HPE Agentic Smart City Solution – Focusing on Real-World Outcomes

This is going to be the real world impact, which I think is what everybody wants to get to the, um, so here are the, here are the, um, kind of four high level use cases. One that we'll not show today 'cause we don't have a video of it, is digital concierge is one that they're very excited about. Uh, when you start to think about having it, you know, if you, if you've ever been in the mountains of Colorado, um, and a snowstorm happens, there is one highway in and one highway out, and that when the snow comes, you may or may not be going down that highway because it's gonna be blocked by wrecks and or snow, um, or impassable.

So how do, how do people get information at that time? If they're at the visitor center and it's two o'clock in the morning, how do they interface and ask questions because they've just kind of been displaced by a snowstorm, those type of things. Or in the library, many will come by and ask questions about the town of Vail, what's going on events, uh, Or I'm a vegan, uh, where should I eat tonight?

Where should I park to go to that particular place? And can you set me up a reservation? And they actually showed those use cases and all, uh, at GTC, it was at the SHI booth with the live digital human, uh, digital concierge, I'm sorry.

Uh, and it was quite, quite stunning graphical interface. It was just built for Vail. That's why we don't have a demo.

Um, and, uh, and so we're, um, we're still in the proof, but that again, can use the com use as comas as the backend kind of intelligence and brain, um, while a much prettier front end in, um, in the digital concierge. And, uh, and so that's something we're working with them to, to roll out their vision for. It is very, um, is very progressive and innovative going exactly to Luke's point, not just in, tell me about the events happening at Vail this week.

Um, but how can you really truly improve the visitor experience and even the citizen experience of interacting with Vail? And, uh, but I think one of probably the most heated debates is do you actually tell people the best place to ski today? Um, from an insider perspective, that's probably still the, the one last question.

We weren't willing that there's a good, a good amount of, uh, question on how to include that or not. Uh, we'll be going into code indeed assistance. Uh, there are, um, affordable housing is a challenge in the Vail Valley and how, and they have a number of land parcels and, um, and housing units that have deed, deed restrictions.

There are restrictions in how you can build on it. There are restrictions in how you can sell it. There are restrictions in what you can charge for rent.

There's just all sorts of different things that, um, can come up. And those deeds can be back from 1960, they can be back from last year. Um, there's all sorts of different, uh, parts and parcels of, uh, of what that looks like.

How do you go and consistently and quickly, um, and quickly identify that? We'll be able to show that what we were just touching on around fire detection and prevention. Uh, we're just rolling that out now with the fire, uh, with a fire team there.

And, uh, they got really excited for what we've been able to show them so far about that real time response. Not only being able to improve the video, to be able to see it not only, um, having good analytics around it, but now bringing context in around weather and, um, and what conditions have been. Because there's a difference if you see smoke and it's been snowing for five days or you see smoke and, uh, and it's been a, uh, red flag day for a week and it's incredibly dry and it's in the middle of, uh, and it's in the middle of timber.

So yes. Um, with the, with the digital concierge and the fire detection system, have y'all, do you have any data or information you can share on anything with 9 1 1 service integration for those situations? So what we're looking, we're, we're taking these in phases.

Mm-hmm. So as, uh, as we just started in August for a workshop and we're, we're here now. Um, so those are the type of conversations of, as we look at further phases, what will that potentially look like?

But I think 8 1, 1 Or 4, 1 1 4, 1 one's being integrated as the next step. And then 9 1 1 is the Sure, Sure. That's pretty critical.

But if, if someone has an assistant right now, it's how are they gonna get redirected over? But yeah. Right.

So I think, you know, those are things that we can look at a redirect, but, you know, this is something that we're, we're working out. So great question. And I think, We'll, the goal is at the information center you can, 'cause there is a cop that comes by every hour, you can flag them just from the assistant and he'll come by quicker.

But like I said, we're gonna incorporate 4 1 1, so information assistance, et cetera, where you would then also get non-emergency calls and support and then full 9 1 1 later. But on the fire stuff, it's totally different. You'll actually see that coming out.

Um, and I think, you know, one of the really interesting things was, you know, by reusing their infrastructure, it makes it, again, much faster to that time to value much faster to that proof of, you know, Hey, how can you use cameras in different ways? And uh, and Vail actually has the largest free bus system in, um, in the US and, uh, which I did not know, but I do now. And, uh, which is very cool.

