you leveraging AI to extract and and build upon what maybe CMSs are missing?
00:16:46.480 --> 00:16:49.919
Yeah, and and I think the word context will come up a lot here.
00:16:50.080 --> 00:17:01.519
It's context to the technician, it's context around the piece of hardware, because I think for the most part, right, coming as a ex-hardware designer, we design a lot of these things in the lab.
00:17:01.840 --> 00:17:10.079
And the the user, the variation in user interaction and environment in which this thing could live, although you test for it, it's kind of infinite.
00:17:10.240 --> 00:17:25.519
You could put this thing in an environment where it will be used and abused by, you know, a a driver because they've had a bad day, someone who is, you know, pissed off every day after work and just like drops this thing on the floor every time they try to charge, right?
00:17:25.599 --> 00:17:26.400
You don't know.
00:17:27.359 --> 00:17:31.839
So it's taking that sort of environmental context into play.
00:17:32.000 --> 00:17:36.400
So everything we layer on to just OCC, OCPP data.
00:17:36.720 --> 00:17:39.279
We do integrate with the hardware as well when we can.
00:17:39.519 --> 00:17:50.880
It's not a game changer if we don't get that though, but we layer on weather data, just general weather data, weather event data, utility data, extreme weather events, and lightning strike data.
00:17:50.960 --> 00:17:55.279
So we have insight into literally everything that's going on around a charger.
00:17:55.440 --> 00:18:02.000
And that really enables us to distinguish whether it is the charger itself that is misbehaving.
00:18:02.240 --> 00:18:04.240
Is it what's going on around it?
00:18:04.400 --> 00:18:07.200
Did the Transformer just experience a lightning strike?
00:18:07.279 --> 00:18:10.079
Or is the grid experiencing a lot of lightning strikes, for example?
00:18:10.960 --> 00:18:16.240
We've been able to avoid a lot of truck rolls just because we're seeing like brownout behavior.
00:18:16.319 --> 00:18:17.920
That's not due to the charger itself.
00:18:18.000 --> 00:18:19.839
It's due to what's going on in the grid.
00:18:21.279 --> 00:18:32.960
So yeah, I think where the machine learning with us comes into play is contextualizing and looking at the charger holistically in the environment in which it has been installed.
00:18:33.119 --> 00:18:49.519
And then it is in finding nuanced patterns within that data set that over time actually are incredibly significant to the expected lifespan or predicted failure date of that asset.
00:18:49.759 --> 00:18:49.920
Yeah.
00:18:50.240 --> 00:19:06.319
Going back to your your comments about the the technicians being a part of the communication and information exchange loop, your platform is leveraging that that field feedback as a part of the infrastructure for your solution, for your product.
00:19:06.559 --> 00:19:21.279
And so you're you're positioning technicians' feedback as that validation layer that is enhancing what the AI is learning and what what the machine learning is doing, you're taking and leveraging that feedback from the field and fortifying it.
00:19:21.519 --> 00:19:34.960
How do you how do you structurally how do you structurally prevent bias or noise or poor inputs from maybe from the field that aren't mess necessarily a line and and those could degrade the model accuracy?
00:19:35.359 --> 00:19:36.960
So how how do you sort through the noise?
00:19:37.119 --> 00:19:38.400
How do you how do you do that?
00:19:38.720 --> 00:19:49.359
Yeah, and it it has everything to do with the line of questioning that we have strategically put in place for technicians in order to in order to ensure that there is very little noise.
00:19:49.440 --> 00:19:52.880
If they're very targeted questions, they're all standardized.
00:19:53.200 --> 00:19:56.799
And via that data set, we can pull keywords.
00:19:56.960 --> 00:19:58.640
So we do a lot of work on the back end.
00:19:58.799 --> 00:20:02.480
So and it it's actually at this point, it's voice to text for them now.
00:20:02.640 --> 00:20:05.200
And but still with the strategic line of questioning.
00:20:05.279 --> 00:20:15.279
And then we do a lot of work on the back end to pull key phrases and key things they said with and we remove all of the what could be interpreted as bias, right?
00:20:15.359 --> 00:20:17.039
So we're looking for very specific things.
00:20:17.119 --> 00:20:23.839
We do a lot of work in the model on the back end to pull out those specific things and then feed it into the training data.
00:20:24.079 --> 00:20:36.640
So I I think it's it's about treating the feedback as data infrastructure and being very firm on some of the things that you're looking for in order to label data sets and then retrain.
00:20:37.200 --> 00:20:41.680
What What's Ups platform is is obviously hardware open.
00:20:41.839 --> 00:20:49.440
It's operating across different ecosystems with different hardware brands and other other software stacks and other and multitude of CMS platforms.
00:20:49.599 --> 00:20:53.680
But um, you know, this kind of goes back to the fragmentation part of our conversation.
00:20:54.240 --> 00:21:09.440
The market is uh is is shown that chargers, charger platforms are unique and they they behave differently, not only as a as a brand and as a product, but then when you introduce different environments, cold, hot weather income locations, et cetera.
00:21:09.599 --> 00:21:23.279
So what uh you know what what has been the hardest layer that you've seen that that is needs to be overcome to standardize across for your for your clients, for your customers who may be operating multiple hardware and multiple software stacks.
00:21:23.519 --> 00:21:30.240
How how do you what's the hardest layer to standardize without losing the nuance about each system operates and consequently also fails?
00:21:30.559 --> 00:21:30.799
Yeah.
00:21:31.039 --> 00:21:38.880
I think the most, you know, and and I should know this already just come coming from with this with OCPP so much.
00:21:39.200 --> 00:21:56.799
Like I think the surprising one that I would have expected to move a little bit faster in development and have provided a little bit more guidance to how to get rid of I I think when everyone implements OCPP, it's like they implemented the to the standard, but it's to their own interpretation.
00:21:56.880 --> 00:21:58.960
So you get different different flavors.
