Utility field teams face increasing pressure to execute with speed and accuracy.

Hello everyone and thank you for joining us today. Welcome to our webinar on AI in the field and how image recognition is transforming digital field work. Before we get started, I've just got a couple of quick housekeeping notes. We'll be holding a Q&A session at the end of today's presentation. So if you have any questions, please feel free to type them into the question box at any time throughout the session. We'll try to answer as many as we can live and for those that we don't get to, we will follow up with you afterwards. You'll also see a few poll questions pop up throughout the session, so be sure to vote on those. And with that, let's get started. I'll now hand over to our host for today. So over to you, Adrian and Thomas. Adrian McNaughty, Thanks Kelly. Good morning, good afternoon everyone. Welcome to today's webinar. My name is Adrian McNaughty. I lead the utility practice at IQGeo. The theme of today's webinar is a really exciting one. It's AI based image recognition, but the focus specifically on utility field operations. It's a topic that's gaining a lot of traction and some of our customers who we've already spoken to on a topic, the feedback has been tremendously positive. Back in March, right, I was actually at Distributech. We announced that we're in the process, IQGeo, in the process of acquiring a company called Deepomatic. And today, we're actually going to present Deepomatic solutions. Deepomatic is incredibly complementary to IQGeo. We're both focused on the same core markets, utilities and telcos, and also focused on improving field operations for our utility customers. So today, Thomas, my new colleague, is going to first introduce Deepomatic, introduce the technology, and then going to give some demonstrations on several different use cases. And we'll ask questions as you go along and we'll save some time at the end to answer those questions. So Thomas, over to you. Thomas Thomas, Thank you, Adrian. And thank you, Kelly. Good to meet you, everyone. Good morning, good afternoon. Good afternoon. So as Adrian said, I'm working with Deepomatic. I've been with the company for almost seven years now. And I've spent the last three years in the US to build our relationship in the market, have new relationship with customers here. And I'm really excited about our new partnership with IQGeo. So to get started, I will tell you a little bit more about Deepomatic as a company to explain where we're coming from, what we've done in the past couple of years, so that you get an idea of who we are. And then I will jump into the solution itself. So Deepomatic is at its core a technology driven company. We've been focused only on computer vision for over 11 years now. So we have a big background in research and development for this technology, which in a nutshell, basically mean that we can train intelligent systems to recognize visual elements, visual patterns in photos and videos. And we've applied this technology to a couple of different industries in the past. But for the past five years, we're focusing exclusively on using this tech computer vision for field work. So we started with specifically telecommunication network, a lot of work on fiber deployment. And then we've had quite a few customers in the utility space, in water network, also electricity network, gas network. And I will show you a couple of use cases that these customers have used Deepomatic for. So since we've started working on field service, we've grown our customer base quite a bit. Today, we're proud to have over 30,000 field technicians who are using Deepomatic for their everyday job. And so that's both for telecom and utilities. And with this number of users, the volume of data of jobs that we processed last year was over 20 million work orders. So you can understand a work order as being the installation of a new asset, the installation of equipment for customers, such as a smart meter or maintenance or inspection of assets, which means in terms of volume of photos, last year we presented a couple hundred million photos that we processed. You'll notice that quite a bit of the logos of our customers are European. If you recognize them, that's because Deepomatic is a company based in Paris. That's where we have the majority of our team. So over 50 people are based in Paris. And I have my team here for North America on the East Coast in New York City. So that's it for the quick introduction of Deepomatic. Deepomatic will jump into telling you what's our approach to computer vision for field work. But first, I think, Adrian, you have a quick poll for us. Yeah, we just want to gauge where the audience is in terms of using imagery in the field today. For the first poll question is during field operations, do you take photos in the field today? So we'll just give that a minute to come in. Okay, let's give it a few seconds and then I'll close the poll. Okay. And then I'll share the results. Great. So 86% of people using them today, that's what we expected. So that's a step in the right direction. So what you'll see today is how you can actually leverage what you're doing today, taking imagery to enhance your field operations even further. Okay, back to you, Thomas. Thank you, guys. So why is computer vision something we have been interested in putting in the hands of field worker? Well, the first element that jumped to us really was taking the approach from the quality control perspective, as well as the documentation of assets that are distributed in a large area. And what we've noticed is, well, very similar to the result of our poll, is that most companies doing field work are capturing quite a lot of photos. And if not photos, asking their field crews to fill forms to capture information because you have to make decisions on what your asset really is like in the physical world. And so that has been the approach so far. But what we noticed is that even if you capture a lot of this data, what happens is to be able to utilize it and make sure that the data you capture from the field, as well as making sure that the field work is done