Eagle County's a very large county and the population is mainly there to support, you know, the winter and the, the, uh, summer sports. So they have to bus 'em in from, I mean, upwards of 40 miles and just the network that's needed to create that and the whole town system, it's, it's quite large. Um, but what's interesting is all of those buses have cameras.

And so now the conversation is, well, what can you see based on those cameras? Because can you now actually detect potholes so that you don't have to wait for somebody to call 4 1 1 instead, you could actually automatically detect that and then be able to schedule that for maintenance. So we're, it's really opening those conversations up around, oh, it's not just a bus for transportation and cameras for safety, but weight.

It can also detect all of these other potential items. And so it, it's really a great, um, kind of discovery together on the different ways to reuse what they already have. Um, and then also kind of, uh, last but not least to touch on that 5 0 8 compliance that I described earlier, uh, that has been separately released but is also a part of this platform.

And, uh, and we're really excited for that. And the ability that it has to help municipalities education, even a lot of enterprise customers to be able to much more quickly, um, make their information accessible. And, and it's one of those things that it is a regulation.

It is compliant. It is something that can open you up to lawsuits or potential impact for contractual awards. Uh, but when I look at it, I'm also like, isn't it also the right thing to do that we help our citizenry be able to get at information that they need from an, from a governmental service perspective.

Luke and Robin, uh, guy Courier Future Group, um, I love these discussions of these case studies. They're focused. I think we all have a lesson to take from this of quite the transformative nature of, of ai, um, focusing, we talk about it all the time, but just actually seeing examples that are really quite focused, um, I think is is is great.

This is public sector though. And a lot of this discussion, especially around those high rates of failure and stuff, which everybody here at least now knows I I take with a grain of salt. 'cause I think piloting and experimentation creates, I think it's a good thing.

Yeah. Yeah. I Actually, I actually think the failure rate is one of those that really frustrates me just to get on like my little soapbox for a Moment.

Keep, keep going. Um, And every time I was literally what, what what I write about is, Yeah. So every time I present next to Luke, I always feel like I need to be on a soapbox anyways.

The um, uh, but the, uh, but, but it makes me a little crazy because it scares people from trying and I'm like, the whole point of it is to fail fast or to experiment quickly. And maybe you experimented with five different things, but then you found the one that actually had value or you were able to dial it in. That's not a failure.

Yeah. Failure in a pilot is not a failure. Right.

Failure in A POC could be a failure because the vendor was trying to prove a concept that did not get proved. My question though is outside of the public sector, um, there's I think a lot of discussion, um, around ROI sometimes it's discussion about how nobody in management seems to care about ROI, the board just wants you to say I'm doing AI stuff. Um, how is first of all the city looking at ROI?

Because there may be considerable upfront investment here and they're looking at long tail. That's the, that's the counterpoint to these being really focused and specific things. You even said like code indeed is what we're looking at next.

No. So no, no. It's now co Code code indeed actually drove the initial ROI on this.

They had three full-time FTE to be hired to just keep up with the code indeed effort that they have. It's also what I call a a, a negative linear effort. The more that homes they bring on from a code deed restriction standpoint, there's a negative correlation to the amount of people you need that goes exponentially in the wrong direction.

It's just more and more files to go and read through. And they had this goal of literally quadrupling it. So it wasn't gonna be the two to three people this year.

It was gonna be two to six people next year and six to 10 the following year. So they had that already mapped out and we literally came to them, uh, when we saw that and said we could literally make it so that you don't have to hire those net new ones. You can repurpose that budget partly to us and then partly to the future.

5 0 8 is another one that's a multimillion dollar three year engagement than most organizations at the small level are having to go through. And, uh, as um, Robin said, we've built that with HPE to be a wedge solution comes in at roughly 250,000 to 300,000 with the server and the agents. And now you have a full-time 5 0 8 compliant agent running on the website, remediating it.

So you go from a couple hundred thousand of CapEx to a multimillion dollar multi-year engagement. So the ROIs have to be driven at the public sector like entirely, even more so than I think the private sector that can actually say, I'm gonna make a longer term strategy on this and hopefully the next one gets ROI or the next one public sector like darn near has to be. 'cause uh, this all leaked actually about a month ago when they went to their board, literally.

And there was a reporter in the board meeting when the town of Vail was saying they were going down this route. That's great perspective. Thank you.