00:21:59.039 --> 00:21:59.759
It's like cocoa.
00:22:00.079 --> 00:22:02.720
Coca-Cola, Coca-Cola cherry, Coca-Cola vanilla.
00:22:03.039 --> 00:22:03.279
Right.
00:22:03.440 --> 00:22:14.720
So I I don't know if one is harder than the other, but the one that I would have expected to I I think root out a little bit of the flexibility within the standard is OCPP.
00:22:14.880 --> 00:22:17.359
Like I think that that would help a lot of people.
00:22:17.599 --> 00:22:26.079
I think that it it still is it's lacking in in some areas, but I think those are solvable with context and adding data layers over.
00:22:26.319 --> 00:22:31.440
But the one that I'm most surprised with is OCPP and how that's given people pain.
00:22:32.000 --> 00:22:32.400
Yeah.
00:22:32.720 --> 00:22:39.039
And yet, and yet that is is kind of table stakes around the the system as a whole.
00:22:39.279 --> 00:22:46.319
You know, obviously for the for the audience listening, you know, OCPP is a converse is a is an acronym you hear often on the on this show.
00:22:46.559 --> 00:22:52.640
Open ChargePoint Protocol, it's really the fundamental language uh of communication be in in the infrastructure.
00:22:52.960 --> 00:23:07.680
But um, you know, kind of kind of zooming out here and and just you know maybe looking at the market as a whole and our industry, service level agreements are our competitive edge in this, um at least from the from the owner operator standpoint.
00:23:07.839 --> 00:23:16.480
They're looking because at this point, many have deployed chargers, they've they have multiple hardware brands that they've worked worked with and and have operating.
00:23:16.640 --> 00:23:24.160
And so kind of the next frontier for them is operation operationally security around operations.
00:23:24.400 --> 00:23:30.799
And so uptime is obviously the the next brand differentiator in in public charging.
00:23:31.200 --> 00:23:31.519
Yes.
00:23:31.920 --> 00:23:44.160
But how do you how do you see your solution or and predictive man maintenance in in general evolving in into a lever around SLAs rather than just controlling costs?
00:23:44.319 --> 00:23:48.240
Like in other words, it's kind of going back to to the to the learning model that you were talking about.
00:23:48.640 --> 00:24:09.920
Yeah, I I think that we could be we could be the lever or the the knob that allows people to start guaranteeing uptime and feel feel more confident about making that guarantee because uh we have machine learning models that have proven themselves to be continuously improving over time.
00:24:10.160 --> 00:24:25.759
I think that we we play very well into a risk mitigation model where we start enabling flat rate fees for service and SLAs while still be able still enabling, excuse me, high uptime guarantee.
00:24:26.079 --> 00:24:38.799
Well, if we were if we were to kind of zoom out and and with a kind of a five-year perspective, or maybe even you know 2030, that seems to be uh a waypoint everyone's marching towards, or at least been a waypoint in our in our space for a while.
00:24:38.880 --> 00:24:50.319
If if if what WhatsApp is executing perfectly over over the over the time between now and you know 2030, what what what what do you predict is is something that needs to change or or will change?
00:24:50.480 --> 00:24:52.319
Is it technician productivity?
00:24:52.480 --> 00:25:00.960
Is it maybe the the operators, their their their operating margins is improving, drivers trust trust brands more.
00:25:01.519 --> 00:25:06.000
If if we're being successful as an industry, what do you think is going to change and what will that look like?
00:25:06.240 --> 00:25:12.880
Yeah, and I hope it happens a lot faster than 2030 because we need to get way more EBs out there.
00:25:12.960 --> 00:25:16.079
And in the end, I think it's it's all one big flywheel, right?
00:25:16.240 --> 00:25:18.160
It's all gonna feed into each other.
00:25:18.240 --> 00:25:21.039
And I do think it starts with technician productivity.
00:25:21.279 --> 00:25:26.319
So and especially when you look at uh the fact that the trades are gonna grow too, right?
00:25:26.559 --> 00:25:33.680
So how do you ensure that these new entries into an electrified world are successful across, like I said, right?
00:25:33.759 --> 00:25:36.880
90 different brands, 40 different, you know, pieces of software.
00:25:37.039 --> 00:25:42.319
I think it's about focusing on productivity of technicians, which the platform has been shown to do.
00:25:42.480 --> 00:25:50.400
We buy technicians more uh capacity on a yearly basis so they can visit more sites, attain more customers.
00:25:50.640 --> 00:25:55.440
That's the start in buying or in making technicians more productive.
00:25:55.680 --> 00:26:05.920
We that leads into the operator achieving more uptime and having less operational spend cutting into that um profitability margin, right?
00:26:06.160 --> 00:26:08.000
So then they're gonna scale more.
00:26:08.160 --> 00:26:11.279
More chargers, happier drivers in the end, right?
00:26:11.440 --> 00:26:19.680
So the driver is is sort of that last point in the flywheel and they don't care, they don't care what's happening in the background and the complexity of what it takes to get that.
00:26:19.839 --> 00:26:21.279
They just want chargers that work.
00:26:21.440 --> 00:26:23.119
So I think that's the ultimate goal.
00:26:23.279 --> 00:26:29.039
But first we start with productivity that if that leads to a very happy side operator as well.
00:26:29.359 --> 00:26:29.599
Yeah.
00:26:30.319 --> 00:26:31.920
You're absolutely right, Hillary.
00:26:32.000 --> 00:26:37.519
When we look over to the fuel and infrastructure on the on the liquid side, it that's already an experience.
00:26:37.680 --> 00:26:43.920
Like you say, the the drivers shouldn't care about what's going on, software stacks and hardware brands.
00:26:44.160 --> 00:26:52.480
I mean, when you pull up to a gas pump at any gas station, the average consumer does not care about the manufacturer of that gas pump.