properly, that requires manual work to review that data. Because if you don't look at this data, if you don't do any action to make sure it's good, any action to translate whatever information is in the photo into a digital form, that just means this data sets idly in your system and it's not very useful. What we also notice is because that review and that usage of the field data is manual, it means for the larger companies, very hard to actually use all of it. You generate a massive amount of data, which means that when you want someone to look at it, work with it, it's going to be on samples. And that creates blind spots. It means you don't always know exactly where things are, how it's built, in what state it is, if it's labeled properly as you would require. And if you have any accident or issues and you want to look back at the data you've captured in the past to make sure everything was done properly, in many cases, it will happen that when you open that package and you look at the photo for the first time, this is when you realize you're missing photos. Or that's when you realize that some of the photos are framed in such a way that they're not useful for your current situation. And what that means ultimately is it creates costs, it creates delays, and overall it generates a bigger potential for costlier problems and more difficult fixes. And so I'm going to illustrate this with two slides. The first one, and I'm putting photo illustration just for the sake of giving ideas of the different things that Deeponic is used for. So right now you're seeing a smart water meter. And so the first application is going to be, well, when someone installs an asset, when someone works on an asset, there is always a set of quality standards that you require them to follow a process. It's just difficult to make sure they follow it every single time. And not necessarily because they don't want to, but also just because they can make mistakes. That happens quite a bit. Everybody's human. And so these mistakes in the field are a first area of cost. The second thing is, I indicated it's hard to make sure you get the right data from the field. So not only do you have little mistakes in the field and they might not be critical, but they will add up. You will have more and more of your assets that are not built exactly as you expect them to be built or maintained or operated. You also do not have the guarantee that all the data, all the documentation that comes from the field is good. So if you are only sampling images, it's very likely that you're missing images. It's likely that you have images of poor quality, which means that at some point you will hit a problem where when you need this data, it's not going to be there. And so the approach with computer vision is to enable the technicians that are there, that are performing the task in the field, that are building assets, installing equipment, inspecting assets, to self-validate directly on the field in their mobile device with the AI. Meaning they take a photo, the AI has been trained specifically to look at this photo, to look at specific elements in this photo, and will provide real time feedback to the crew, to the technician in order to validate they took the photo properly. So if the photo is missing, if it's not taken properly, they will be notified so they can correct that. And if there is any mistakes in the work itself, visible in the photo, then this will also be flagged in real time and will be indicated to the technician how to fix it in order to leave the site in the best state possible with all the information that you need. And the benefit of this approach is whenever you work on assets physically, you get a higher rate of first-time rights work. Meaning you have much fewer situation where a crew will leave the site with some issues. Issues that maybe you'll find with the quality control you perform afterward or the inspection you perform afterward. Maybe you won't find, but even if you find it, it's going to be after the fact and you need someone to send someone back. So this will be prevented. It also means you get much higher visibility. The AI inspects the photos in real time, provide feedback on your assets to your back office crews in real time. Which means that if there is any maintenance that needs to be done on certain assets degrading, you'll know right away. And you won't have to guess if the data you capture in the field is good or not. You will have the confidence that every piece of data has been looked at. So how do we do that? I'll give you a quick overview of the DiPomatic solution and that's where we'll be jumping in a demo. But just to give you an idea before I jump into the demo, and I believe we have a little poll just after this one. Really what DiPomatic does is we train specific neural networks. So the neural networks are this intelligence system, this AI that we train specifically for the tasks that you need us to analyze. We'll train these neural networks for different photos, different elements in the photos that we will then integrate into your field mobile workflow. So that every time the technician takes a photo, you can see in the screenshot on the right, they get a feedback. We can help them fill form, et cetera. And this is what I will be showcasing. And all this data then gets centralized in the back office and we can then tie it back into other systems such as a GIS system like IQ. Geo to improve your system of record. But that can be sent back into a whole range of other system, maybe to improve your business intelligence dashboards. So we'll do a quick poll and then we'll be seeing a demo from me. So Adrian, take it away. Yes, the second poll question again, just to really gauge where we all are in terms of using AI technology in the field. So do you or your organizations use some type of AI to support your field operations today? We'll give you a minute to respond to that one. Okay. If in the meantime, you have any question regarding what I said, it's also a good time to type your ideas, type your thoughts. Okay, just a few more seconds and then I'll close the poll. And then you should be seeing the results