You know, Toma, there's some human processes that have to change in order to adopt some of these things. The fire detection and prevention for instance. I mean, how you plug, you know, the, I don't know, a thousand cameras into, you know, somebody's gotta at least take a look at it and verify it or even validate that this is a real problem.

And then, uh, dispatch things, ai It all and we'll show it to you once again guys. You have to sort of disbelieve chat interfaces and what people call ai. This is true Agent ai.

Um, the next slide, I think we're almost there. It's gonna start to show some of these use cases and they should literally put you guys in your backseat. Yep.

So this is on the 5 0 8 compliance. We've, uh, we've, we've talked about this, so I won't go too much into, uh, into kind of the description 'cause I've already done it for you. Uh, but I think this is one that we really have, have a great deal of passion about.

Not just because it cuts across so many different areas, but because in the end, it, it really is AI for good and, uh, and trying to do it in a good way. Um, so I think we have, we can, we can watch the sizzle video. It's like 45 seconds and super fun.

Um, but then we can also show the demo of how it actually works. So this is going to be the, um, the actual compliance demo and I'll let you Yeah, I'll try to help narrate this. It's getting there.

There we Go. Um, so this was pre-release on, on Vail. So we actually just sort of picked on the poor city of Lakewood.

Um, so it actually is an agent, you pointed to a particular website. That agent then loads up a computer use and actually opens up a web browser and then starts to interpretate not only the code, but it downloads every PDF and image that's sitting there and actually inspects the metadata from a visual language perspective. So we have a visual language model mapped up with a large LLM mapped up with a small LM that's actually driving, uh, the computer use.

By doing that, then it will actually give you the concept. And from here you can define the scope because very large websites further, this is the individual compliance level. Each state, uh, has their own level of 5 0 8 compliance, and then they've been wrappered into about three different tiers across the country.

Then you select which, uh, level you want, the actual website, how many pages you want the computer use one to go through. 'cause this can chunk for days and weeks and months, depending on the hardware you have and the amount that you're actually having it try to do gives you a quick pre-audit. This is a overall glance just to see, all right, where might the website sit?

So then you can tell it, Hey, I want it to do a much deeper dive or just focus on this particular stuff. And this is if you've run it many, many, many times, 'cause we expect once it's audited, you've gotta keep it audited and keep it going. So before remediation gives you a pretty good outlook on what you have to do.

It goes into the individual code of the HTML. It goes in, if the metadata structure's there, it goes in, if the individual documents and PDFs even have metadata, if they can't be at it because they're an older version of Adobe, we actually reach out to Adobe's, API, we reconvert all the PDFs, redownload them. Now they're metadata structured enabled, and then we can start applying the actual metadata in the remediation.

The automatic detection will then understand is the actual metadata that's put into the documents matching up with what the visual model sees. So now you have a visual model interpretation of the metadata to match the actual documents, to match the pictures, to match the graphs. Then you actually get into all the tagging and underlining code.

Does the actual text in the, uh, HTML have enough contrast level to meet certain, uh, visual issues, et cetera. So it's measuring anywhere between about 500 and 800 different remediations per page. And even the metadata structure it has to understand are you oversubscribing it because you could literally have a, a tool that now reads what a picture is and if it takes 25 minutes to read the picture because it put too much in there, there's even a structure around that.

So is it enough, not quite too much, and then how do you apply it? Then all of it's downloaded. And as Robin said, there is the human and loop feature.

You still have to actually as the user push the HTML document back up, push the actual pictures back up into it, and then it reruns the compliance and see did you change it? Did you not change it from the current one that's there? This whole process, even for a small organization, let's give it three to four weeks.

It's probably two to three weeks of just the tokenization of it running. If you're using our smaller wedge solution server that's in market, and then you gotta upload it and then you wanna rerun it and rerun it for every website change every municipality difference, et cetera. And it really, like I said, transforms the game from many, many people constantly trying to go through over 800 compliance checks on every single page, on every single document to a full on multi-agent, agent system processing it.

So the challenge is, like most of these websites, there's probably, I don't know, a hundred different websites solutions to use to define the websites I use WordPress. Mm-hmm. I've used other solutions in the past.