00:26:52.559 --> 00:26:53.519
They don't even know who it is.
00:26:53.759 --> 00:26:56.640
They don't care about any of the upstream products.
00:26:56.799 --> 00:26:58.559
They don't care about any of that.
00:26:58.720 --> 00:27:03.839
It's just the pump works, they get fuel, they go on, and they're not worried about everything else.
00:27:03.920 --> 00:27:05.279
It's just it's kind of immaterial.
00:27:05.359 --> 00:27:08.640
And so getting getting to that point in our space is is going to be important.
00:27:09.039 --> 00:27:21.599
Well, there's been a good good conversation for those that are listening and want to connect with you directly, want to learn more about what's up, want to hear about how they can possibly integrate that solution and into their operations.
00:27:21.759 --> 00:27:24.160
What's where do where do they find find you?
00:27:24.319 --> 00:27:25.200
How do they connect with you?
00:27:25.519 --> 00:27:27.920
Yeah, you can reach out in two ways.
00:27:28.079 --> 00:27:36.079
You can reach out directly to me on LinkedIn, or you can email info at WhatsAppEV.io.
00:27:36.640 --> 00:27:37.119
Got it.
00:27:37.359 --> 00:27:39.519
Well, thank you for coming on Field Frequency.
00:27:39.599 --> 00:27:40.880
It's it's been a good conversation.
00:27:41.119 --> 00:27:41.680
Appreciate your time.
00:27:42.000 --> 00:27:42.880
Thank you for having me.
00:27:43.039 --> 00:27:43.359
Thank you.
00:27:44.000 --> 00:27:47.119
This episode was produced and edited by the team at Atozi.
00:27:47.519 --> 00:27:52.480
To find out more, visit autosy.co a-t-o-z-y.co.
00:00:00.160 --> 00:00:04.480
It's a bit reductive to think EV charging reliability is just a hardware problem.
00:00:04.719 --> 00:00:06.879
It's not, at least not primarily.
00:00:07.040 --> 00:00:09.119
The bigger issue is operational fragmentation.
00:00:09.279 --> 00:00:12.960
There's too many systems, too many stakeholders, too many gaps in accountability.
00:00:13.199 --> 00:00:19.839
In this episode, Hillary Taylor, CEO of What's Up, breaks down why uptime isn't just about fixing broken chargers faster.
00:00:19.920 --> 00:00:25.920
It's about better data, better workflows, and giving technicians the right context before a truck ever rolls.
00:00:26.079 --> 00:00:31.920
AI won't fix broken operations by itself, and if we don't solve that, scale really won't matter.
00:00:32.159 --> 00:00:33.359
Let's get into it.
00:01:01.759 --> 00:01:02.320
What's up?
00:01:02.479 --> 00:01:03.439
Clever play there.
00:01:03.759 --> 00:01:04.959
How are you doing, Hillary?
00:01:05.200 --> 00:01:05.680
I'm good.
00:01:05.840 --> 00:01:06.239
I'm good.
00:01:06.319 --> 00:01:06.719
How are you?
00:01:07.120 --> 00:01:07.599
Doing well.
00:01:07.760 --> 00:01:09.280
Thank you for sharing some time.
00:01:09.599 --> 00:01:17.120
We tried to get you on the episode or on the show, uh seems like a little bit, but stars wouldn't align, but finally it's come together.
00:01:17.200 --> 00:01:18.799
So I'm I'm glad it it did.
00:01:18.879 --> 00:01:26.400
And uh, you know, we appreciate the uh the the connection between Field Advantage and what's up, and we're glad to have you on the show.
00:01:26.480 --> 00:01:28.640
And and uh let's just get into it.
00:01:28.799 --> 00:01:41.200
So I'd uh I'd like to start off the conversation with you walking us through your personal origin story, share with the audience what pulled you into the EV charging space uh, I guess over a decade ago, but now it's been a long time.
00:01:41.680 --> 00:01:45.359
You're engaging me, you know, giving me away.
00:01:45.680 --> 00:01:49.680
But yeah, so right out of college, actually, I'm a Georgia Tech grad.
00:01:49.840 --> 00:01:58.319
I graduated with a degree in electrical engineering, and I went straight from being a college student to working in a late-stage startup called Whitricity.
00:01:58.719 --> 00:02:04.239
So that that was my entrance to vehicles and specifically electric vehicle infrastructure.
00:02:04.319 --> 00:02:14.560
And this one was very futuristic in that it was wireless charging of electric vehicles, which I still think is on the horizon in terms of technology deployment for us.
00:02:14.719 --> 00:02:21.039
But from there, I went to Spark Charge, uh, which is mobile charging of electric vehicles.
00:02:21.199 --> 00:02:25.599
And that's where I got the first opportunity to lead an engineering team.
00:02:25.840 --> 00:02:36.159
So I've always been on the technical side of the EV charging game, where, and that's where I built mobile grid list uh DC fast chargers.
00:02:36.319 --> 00:02:41.039
And then from there, I went to megawatt level charging in new.
00:02:41.360 --> 00:03:10.879
But where where it really clicked for me that uh there was a reliability issue that as a hardware provider, you know, I was contributing to an ish, uh a big issue that was global, was a a drive from Boston to Buffalo, which should have taken seven to eight hours, but it ended up taking 17 hours and I was running a fever and I was in a Chevy bolt, which at the time you couldn't charge the Chevy bolt due to uh uh fear of the something happening with the battery.
00:03:11.120 --> 00:03:14.719
Um, so you can only charge it to 80% and it was 23 degrees outside.
00:03:14.960 --> 00:03:24.240
So it was just a nail biter the entire way, and my drive effectively doubled because the chargers we were encountering along the way were not working.
00:03:24.400 --> 00:03:41.120
And that was the click for me that there is something going on here that's really serious with the infrastructure and how I mean it it really put things into perspective for me, then as the driver, how can we ask others to experience something like I just went through?