now. So, okay, that's 83% no and the rest in the future. So that's interesting. So that's why you've attended today to learn more about it. So now we're going to jump actually into demonstrations of some different use cases. Yep. So let me switch screen. I will go straight away into a presentation of our standalone mobile application and back office tools. Adrian, could you just confirm that you're all seeing my screen with the demo right now? Yep. I see the phone app. Perfect. Thank you so much. So this is the mobile application that you can see. What you're seeing right now is our web app. Typically, customers will integrate this into the mobility tool that they already use so that technicians don't have to switch between several different tools. We don't want to create extra friction. So we can get into that maybe in the Q&A, how this can be integrated. But for the purpose of the demo, you'll see our standalone application. So I can get into a specific job. You can see that the job is already populated with several photos that I'm required to take as a technician. This is going to be customized depending on your use case. And directly from there, what the technician would do is capture new photos through the diplomatic application. That means we'll be able to geotag it, timestamp it, and we'll have the ID of the technician as part of the metadata. So as you can see, I took a new photo. This is the photo I just took from the field. And I'm getting the feedback from the AI pretty much right away. And that will allow me to validate that my job is done properly and that the data I'm collecting from the field is done properly. In this use case, I'm just doing a quick poll inspection. We've set up that example just for the purpose of our webinar today. And let me explain a little bit what are the different feedback you're seeing here. So the first one, image type validation, this is pretty much going to be for every single photo that we will configure for your solution. This is the one where we have a specific neural network that will validate that the photo captured is acceptable. It's the photo we expect. It is usable, meaning it's framed properly. It's not too dark, not too blurry, etc. So did I take a good photo? This is the very first thing we'll be checking. Then within the photo, we'll be able to check the different elements visually that the AI is. So what you see here, this square, this is what we call a bounding box. It's the technical term basically to indicate what the AI is seeing on the photo. And that's another thing that's very powerful with Diplomatic is you can see very clearly what the AI sees. For different use cases where potentially we're checking the quality of work, it's going to be very important to be precise when we indicate to the technician, we detected something wrong on that photo. You can click on it. The AI will show exactly what we're seeing so that technician doesn't have to guess what they have to fix. So for this use case, all we're doing is detecting the neutral line. We're going to check the attachment for joint use. And then we're doing a quick analysis of distances just in order to make sure there is enough clearance. So again, this is something we've just put together for the purpose of that demo. We could be looking at many different things on that image, different type of equipment, different elements that would help for whatever task they're performing at the given time. So once I've done that, I can be done. I will go back into my job and continue the rest of the work I'm assigned. I will show you quickly now the back office part of our application. This is where as a project manager or a supervisor, I will have access to all the jobs that are being performed on the field. And this will be stored indefinitely as well, which means you create a system of record, a repository, let's say, of images, of analysis of all the jobs that are being done. So you can see that here I have the job I was working on just before. We have the timestamp. We have the progress of the job and we have the photo that I just uploaded with all the information being captured as well as the AI feedback on the photo. So this are all the type of information we would be able then to push to other systems for further automation, to update your design, to put into your GIS system in order to slowly create a very accurate system of record of your asset. It's also very useful when constructing a new piece of your network, installing new asset to monitor both your internal workers, but also contractors, subcontractors. And make sure they are performing their job properly. So this is it for the standalone diplomatic application. I will jump into a different screen and quickly show you what a general integration into your tools can look like. And more specifically, what an integration into IQgeo's workflow manager would look like. So here I'm on the map based mobile device, mobile interface of IQgeo. This is something I would be able to access from my tablet, for example. And we'll have different tickets assigned to me. And we can see that here, once I get into the ticket, we'll have a diplomatic button. This is the one you're seeing here where I can click on it and it automatically takes me into the diplomatic interface. And so that's what I was talking about when indicating that you don't need to navigate between different applications installed on your phone. Within a single application, you'll be able to capture this information without having to log in and log out. So I'll take a new photo and you can see that in this one I'm looking for slightly different things. I will be counting the assets here that are on this poll. You can see also that the view is a little different. I was just in the mobile format before. This just adapts to my computer screen. It will adapt to any size of screen that you're using. So on the tablet, it will be no problem as well. So once I've captured my photo, I will go back and you can see I'm taken instantly back into the workflow manager, into the ticket I was working on. So again, I'm not disrupted in my flow. And you can also see that we have captured in the