I mean, uh, going into PDF and uh, me adding metadata, going into an image, adding text, I mean, that's all good because it's kind of behind the scenes, but you're gonna change the contrast. You're gonna change some of the, the attributes of the webpage we're talking now you have to play at the, at the level of whatever the web solution Is. Correct.

Uh, that's a cool part about actually using computer use on this. It actually looks at it first as the user, then it downloads the actual HTML code and it can recognize from within the HTML code what systems being sub used, and then it makes a recommendation being actually changed that to HTML code or here's the reference documents for WordPress or whatever to make those particular unique changes. So what have been the day two challenges?

Because this is a complex distributed system. Yeah, and this is great once HPE and KAA is there, but when you leave, someone has to maintain this. So SHI full ongoing support through, uh, the lifecycle of this from, uh, uh, integration and support perspective, uh, KAA does a step away.

Anything that's a bug in the outcome and the outcome-based support, we're there. You buy our license on an a RR model between those two things. And in fact, I believe SHI even is the point that they'll get into those post remediation services and helping them keep updated on it.

So we're really trying to bring in the Unleash AI partner ecosystem and make this sort of be a literal turnkey. And once again, this is a wedge. This is gonna show amazing use of multi-agent capabilities.

They can upgrade our license, they can upgrade the server, and they can start to expand that out to more agents, more capabilities. But we come in with this and we go full smart city, it's, Yeah, Shout out to the implementation partner too, because we focus on the technology here, but, uh, The actual implementation and making it real and putting it in place is very important. And that, and that's why we took, I took a strategy of engaging with SHI and they actually did the PR on it, and we did supporting blogs and, and focus was because of that implementation and integration strategy.

Um, exactly to your point, Keith, around being able that we can go and get it day one, but people forget that, then the next organization pushes up whatever in heaven's name, they might push up and we can almost assure that it won't be in compliance because, and you know, and that this is, this is something that because you're investing in an ag agentic solution, you now have this at the ready to continue to run every week, every month, it doesn't matter. Go through that remediation and continue to service it. It'll be interesting to revisit this three years from now and see what tentacles have gotten itself into this project.

Yeah. Because this becomes kind of a, without, without it trying to become, it becomes a system of record or a core system within enterprise. Most agents, actually, I've sort of reversed some thinking on that.

The agents themselves start to become their own systems. Yeah. I think we have a couple more to, yep.

I was gonna ask a quick about the, are you looping this in at all with the development pipelines and having, you know, pull requests being generated? Sounds like you've still got A, you actually literally can do, uh, not only automatic pull requests, but you can actually connect it to your GitHub system and it can even update documentation and everything. Okay.

I was surprised you Didn't mention that out of the box with this. I mean, there's the, there's the, there's the practical use and then there's the, what I would consider more tactical, expansive use. I mean, town of avail, very extensive website to actually, I mean, the amount of languages, 35 languages, like et cetera that they have to go through.

But now take this up to the state of Colorado, LA County. Like it starts to get very, very, you know, integrated that you have to implement. Mm-hmm.

Come on. There we go. So now, digital concierge, as we said, we've touched on this, we kind of explained the use case Hamza as the agentic brain and, um, an SHI with the digital, uh, civic ambassador, uh, who can, uh, who can then put kind of the interface, everything natural language, asking the natural language questions, but again, making sure that everything is guardrail to that responses only come back from, as we described earlier, what the, what the ambassador has been exposed to and what's appropriate for that particular use case.

But there's not a demo of that because that's something that is, uh, that is, uh, kind of live interaction and is a little hard to do that on a real video. The, um, so the next is around the fire detection and prevention. We'll lean in really first here with the fire detection.

That's where we've been spending the most of our time, where we're able to reuse cameras that they already have that are pointed at different spots in Veil. And working with the pro HC and VIO combination in combination with Kami Waza now being able to look at the, the impact of that. They've detected an incident, they've detected something, but now let's give context to it and let's be able to build it out.

So here we go. So, but just back up real quick. So who collected this amazing stack of 12 vendors too?

I did. You did. Well done.

Thank you. That's that, that's part of the Unleash ai. That's why we started with the Unleash ai and then, and then my proposition was that, um, that cities were getting, they were getting a lot of point solutions, but they weren't getting a city solution.

And that's why I wanted to bring in this different approach from working with like Kami Waza to say the agents can start to bring this together and start to harmonize all of these solutions. So HPE, This is, this is our sexy one and our unsexy ones the deed stuff. But it's amazing the impact difference, uh, of Bat 'em all.