00:03:41.280 --> 00:03:45.759
And that's where, you know, the seed was planted and what's up was eventually born.
00:03:46.000 --> 00:03:46.159
Yeah.
00:03:46.560 --> 00:03:52.159
So what year would that trip have been if you were in the in that model of vehicle that could only go to 80%?
00:03:52.560 --> 00:03:53.199
What year was that?
00:03:53.599 --> 00:03:57.439
That was like late 2022, probably, or early 2023.
00:03:57.840 --> 00:03:59.919
Um and it was still cold in Boston.
00:04:00.080 --> 00:04:03.120
And then in 2024 was when What's Up was founded.
00:04:03.360 --> 00:04:04.240
Yeah, cool.
00:04:04.800 --> 00:04:12.800
Well, let's uh let's talk about, you know, you've kind of already alluded to it, but you know, what you saw in the field that led you to to building what's up.
00:04:12.879 --> 00:04:18.240
But what, you know, what where does What's Up sit in the ecosystem and what problem is it solving?
00:04:18.480 --> 00:04:18.639
Sure.
00:04:19.040 --> 00:04:19.839
Share that with the audience.
00:04:20.160 --> 00:04:20.399
Sure.
00:04:20.560 --> 00:04:21.040
Yeah.
00:04:21.360 --> 00:04:31.519
So I think fundamentally that the position we take with what's up is that there's a fundamental gap and layer missing in the ecosystem.
00:04:31.839 --> 00:04:35.120
So we're not integrating into an existing layer.
00:04:35.279 --> 00:04:51.519
We are a new layer that sits up top of the ecosystem and solves the reliability issue plaguing literally everybody on all layers, the software, the existing software layer, the existing hardware layer, the technician layer.
00:04:51.759 --> 00:04:54.240
We are the unifier to all of that.
00:04:54.480 --> 00:05:01.279
And with that, with that unification and standardization powered by AI, we will solve it for everybody.
00:05:01.600 --> 00:05:01.839
Yeah.
00:05:02.000 --> 00:05:06.319
So, you know, I understand what WhatsApp does for the audience.
00:05:06.480 --> 00:05:17.839
I I would just, and I don't want to oversimplify this, so you correct me if I'm if I'm wrong, Hillary, but but your your platform is not a replacement to the EMSP platforms that maybe CPOs are operating right now.
00:05:17.920 --> 00:05:20.160
It's it's a it's a supplement to that.
00:05:20.240 --> 00:05:28.959
It is it is uh and and and it's bolstering some of the gaps uh outside of just a standard CMS out-of-the-box CMS platform.
00:05:29.279 --> 00:05:29.519
Right.
00:05:29.920 --> 00:05:41.040
So not a replacement to a CMS, you're bolstering what CMS is are are trying to do in the space now as as maybe a layer below and a layer above, and maybe even a layer in between.
00:05:41.199 --> 00:05:41.519
I don't know.
00:05:41.600 --> 00:05:48.959
I don't I don't want to oversimplify it, but I want to kind of connect the dots for the audience that I don't can understand exactly and and we'll we'll get into it.
00:05:49.439 --> 00:05:56.800
Obviously, we'll we'll we'll break it all down, but uh uh is that a is that a good good breakdown of of the I I I think so.
00:05:57.040 --> 00:06:00.560
I think we are a connector in a fragmented ecosystem.
00:06:00.959 --> 00:06:01.360
There you go.
00:06:01.839 --> 00:06:03.839
We do we do three things, right?
00:06:04.079 --> 00:06:08.959
We bring proactivity to a very reactive ecosystem.
00:06:09.279 --> 00:06:24.319
We take action, and then that action is a continuous improvement mechanism where every future action is always smarter, better, faster, and will lead you to just increase success.
00:06:24.720 --> 00:06:25.279
Yeah.
00:06:25.839 --> 00:06:26.720
Good, good.
00:06:26.959 --> 00:06:41.600
You know, in your in kind of your catalyst, the experience on the drive that you mentioned, uh you know, it's it it signals something that was happening then, something that was happening before then, and something that really is happening today, and that is public charger reliability.
00:06:42.240 --> 00:06:48.480
And the data around reliability is is really uncomfortable when you look at it from an industry standpoint.
00:06:48.720 --> 00:06:56.399
You know, 78% up time, 22% non-functioning, flatline, just all of these things, these data points.
00:06:56.560 --> 00:07:02.160
And so from your vantage point, what is the what is the root cause of some of this reliability problem?
00:07:02.240 --> 00:07:08.879
I mean, you know, there's more than one, but what you know, kind of high side, what is where where is that reliability problem happening?
00:07:08.959 --> 00:07:10.879
Is it is this a hardware issue?
00:07:11.199 --> 00:07:13.279
Is this a workflow fragmentation?
00:07:13.360 --> 00:07:15.920
It is all of the above plus what's that look like?
00:07:16.160 --> 00:07:16.879
Yeah, yeah.
00:07:16.959 --> 00:07:20.879
And I think it's important to understand it's it's a compounding problem, right?
00:07:21.040 --> 00:07:26.959
So, you know, there are hardware failures, there are going to be software failures when you combine them.
00:07:27.120 --> 00:07:32.560
There are in environments and likelihoods where they will bring out the worst in each other, right?
00:07:32.720 --> 00:07:41.040
But the deep, I think the deep rooted, most painful part of this is the lack of standardized operational strategy and workflow.
00:07:41.199 --> 00:07:50.639
So we need to accept that this is a fragmented, I think there's 90 different hardware brands and 40 different software, you know, brands to pair with that, which is wild, right?
00:07:50.800 --> 00:07:54.319
We need to accept that this is a fragmented ecosystem.
00:07:54.560 --> 00:08:13.040
Now, how can we standardize the operational action workflow that sits on top of that and ensure that with every anomaly and every issue and every service call, there is a paired and strategic operational outcome with that need?