form the information. So everything the AI sees in the photo is something that we'll be able to fill in a form as well. Read labels, read meter indication. Maybe you're using a ruler to measure things. All these information will capture and will help you fill that instantly without having to manually input it. And the other thing is you can see that for that specific poll, if we look at the number of assets that were recorded previously as being associated with that poll, is not exactly what we're seeing in the field in the photo. So this is an example of how with something like Diplomatic, you'll be able to update your system of record as you go. Meaning that if you have some problems with your data today in terms of the confidence as to what you have, having Diplomatic as part of your day-to-day work means that we'll be able to implement systems that will look at everything that we see in the photo in order to make sure that you record the real world and update your digital system of record with accurate data. So now I can close that ticket and I'm done and I can jump to the next job, to the next ticket. That being done, I will go back into my presentation. You'll see that we put a couple screenshots of what I just showed you. This is for you to have as reference once we send you that deck. So you have here pretty much the whole workflow I was showing. It's just on a different example. As well as the back office indication as to on the left side, this review dashboard that I was showing where all the data is centralized, where you can review the photos that are being taken and analyzed by the AI. So if you ever have an incident on a specific asset, you'll be able to really easily go back to the photos taken at that site for that specific asset. And you'll be sure to find good quality photos that the AI has approved as well as a lot of additional information. On the right hand side is an example of what a BI dashboard would look like. This is something that we also offer as part of our solutions package. We can customize different graphs, different indicators that would allow you to monitor your operation, monitor your contractor. But obviously, if you have your own BI capabilities, that's something where we would connect and send you all that data so you can use it directly in the tools you already have. And that's something I'm exemplifying here. This is a pretty simplified version of our entire solution, but it's really to indicate to you the entire architecture. So at the very center of it is the heart of Deepomatic is where we train the AI, where we package the data set, where we update and upgrade the different performances of the solution that we deploy. And it connects on the left side, so what we call lens is going to be the area where the technician are using the photo capture and the AI feedback on the field. On the right hand side, what we call engage is going to be that back office capabilities in terms of reviewing photos, reviewing job, as well as building the ash dashboard. And at the bottom, this is really all this data that we capture can be leveraged into other system. So all our different customers have connected Deepomatic to a whole range of different tools in order to provide further automation, further reporting capabilities, etc. So this is what we're showing in this slide. Again, will be shared with you at the end of this presentation, so you'll be able to look at it in a little bit more details. So once, now that I've presented the general idea of what Deepomatic is, I hope that was very clear for everyone. If you have any doubt, any question, please write your questions in the chat and I'll be happy to address them at the end of the call. But now what I'd like to do is give you a couple examples of use cases that our customers have implemented Deepomatic for. Some of them are use cases that we've been, you know, asked to produce and we're working on. And so I hope these will resonate with you. And I also have some videos to exemplify some of them. So the first one is that Street View Poll Inspection. We're very excited about this one because we've recently developed an agent that allows us to basically look in an area at all the different polls automatically with computer vision. And just the coordinate of the poll will allow us to do a lot of the inspection of those polls. So not all of them, there's always going to be little difficulties, but a big majority could be done straight away from the back office automatically. But we'll tell you a little bit more about this one down in the rest of the presentation. Another use case is using drone data. We can do batch upload. You don't have to individually upload the photo one by one, like I showed you in the demo. This is really the use case when a technician is using Deepomatic. But if you capture data in any other way, we can upload them in a batch at the end of the day, at the end of the week. Already, if you have a big data set you want to process, that's something we'll be able to do. Another thing I wanted to show you is the deployment of smart meter. So that's an area where Deepomatic has been deployed quite a bit in Europe. We're currently part of three big AMI projects on water, gas and electric, where we're validating that the asset is installed properly. Those are very impactful deployment in terms of return investment. And if you're about or thinking of doing an AMI project, I think it would be extremely valuable to have a discussion with us. We can share return on investment metrics from our customer and give you an idea of how this could be leveraged. So this, what you're seeing right now, is basically a complete workflow for the swap of a water meter. So this is what a technician would have to take every time they swap a meter. There would be photos of the meter before, its ID, its current reading, and then you have the photo of the new meter. Again, ID reading and then closed. So this would be a full closeout photo package. And for each photo, you would have a lot of different visual elements that the AIs will be able to look at to ensure that the job was done properly. So this is the new meter being installed. So