But this one's really impressive. So what you're seeing here is, uh, VAIO is actually doing a live detection. It uses ML models to put descriptors and tagging in it.

Those tags are then fed, uh, into pro hoc, pro hoc, uh, upscales them effectively. Uh, once again, not changing the pixels, but making it more, uh, digitally transparent here. We can do another pass on it so that the visual models will get the metadata tag and actually be able to see it.

And now we can actually apply, um, uh, this is the 3D rendering, the black shark, which then gives us a geospatial on what's around that fire, what is the impact of it, and what we should then know could be, uh, um, a possible larger cause. Then we actually pull in live time feeds from noaa, space, transportation, satellite imagery, um, moisture responses, and about six other things. From there, we actually are able to create a live approach, which we notify then, um, the, uh, fire department, uh, two different levels.

First, we're able to give them a full report on where the fire was detected, what the impact analysis is, and what the urgency rating of it is. Uh, they literally, during red flag days, actually have four time people that are just driving around the county to respond to these things. So actually being able to redirect them, uh, have even less is gonna be a massive ROI uh, difference.

But more importantly, if all of that data comes into effect and it says it's an urgent one because there's high timber rating around there, there's high impact associated with, uh, the homes. Uh, and it's a red flag day, it's low, and we have even more coming in. You can literally just have a full report go out to the city planners, the city organizations reverse 9 1 1, and the whole gambit sort of raised the army up to go at it.

And this is the auto generation that auto report then generates. And it's really neat because now you also have the digital concierge, which could actually answer questions about, uh, where the smoke in the fire's going on as, uh, local residents are asking about it. You have the city planner and the city groups, which are gonna get their immediate updates on the fly about what's going on with there.

And then you get that report sent immediately to the fire department so that they can get a, uh, interaction and reaction off of it. Uh, tying all that together blew their mind. If, if, if you think about it, it's not just a picture, it's that full analysis.

It's that full reporting, it's the full updating structure, it's the full workflow that, that kicks off is something where the fire department went from, well, we have this, we have these people doing it to, oh my God, we get this, what else? And they just started asking us, just use case after use case after use case after this. And it's sort of that hair on the back of the neck moment where you've got that breakthrough with the client customer and they see the power of what AI can sort of do.

5 came out is that it was really a lot more of a quality improver than a productivity improver. I mean, productivity follows. This seems like a great example of that where, um, you don't necessarily have an immediate staff impact, but the vision, you know, or the, the Yeah.

The vision of each human actor involved here, manager on, on down in increases terrifically. Um, yeah, I think one of the interesting things that, um, when we was, um, Russ and I have spent a lot of time together at GTC for the last two days. The, um, he was talking about that in, in July, which is typically, you know, that fires are now way of life, you know, in, in, in, in the west and, and in the mountains.

And it's not a question of whether you're going to get a fire, it's a question of when you're going to get the fire. And when we actually started these conversations, it was incredibly impactful because they actually had a fairly good sized fire about 30 miles from them. Yeah.

Why were in the meeting, having our conversation, literally fire couldn't even show up that one day because they were having to deal with it. Um, and so the, so it, it was, it's very germane to what they're looking at. And their highest rate of, uh, fires, of, of fires being kicked off is from lightning.

You know, dry trees, beetle kill trees, lightning comes down, bad things happen. The, um, and so they had to, there was a report that Lightning had struck and they had to send people out into the woods at like two o'clock in the morning. And, and that's actually fairly dangerous.

You know, how many people are you sending out there? You don't know. Was it really a strike?

Is there really, is it more fog coming in? Is it actually smoke? Is it The brains supposed to go on for the next four hours?

So it's a remediated issue that you then can watch in the Well, yeah. What struck is this mix of that, what struck me in this example is I could imagine the, the whole process you're describing happening without ai, without an agent, because it's somebody looking at something, there's eyes on glass saying, oh, here's 24 hours A day, Three or four people, and then they're consulting with other people, and then a manager's coming in who's read the recent NOA, you got it for all these information sources are coming in. There's a speed factor here.

But I also feel like, um, the, the, when I said like, the vision imp improves or increases, granted, you know, you might need to deploy more cameras in more places and everything. I understand all of that. But, but the, the, the, the speed and the volume information, volume, unnecessary information being filtered out, a certain amount of decision making happening, not decision making, but I guess filtering is the Right word.