00:08:13.279 --> 00:08:13.519
Yeah.
00:08:14.079 --> 00:08:14.399
Yeah.
00:08:14.879 --> 00:08:18.560
Gosh, you know, the those stats are scary when you when you think about it.
00:08:18.639 --> 00:08:28.160
The the the the quantities of hardware, the quantities of software, multiple payment platforms and payment processors, and so many fail points in all of that.
00:08:28.319 --> 00:08:34.960
And and yet in in that in just that space, just that large space, there there is truly gaps.
00:08:35.120 --> 00:08:37.039
There's a there's an accountability to gap.
00:08:37.120 --> 00:08:43.120
And so when when we as a field service company, field advantage out in the field, we we get we we experience that firsthand.
00:08:43.200 --> 00:08:47.120
But the operator, the customers, the owner operators of these of these sites are dealing with it.
00:08:47.200 --> 00:08:56.080
And so between the the charger OEM and and the CMS and you know the site host that is just there, kind of captive to the process, whether they own the hardware or they don't.
00:08:56.240 --> 00:08:59.679
They're they're a part of the process and the part of the value chain and the field technicians.
00:09:00.000 --> 00:09:04.960
Accountability and and and ownership of the problem gets gets lost.
00:09:05.120 --> 00:09:06.639
It it gets lost in the noise.
00:09:06.720 --> 00:09:08.159
It's and it's not always clear.
00:09:08.240 --> 00:09:15.759
And so, you know, you're building a platform and you have a solution that is is kind of closing those accountability gaps.
00:09:15.919 --> 00:09:22.639
Where from your perspective, where does responsibility break down in in the model that we have today?
00:09:22.960 --> 00:09:23.679
Yeah, yeah.
00:09:23.919 --> 00:09:25.039
Really good question.
00:09:25.279 --> 00:09:28.879
And just going back to yeah, there's so many different stakeholders here.
00:09:29.120 --> 00:09:32.799
And I think having grasping that that is also like a fragmented.
00:09:33.039 --> 00:09:44.799
So we're we're just stacking fragmentation, honestly, on top of each other and acknowledging and realizing that with any fragmentation is a natural gap forms, whether you like it or not.
00:09:44.960 --> 00:09:50.399
And I I think the breakdown is that the problem for people lives in different tools.
00:09:50.639 --> 00:09:54.960
So the the problem for a site operator may live in the CSMS.
00:09:55.039 --> 00:10:01.919
For a technician, it'll live in Salesforce or whatever asset management or field service tracking system, right?
00:10:02.000 --> 00:10:04.799
For a warranty provider, it's living somewhere else, right?
00:10:05.120 --> 00:10:18.080
I think creating this unified and standardized source of truth, no matter the hardware brand, software brand, site operator, technician, a unified source of truth where everyone is communicating in.
00:10:18.320 --> 00:10:29.360
And despite communicating in differently, the way that it's communicated out via our platform, for example, is standardized to everyone, you start to bridge and close those gaps.
00:10:29.600 --> 00:10:29.840
Yeah.
00:10:30.159 --> 00:10:31.600
That's that's good.
00:10:31.679 --> 00:10:42.639
And that the closing of those gaps is is key because there's so many different, not only the fragmentation of the, of the, of the product, but then also how we as a field service company are responding to the needs.
00:10:42.879 --> 00:10:51.759
We get, like you said, the example of whether it's Salesforce as the as the ticket managing platform or Zendez Zoho, what whatever it may be, there's multiple platforms.
00:10:51.919 --> 00:11:05.200
And so we're we're we're coming into the service need and we're reacting to it, but there is sometimes as as a as a field service provider, we we are captive, a part of, you know, part of that bottleneck.
00:11:05.279 --> 00:11:13.679
And I know that in our conversations in the past, as well as just what I've seen from presentations, you've cited kind of vague work order development.
00:11:13.840 --> 00:11:20.399
So basically curating the work on the front end, developing that scope of work and what the SOPs, what the field service procedures.
00:11:20.639 --> 00:11:29.759
And so I've I've heard you cite that vague work order development is and and too much on-site troubleshooting is a part of that butt bottleneck in in service work.
00:11:29.919 --> 00:11:35.600
And so what information needs to be established before the truck even leaves, before the truck even rolls?
00:11:35.759 --> 00:11:38.559
What are kind of some of the fundamentals that needs to be established?
00:11:38.799 --> 00:11:42.559
Yeah, I I think that uh I think fundamentally it's like a mindset shift, right?
00:11:42.799 --> 00:11:49.840
So the technician, and this is how I kind of went into this and how we designed WhatsApp's platform, which is a two-sided platform, right?
00:11:49.919 --> 00:11:56.720
So we cater to the site operator, but also the technician with the idea that the technician is the sleeper, you know, agent.
00:11:56.879 --> 00:11:58.960
They're the most important part of this problem.
00:11:59.120 --> 00:12:05.200
So why are we giving technicians instruction instead of context and instruction?
00:12:05.440 --> 00:12:07.039
So yeah.
00:12:07.200 --> 00:12:17.519
So why aren't we contextualizing the things that we're telling people to do out in the field so that there is a not a better understanding of why they're doing what they're doing?
00:12:17.840 --> 00:12:25.440
Because I I think fundamentally technicians are capable of making really smart decisions because they're the ones, they're the only ones looking at the problem.
00:12:25.840 --> 00:12:34.080
So, with that, I think uh the the platform, right, has been designed to provide all of the history with with respect to this charger.
00:12:34.240 --> 00:12:39.919
Any sort of previous work orders on that charger, the technician should have access to.
00:12:40.000 --> 00:12:47.120
So an entire it's basically like a library or archive of all the work that's been done.
00:12:47.200 --> 00:12:51.919
And if there's repeat errors, the technician should know about that before going on site.