for water, you'd make sure that the valves are in the same position as before, that there is no leak, that it's well positioned, that the antennas is well installed, as well as the pit clean so the signal can go through. This is a very specific example. It can be, again, customized to your specific setup, specific equipment, specific type of network. But it gives you a pretty good idea as to how the solution is thought about in terms of identifying into each photo what it is that the AI will have to look at to validate that the job has been done properly. Another example that we're working on is identifying specific equipment. So we're working with a large utility that has identified that some very specific insulators have potential problems. And so what we're doing with them is defining how to recognize the exact model and manufacturer of some of these potentially problematic insulators so that they can be detected. Very different return on investment, very different implementation as to the water meter installation, for example. But it's just to give you another idea of how you can deploy something like Deepomatic into your system. And I'll just spend a little bit more time on the Street View poll survey because, like I said, we're pretty excited about it and we think that there is big potential. And basically the idea is if you have an area where we have several polls, you could draw a polygon to select this area where we could connect to all the coordinates of these polls and extract them and automatically do a survey based on the Google Street View images that are available. The idea is to support your inspection and survey work and lighten the burden, lighten the load in terms of inspecting all these assets and do some of it in the back office. And so that's what I'm going to show you right now. This is a recording of the system we've put together. So as you can see, I put the coordinates of one poll and the system will drop there and we'll look around to try and find the poll. Once the system finds the poll, it will zoom on it and we'll do the processing of the image. And that's where the AI is leveraged to extract some information. And so just like I showed you earlier, what we look at in the image can be customized and can be different depending on what you're looking for. But I think what's interesting with this is just imagine that you don't have to do this one by one. You just input a thousand poll coordinates, ten thousand poll coordinates. The system will run in the background and will provide a lot of information on all those polls. We can indicate how old the photo is from Google Street View. We can indicate if there's a poll there, if the coordinates slightly different, what equipment are on there. A whole different range of things depending on what you need. And the last little demo I will show you is another element we're very excited about is what we call the business intelligent agent. And so basically once you start using Deepomatic and you've captured a large amount of data that the AI will have processed in terms of quality, when it's been done, who's done it. Then we have an agent where you can ask very precise question about your infrastructure and you'll get an instant response. So if you're looking at the video right now, what you'll see is we first asked if there was a correlation between the state of a specific asset and the region where these assets are. And the agent will be able to scour all the information that we've captured to give you a response in terms of what they're seeing in correlating geography and asset quality. And the next part of the demo, what we're doing is going a little bit further and we're asking the agent if there is a correlation between the quality of the current health of the asset and the contractors that are responsible for this specific region. So this is something that we used to configure manually in this BI dashboard that I was showing earlier to have these specific graphs, specific KPIs in order to make this decision and monitor your work and workers, contractors, etc. But thanks to this agent, we can still do this dashboard setup. And if you have any spot question where you're at that specific moment wondering if there is a specific correlation between the information that you have, this agent will be able to give you a very good answer on the spot, which might help you actually decide if you need to monitor this down the line, put together a new dashboard or just make a decision as you need to and as you address current situation, problem issues, etc. So we have a final poll, I believe, Adrian. Yeah. So before we get into some questions, and there's been some in the background, final poll question, you really, really see what your interest is in different use cases. And it's just some of examples of what we've been talking to our ArcGEO customers. Asset inspections, and that's got a variety of different forms. Asset installs, for example, meters that Thomas just showed, even taking it into damage assessment through an outage conditions. If there's any other ideas you have, or all of the above, we're interested to see where you think you could apply this technology in your organization. Okay, a few more seconds. You should be seeing the results now. Yeah, pretty much all of the above. So this really confirms what we've been talking to some of our existing customers already. And I'm going to post my email address in the chat and open it to anybody who, if you would like any more information, reach out to me directly or anyone at IQgeo. I think we're going to address some of the questions that were asked during Thomas's presentation. Not in particular order. So Thomas, we'll try and do like three or four questions before we let everyone go. So the first two are related. I'm just reading from my screen. So can you, I think you answered this. Can you use images that we already have from a different source, i.e. drone capture? So I think Thomas touched upon that around the batch processing. And then the second question is related to that. If batch loading drone images, do you use the metadata to tie that structure to the GIS or