They gonna noise ratio go exponential. Yeah. That, that allows for the decision to send people out dangerously in the dark.

Right. Happen less frequently and improves the quality of the service. Mm-hmm.

Exactly. And exactly, that was exactly where I was going with that. And, and that's very much where, where their focus is, is you're not gonna be reducing people, but you're going to improve what they're able to do.

Mm-hmm. And to your point, the speed, um, because before you had different inputs of data and this is able to take all of that data at once and be able to bring that context to it immediately. And, uh, and I think that that's really, um, I just get so excited working with them and seeing what we can do.

Yeah. I want to, I want to add that it, it's always a little contradictory sounding. Um, when, when someone like me, uh, complaints about how dumb AI is in general without a human loop, especially generative ai, but then says that AI improves quality.

The way I square that is that, um, with more review cycles or more ability to go repeatedly over the same thing, that's where the quality improves and that's where the speed and, and new data, And once again, unbound tokens, we can do that such at an exponential level, the re-scan, the reprocess, the re capability, the re-input of net new data. Mm-hmm. Yeah, Exactly.

And a lot of that's also why it's so important to have it within their data center. And to be able to do that, You're not gonna do 10,000 data pulls to the cloud and try to, you know, one extra millisecond would make that a 10, literally 10 minute increase. Yeah.

So, so that's why it's so important, kind of like bringing this whole, this whole ecosystem together. And then here's the code indeed assistance. This is, this will be the Code is like code enforcement, not not programming code.

I wanna make sure. Yes. Yeah, I was, I was like, and let us be very clear on what that code is.

The, um, and so this is again, um, really focused on Kami wasa much like the 5 0 8 was, uh, but on a really different use case. And, um, and as they, they keep saying, they keep apologizing that it's not a sexy use case, but it is sexy to them. It's sexiest, but it is because of the impact and the ability that they have to really look at redeploying people to, uh, to customer service rather than, um, manual lookups.

Uh, so literally everything is still either written. The vast majority of their data is in microfish, and you guys are gray enough hairs. I know you know what that means.

So we actually plugged, uh, the API into their microfish system. And you can see here a document all the way back to 1976. You can see the barcodes and everything else.

And it's actually live interpretating them, even though it has no idea about the previous system and the setup on any of these, puts it all into our ontology system. The ontology system then puts it all into a metadata structure and a parkade table. And now every single deed goes through these 60 compliance steps rebuilt out.

You can see it was about 7 million tokens per process that it actually runs. Just to do that builds a full ontology schema breakout, cross mapping every like deed for like service functionality. And from here, literally now, a single person could do the job of nearly 10 people on the fly just by inter interrogating it.

We also put a nice little chat interface on it so that eventually the public will be able to ask about their own deed. If they're trying to buy a house, they could ask about what is the control and the restrictions on it. If they're trying to rent it, they could see it.

But furthermore, you saw that sort of compliance mesh that we put together, the ontology mesh. Now they can see all of the individual deeds and licenses. They can make sure that they're all rent controlled and rented to the proper people.

They can all make sure all the purchases were done within the rate of inflation that's allowed. All of that's completely done. And like I said, via all of the documents where every single time, every single question, every single sell, every single renter, every single process was a manual lookup of a physical document or microfiche.

And unless it was in the last eight or nine years, then they had a digital, uh, of it. Now it's completely interpreted and all run through that, and the agents literally do a live process feed off of it. We didn't displace the actual staff that's there.

We've displaced the net new that they were, uh, already qualified and budgeted to hire, and they have 5,000 properties and they're trying to move to 25,000 by end of next year. And now they can do that completely non-linear from a, a growth of actual, uh, uh, FTE count. Uh, and I think it's somewhere in the range of nine to 10,000 impacted people yearly.

'cause these people come in seasonal, they come into rent, they come into either buy or, or work within it. And now this whole processing thing can be run 24 hours a day, seven days a week, and they can get that access. So from a real practical perspective, specifically around somewhere like Vail, I can have a inverse require inquiry.

If I want to buy a property somewhere or rent a property somewhere, I can ask the question, how many properties around this address can be short term rent? You got it. So I can now make a personal decision.