00:12:52.399 --> 00:13:00.240
All the contextual data that was flagged at the time of the work order creation, like why did this get created?
00:13:00.399 --> 00:13:02.720
I think the technician should have insight into all of that.
00:13:02.960 --> 00:13:27.440
Of course, we don't want to overload them with inundation of information, but we want to do it in such a way that they can, you know, very quickly obtain a contextual view of this piece of hardware and not just see it in a moment in time and then allow them to also have influence in in what is is done when they get on site instead of just someone telling them what to do.
00:13:27.840 --> 00:13:28.159
Right.
00:13:28.399 --> 00:13:29.200
Absolutely right.
00:13:29.360 --> 00:13:35.679
Because again, going back to the comments around all the stakeholders and all those involved in in processes.
00:13:35.840 --> 00:13:48.720
So you think about like an OEM who has updated a field service procedure for how maybe a particular component is replaced or how a particular component is troubleshooted in the field, and maybe the field service procedure has changed.
00:13:49.039 --> 00:14:04.960
And so that that those predictive insights that you're you're talking about and is is key to really cost containment and influencing overall the the success of the site and and the opex for the for the site host or the owner operator.
00:14:05.600 --> 00:14:08.879
Predictive maintenance has has really been aspirational.
00:14:08.960 --> 00:14:14.000
And so I like how you're you're you're you're putting guidelines around the information set.
00:14:14.159 --> 00:14:23.679
You're saying context, but then also uh historical data, but not just a data dump that is is so much you don't even know where to start because it's so much information.
00:14:23.840 --> 00:14:32.480
And so when when you're looking at predictive work versus reactive work and kind of the economics in there, how do how do you operationalize that?
00:14:32.559 --> 00:14:36.080
Or and how does WhatsApp help its customers operationalize that?
00:14:36.159 --> 00:14:43.279
So so predictive insights are being used and and maybe false starts aren't aren't a part of uh the process anymore.
00:14:43.519 --> 00:14:43.840
Yeah.
00:14:44.080 --> 00:14:47.360
So I I think fundamentally, and this is something we learned over time.
00:14:47.519 --> 00:15:05.120
So WhatsApp started just as a pure predictive maintenance platform, but we realize that predictive analytics and data in this ecosystem don't solve everything and and aren't super useful unless there is a very prescriptive operational action to go along.
00:15:05.279 --> 00:15:08.879
So I think that's the emphasis in and that's what we've learned over time.
00:15:08.960 --> 00:15:21.360
And that's really how the technicians started to come center, front and center in in the way that we approach solving this really big issue is there needs to be a very strategic operational workflow paired.
00:15:21.519 --> 00:15:35.360
And then using the technician feedback loop to ensure that the economics of what we just told them to do makes sense any future time that something similar arises in terms of anomaly or problem set.
00:15:35.759 --> 00:15:36.000
Yeah.
00:15:36.159 --> 00:16:01.840
Yeah, because they're you know, back to my my comments around OEMs maybe changing or updating rather a field service procedure for how something is if that if that's not ingested um in into a platform, I mean that that that knowledge is living with the OEM, but maybe it hasn't transferred out across all platforms, or maybe maybe they've shared that with the the owner of the of the hardware, or maybe it's available in some place, but how does the technician access that?
00:16:02.000 --> 00:16:13.759
And so it's it's getting into that predictive maintenance and and and and kind of building building that that that fence around this information, or at least a gateway for that information to come inside the fence.
00:16:13.840 --> 00:16:14.240
I don't know.
00:16:14.320 --> 00:16:23.360
But I know I know your platform leans into machine learning, uh machine learning driven insights, because to your earlier comments about learning and relearning.
00:16:23.440 --> 00:16:30.559
And so it's it's it's it's compounded problems, so you're compounding the knowledge base uh to exp you know to to prov provide that solution.
00:16:30.720 --> 00:16:35.440
What what signals are you extracting from the CMSs?
00:16:35.840 --> 00:16:39.279
Or maybe what are you seeing in the CMSs that are missing that are creating that gap?
00:16:39.440 --> 00:16:46.159
How is how are you leveraging AI to extract and and build upon what maybe CMSs are missing?
00:16:46.480 --> 00:16:49.919
Yeah, and and I think the word context will come up a lot here.
00:16:50.080 --> 00:17:01.519
It's context to the technician, it's context around the piece of hardware, because I think for the most part, right, coming as a ex-hardware designer, we design a lot of these things in the lab.
00:17:01.840 --> 00:17:10.079
And the the user, the variation in user interaction and environment in which this thing could live, although you test for it, it's kind of infinite.
00:17:10.240 --> 00:17:25.519
You could put this thing in an environment where it will be used and abused by, you know, a a driver because they've had a bad day, someone who is, you know, pissed off every day after work and just like drops this thing on the floor every time they try to charge, right?
00:17:25.599 --> 00:17:26.400
You don't know.
00:17:27.359 --> 00:17:31.839
So it's taking that sort of environmental context into play.
00:17:32.000 --> 00:17:36.400
So everything we layer on to just OCC, OCPP data.
00:17:36.720 --> 00:17:39.279
We do integrate with the hardware as well when we can.
00:17:39.519 --> 00:17:50.880
It's not a game changer if we don't get that though, but we layer on weather data, just general weather data, weather event data, utility data, extreme weather events, and lightning strike data.
00:17:50.960 --> 00:17:55.279
So we have insight into literally everything that's going on around a charger.
00:17:55.440 --> 00:18:02.000
And that really enables us to distinguish whether it is the charger itself that is misbehaving.
00:18:02.240 --> 00:18:04.240
Is it what's going on around it?
00:18:04.400 --> 00:18:07.200
Did the Transformer just experience a lightning strike?
00:18:07.279 --> 00:18:10.079
Or is the grid experiencing a lot of lightning strikes, for example?