the IQgeo client asset? Can you repeat the first part of the question? I'm not sure I got that. No, basically we're using, are we using, if we take in the batch processing, you know, and we're using the images from the drone. Are we using the metadata to tie it to the feature? Yeah. Yeah. The answer is yes. And in regards to the batch processing. So yes, we can absolutely process batch of photo. So a use case would be a drone capture. And at the end of the day, or during the drone capture, you would upload a batch of photos. We can very well do this. If you have an existing data set of photos, we can also process them. But keep in mind for that second one, that we have to train the specific neural networks to handle specific photos and specific things we're looking for. So we've done in the past looking at a big data set of photos, but you need to be looking for specific things, right? So we'll have to train the AI to be able to provide you the information that you need. So that's something we should discuss and see what you would want to be processing. And that's something we usually do as we move forward, as we deploy new use cases for the field that also augment our capacity to look at the data you captured in the past to also extract information from those. And one thing just to add to Thomas, your comment there is with any given use case, there could be multiple models that are trained to look at different aspects within the actual images that have been taken. And this helps with performance. So it leads me to my next question, Thomas. How long does it take to process the results and present the results to the field tech? Yes. Well, I mean, you saw the demo I just presented, so it's very quick. Basically, the analysis of the AI is negligible. It's 0.1 second. What takes time in the current scenario I showed is uploading the photo. So you saw the demo took maybe a second, a second and a half. The biggest amount of time is just to upload the photo. So and we're aware that some of you have situation where you're taking photos completely offline. And that's a big part of why we've focused the last 12 months of R&D building the capacity to deploy the AI, the neural networks directly into the mobile device. And that's something that is now alive. If you work with us, we'll have offline capabilities where basically when the technician takes a photo like this one and they're completely offline, we'll be able to have some of the AI doing the analysis. Something like this photo I'm showing right now would probably be able to do all of it offline. I'm saying some because sometimes we're looking at very big asset with many different things to look at, which require a little bit more processing power. And the AI wouldn't be able to function properly on the mobile device. But yes, it's and if it's offline, it's also going to be the same speed. It might actually be faster because now we don't need to upload the photo. So answer is on average about 1.5 second. You take the photo, you get feedback. Thanks Thomas. And then Adrian, just to compliment your answer, which was really important and very good is to show this photo. We're saying we need to deploy several different AI to look at a photo. That's very true. And that's what makes Deepomatic very efficient, very quick, high results and nimble as well, because we can update the solution with a scalpel basically. And so if you look at a photo like this, every bounding box is at least one AI looking at specific element of the image. So it means we can turn it on, we can update them. But so each image on average will have sometimes as low as five AIs and sometimes as more as as much as 25 different neural networks working on a given image. All right. I think we've got two more. One of them you actually answered just then. We use IQgeo anywhere. Is this application available in an offline mode? So what you showed was the online browser, but you just mentioned if you want to talk about lens for a second. Yeah. So like like we talked about before, I think so IQgeo anywhere is available offline and Deepomatic will be available offline as well. So we'll have the same connector that I showed earlier on WFM where I'll just go back to the end of the deck. So you'll be able to access Deepomatic from anywhere in the same fashion as I've showed and Deepomatic will be available offline as well with the AI offline. And then the last question, there's a couple more questions that we'll send out responses to the attendees after the after the webinar. The last one, Thomas, is are the photos searchable? Yes, absolutely. And so what you saw in that demo I was showing in the back office is you get everything centralized here. So we'll be able to manage access. This is not going to be accessible to everyone, both across your organization and your contractor. So depending on what level of access you'll have, you'll have access to only a certain amount of job. But ultimately an executive would have access to all the jobs, all the photo, and it's going to be searchable. So you'll be able to search your job based on geolocalization, asset type, contractor, defect, anything you can think of that the has captured. So anything we captured visually, any label we read, as well as all the metadata. So again, timestamp, ID, geotag, etc. You'll be able to search and pull a report where you see only the photo you're interested in. Perfect. Well, thank you, Thomas. And thank you to everyone for attending. I just posted my email address in the chat. Feel free to reach out to me if you've got any questions or like any further follow on sessions, where you can dive into any specific use cases in a bit more detail. And I think Kelly is going to share how we'll share the links to the webinar now. So appreciate everybody attending. Thanks, Adrian. Thanks, Thomas. And thank you to all of you for joining us today. We have recorded the session, so we'll make sure a link of that is sent out to you in the next few days. And as the team said, if you've got any more questions, you can reply to that email. And you've also got Adrian's contact details as well. Many thanks. Thanks, everyone. It was a pleasure.