It's on the code base, whether not I can buy, I, whether not I want to buy that property or not. Based on the trends around that Was that rational, uh, spiderweb you saw that breaks that all out and allows all of that from a full ontology understanding. So it literally understands all of that data and its interpretation and then how you access it or what questions you want, it can give you a full result.

And so I think, and that's why the goal is, was first to help the, the actual housing group to be able to, to better service requirements and needs. But as Luke was saying, now the next phase or the next step is, is that chat interface is being able to enable the public to re to access publicly accessible records. To, to share that type of yeah.

Information so that you don't have to call the nice people at the housing organization. You can actually just look it up yourself. And then code enforcement and goals.

Um, working with Black Shark, we have 3D models of all of the properties, et cetera. We also will understand all of the use of energy outdoors for like snow melt, et cetera. They're only permitted, they have so much net new homes.

As you uh mentioned, Keith in that particular area, can't get a permit for any external usage. Now you'll have full understanding of that. They're also working with us and Black Shark to get a full understanding of all of fire prevention tiles on the homes, uh, cross area from, uh, how far the tree is to the home.

And they'll want to pass a law that actually says you have to build back so much and, uh, clear so much. And now people will get realtime 3D rendering of their individual properties, how they'd be impacted so they can see how the vote would actually impact them. I mean, this just changes the entire policy and process and just living in that particular area.

And it does it nearly overnight at such a high level that it's just astonishing. A question online here, is there any chance of encapsulating the bad part of redlining, Bad part of redlining? I don't know if there's a good part of redlining.

Well, red, red lining is yes. Uh, yeah. It was like, wasn't that the inherent definition of red lining?

Yeah. Um, the, the, the way I, the the way I think about it is I, I, I think policy, um, should be implemented and voted on for the good of the community. And, uh, and the enforcement of it, I think is now being able to be driven by ai.

So that might be sort of a bad part of redlining, but more importantly, like I said, this new policy that they're trying to implement, they want everyone to really understand how that's gonna impact their particular, you know, trees and their particular, uh, shingles and everything else. And to now get an actual house by house visual image of what they would actually require of them allows the, uh, uh, um, residents to now make a real informed basis and decision. And I don't think we could literally say that almost anywhere in our electorate system.

And the fact that this could be rolled out nationwide, uh, the fact that, you know, LA County could understand what it's gonna require to avoid another palisade and that kind of stuff, I, I, and actually have a visual representation of it, I, I think is where AI has taken AI for good in the long term. Yeah, I think that was one of the things in, I I really, um, you know, the partnership with Fail has just been, you know, I can't say enough about how terrific it's been to partner to truly collaborate and partner with them, um, in doing this. And, and you see kind of what we've been able to, to accomplish together, um, from, from kind of August to now and then then announced yesterday at GTC.

But continuing, you know, as, as Luke's talking about, you know, kind of the vision, the next phases of, you know, how we support kind of that citizenry, because it sounds kind of silly to say, well, you know, you actually need to have your trees like four feet away and like branches cut down and things. But when that fire's coming, it actually really matters and then it's really gonna be too late to do it. And that's what their passion is about, is how do they keep people safe?

But by bringing in everyone and by being able to leverage AI to do it in a more holistic and more informed way, is, is to me something that that is, that is truly special in how they're looking at approaching this. And I just wanna double down. It doesn't get more dirty than the data from a Microfish if we were able, uh, this isn't a data lake first conversation, it's an outcome, first conversation.

And AI literally makes that transformative in thinking, and we can't express that enough. That is literally the mental incumbent that most enterprises and organizations have to understand. Right?

And so this is, you know, as we go forward, we're already having other conversations with other municipalities. Um, one of the unique things about Vail if, uh, if you've never been there, is that they don't have any traffic lights. Uh, so, but that's something that is roundabouts.

Yep, roundabouts, lots of roundabouts. Um, but the, um, but as you start to think about, you know, kind of traffic and parking, parking's definitely an issue there. Uh, being able to continue to scale out in all different ways.

And what we look at is essentially building that community that as we add additional use cases and across this, um, across this overall solution, that those will then be added for everyone who, who has the solution. So we're continuing to look out across all of the different points of, of areas where people are interested and being able to build out kind of through the app garden for where people want to point, integrate, you know, kind of those departmental solutions into this. Making sure it's a very open approach, uh, to, uh, to do that and, uh, and build this out.