00:18:10.960 --> 00:18:16.240
We've been able to avoid a lot of truck rolls just because we're seeing like brownout behavior.
00:18:16.319 --> 00:18:17.920
That's not due to the charger itself.
00:18:18.000 --> 00:18:19.839
It's due to what's going on in the grid.
00:18:21.279 --> 00:18:32.960
So yeah, I think where the machine learning with us comes into play is contextualizing and looking at the charger holistically in the environment in which it has been installed.
00:18:33.119 --> 00:18:49.519
And then it is in finding nuanced patterns within that data set that over time actually are incredibly significant to the expected lifespan or predicted failure date of that asset.
00:18:49.759 --> 00:18:49.920
Yeah.
00:18:50.240 --> 00:19:06.319
Going back to your your comments about the the technicians being a part of the communication and information exchange loop, your platform is leveraging that that field feedback as a part of the infrastructure for your solution, for your product.
00:19:06.559 --> 00:19:21.279
And so you're you're positioning technicians' feedback as that validation layer that is enhancing what the AI is learning and what what the machine learning is doing, you're taking and leveraging that feedback from the field and fortifying it.
00:19:21.519 --> 00:19:34.960
How do you how do you structurally how do you structurally prevent bias or noise or poor inputs from maybe from the field that aren't mess necessarily a line and and those could degrade the model accuracy?
00:19:35.359 --> 00:19:36.960
So how how do you sort through the noise?
00:19:37.119 --> 00:19:38.400
How do you how do you do that?
00:19:38.720 --> 00:19:49.359
Yeah, and it it has everything to do with the line of questioning that we have strategically put in place for technicians in order to in order to ensure that there is very little noise.
00:19:49.440 --> 00:19:52.880
If they're very targeted questions, they're all standardized.
00:19:53.200 --> 00:19:56.799
And via that data set, we can pull keywords.
00:19:56.960 --> 00:19:58.640
So we do a lot of work on the back end.
00:19:58.799 --> 00:20:02.480
So and it it's actually at this point, it's voice to text for them now.
00:20:02.640 --> 00:20:05.200
And but still with the strategic line of questioning.
00:20:05.279 --> 00:20:15.279
And then we do a lot of work on the back end to pull key phrases and key things they said with and we remove all of the what could be interpreted as bias, right?
00:20:15.359 --> 00:20:17.039
So we're looking for very specific things.
00:20:17.119 --> 00:20:23.839
We do a lot of work in the model on the back end to pull out those specific things and then feed it into the training data.
00:20:24.079 --> 00:20:36.640
So I I think it's it's about treating the feedback as data infrastructure and being very firm on some of the things that you're looking for in order to label data sets and then retrain.
00:20:37.200 --> 00:20:41.680
What What's Ups platform is is obviously hardware open.
00:20:41.839 --> 00:20:49.440
It's operating across different ecosystems with different hardware brands and other other software stacks and other and multitude of CMS platforms.
00:20:49.599 --> 00:20:53.680
But um, you know, this kind of goes back to the fragmentation part of our conversation.
00:20:54.240 --> 00:21:09.440
The market is uh is is shown that chargers, charger platforms are unique and they they behave differently, not only as a as a brand and as a product, but then when you introduce different environments, cold, hot weather income locations, et cetera.
00:21:09.599 --> 00:21:23.279
So what uh you know what what has been the hardest layer that you've seen that that is needs to be overcome to standardize across for your for your clients, for your customers who may be operating multiple hardware and multiple software stacks.
00:21:23.519 --> 00:21:30.240
How how do you what's the hardest layer to standardize without losing the nuance about each system operates and consequently also fails?
00:21:30.559 --> 00:21:30.799
Yeah.
00:21:31.039 --> 00:21:38.880
I think the most, you know, and and I should know this already just come coming from with this with OCPP so much.
00:21:39.200 --> 00:21:56.799
Like I think the surprising one that I would have expected to move a little bit faster in development and have provided a little bit more guidance to how to get rid of I I think when everyone implements OCPP, it's like they implemented the to the standard, but it's to their own interpretation.
00:21:56.880 --> 00:21:58.960
So you get different different flavors.
00:21:59.039 --> 00:21:59.759
It's like cocoa.
00:22:00.079 --> 00:22:02.720
Coca-Cola, Coca-Cola cherry, Coca-Cola vanilla.
00:22:03.039 --> 00:22:03.279
Right.
00:22:03.440 --> 00:22:14.720
So I I don't know if one is harder than the other, but the one that I would have expected to I I think root out a little bit of the flexibility within the standard is OCPP.
00:22:14.880 --> 00:22:17.359
Like I think that that would help a lot of people.
00:22:17.599 --> 00:22:26.079
I think that it it still is it's lacking in in some areas, but I think those are solvable with context and adding data layers over.
00:22:26.319 --> 00:22:31.440
But the one that I'm most surprised with is OCPP and how that's given people pain.
00:22:32.000 --> 00:22:32.400
Yeah.
00:22:32.720 --> 00:22:39.039
And yet, and yet that is is kind of table stakes around the the system as a whole.
00:22:39.279 --> 00:22:46.319
You know, obviously for the for the audience listening, you know, OCPP is a converse is a is an acronym you hear often on the on this show.
00:22:46.559 --> 00:22:52.640
Open ChargePoint Protocol, it's really the fundamental language uh of communication be in in the infrastructure.
00:22:52.960 --> 00:23:07.680
But um, you know, kind of kind of zooming out here and and just you know maybe looking at the market as a whole and our industry, service level agreements are our competitive edge in this, um at least from the from the owner operator standpoint.
00:23:07.839 --> 00:23:16.480
They're looking because at this point, many have deployed chargers, they've they have multiple hardware brands that they've worked worked with and and have operating.
00:23:16.640 --> 00:23:24.160
And so kind of the next frontier for them is operation operationally security around operations.
00:23:24.400 --> 00:23:30.799
And so uptime is obviously the the next brand differentiator in in public charging.
00:23:31.200 --> 00:23:31.519
Yes.
00:23:31.920 --> 00:23:44.160
But how do you how do you see your solution or and predictive man maintenance in in general evolving in into a lever around SLAs rather than just controlling costs?
00:23:44.319 --> 00:23:48.240
Like in other words, it's kind of going back to to the to the learning model that you were talking about.
00:23:48.640 --> 00:24:09.920
Yeah, I I think that we could be we could be the lever or the the knob that allows people to start guaranteeing uptime and feel feel more confident about making that guarantee because uh we have machine learning models that have proven themselves to be continuously improving over time.
00:24:10.160 --> 00:24:25.759
I think that we we play very well into a risk mitigation model where we start enabling flat rate fees for service and SLAs while still be able still enabling, excuse me, high uptime guarantee.
00:24:26.079 --> 00:24:38.799
Well, if we were if we were to kind of zoom out and and with a kind of a five-year perspective, or maybe even you know 2030, that seems to be uh a waypoint everyone's marching towards, or at least been a waypoint in our in our space for a while.
00:24:38.880 --> 00:24:50.319
If if if what WhatsApp is executing perfectly over over the over the time between now and you know 2030, what what what what do you predict is is something that needs to change or or will change?
00:24:50.480 --> 00:24:52.319
Is it technician productivity?
00:24:52.480 --> 00:25:00.960
Is it maybe the the operators, their their their operating margins is improving, drivers trust trust brands more.
00:25:01.519 --> 00:25:06.000
If if we're being successful as an industry, what do you think is going to change and what will that look like?
00:25:06.240 --> 00:25:12.880
Yeah, and I hope it happens a lot faster than 2030 because we need to get way more EBs out there.
00:25:12.960 --> 00:25:16.079
And in the end, I think it's it's all one big flywheel, right?
00:25:16.240 --> 00:25:18.160
It's all gonna feed into each other.
00:25:18.240 --> 00:25:21.039
And I do think it starts with technician productivity.
00:25:21.279 --> 00:25:26.319
So and especially when you look at uh the fact that the trades are gonna grow too, right?
00:25:26.559 --> 00:25:33.680
So how do you ensure that these new entries into an electrified world are successful across, like I said, right?
00:25:33.759 --> 00:25:36.880
90 different brands, 40 different, you know, pieces of software.
00:25:37.039 --> 00:25:42.319
I think it's about focusing on productivity of technicians, which the platform has been shown to do.
00:25:42.480 --> 00:25:50.400
We buy technicians more uh capacity on a yearly basis so they can visit more sites, attain more customers.
00:25:50.640 --> 00:25:55.440
That's the start in buying or in making technicians more productive.
00:25:55.680 --> 00:26:05.920
We that leads into the operator achieving more uptime and having less operational spend cutting into that um profitability margin, right?
00:26:06.160 --> 00:26:08.000
So then they're gonna scale more.
00:26:08.160 --> 00:26:11.279
More chargers, happier drivers in the end, right?
00:26:11.440 --> 00:26:19.680
So the driver is is sort of that last point in the flywheel and they don't care, they don't care what's happening in the background and the complexity of what it takes to get that.
00:26:19.839 --> 00:26:21.279
They just want chargers that work.
00:26:21.440 --> 00:26:23.119
So I think that's the ultimate goal.
00:26:23.279 --> 00:26:29.039
But first we start with productivity that if that leads to a very happy side operator as well.
00:26:29.359 --> 00:26:29.599
Yeah.
00:26:30.319 --> 00:26:31.920
You're absolutely right, Hillary.
00:26:32.000 --> 00:26:37.519
When we look over to the fuel and infrastructure on the on the liquid side, it that's already an experience.
00:26:37.680 --> 00:26:43.920
Like you say, the the drivers shouldn't care about what's going on, software stacks and hardware brands.
00:26:44.160 --> 00:26:52.480
I mean, when you pull up to a gas pump at any gas station, the average consumer does not care about the manufacturer of that gas pump.
00:26:52.559 --> 00:26:53.519
They don't even know who it is.
00:26:53.759 --> 00:26:56.640
They don't care about any of the upstream products.
00:26:56.799 --> 00:26:58.559
They don't care about any of that.
00:26:58.720 --> 00:27:03.839
It's just the pump works, they get fuel, they go on, and they're not worried about everything else.
00:27:03.920 --> 00:27:05.279
It's just it's kind of immaterial.
00:27:05.359 --> 00:27:08.640
And so getting getting to that point in our space is is going to be important.
00:27:09.039 --> 00:27:21.599
Well, there's been a good good conversation for those that are listening and want to connect with you directly, want to learn more about what's up, want to hear about how they can possibly integrate that solution and into their operations.
00:27:21.759 --> 00:27:24.160
What's where do where do they find find you?
00:27:24.319 --> 00:27:25.200
How do they connect with you?
00:27:25.519 --> 00:27:27.920
Yeah, you can reach out in two ways.
00:27:28.079 --> 00:27:36.079
You can reach out directly to me on LinkedIn, or you can email info at WhatsAppEV.io.
00:27:36.640 --> 00:27:37.119
Got it.
00:27:37.359 --> 00:27:39.519
Well, thank you for coming on Field Frequency.
00:27:39.599 --> 00:27:40.880
It's it's been a good conversation.
00:27:41.119 --> 00:27:41.680
Appreciate your time.
00:27:42.000 --> 00:27:42.880
Thank you for having me.
00:27:43.039 --> 00:27:43.359
Thank you.
00:27:44.000 --> 00:27:47.119
This episode was produced and edited by the team at Atozi.
00:27:47.519 --> 00:27:52.480
To find out more, visit autosy.co a-t-o-z-y.co.
