60: The Future of AI Starts With Real-World Data (with Nico Posner)

Polly Allen: I love seeing is that that focus on the customer pain points with their current solutions is really what's driving innovation. And it speaks to the larger question of just like you've been managing innovation now for decades, generally at the forefront of this is the management of innovation changing with these new changes of AI,

Nico Posner: there's a lot you can do with the wonders of software and compute once you have a digitized version of the reality around us, uh, in order to do that, there's a lot of development that are vertical specific for very specific niches. AI can accelerate a lot of that grunt work and the research work that previously would have taken days or weeks of research and might not have been exhaustive or uncovered all kinds of data that AI tools are very, very good at surfacing data from a much larger corpus of underlying data that they have available to them.

Polly Allen: Welcome to another episode of AI Product Leader where every week we shine a light on what it really takes to join the AI world. Real conversation with real AI builders, founders, product leaders, AI experts, technologists and everything in between to find out what that path looks like, whether you're building AI products or career in AI. Have a question or topic? Email us at AI Career Boost SupportI CareerBoost to join in on the conversation. I'm your host, Polly Allen, the founder of AI Career Boost where I spend my time helping product leaders navigate the world of AI and thrive in AI leadership roles. And I've spent my career on both the product management and development sides of the fence. I'm an ex Alexa AI principal Product manager where I launched the very first generative AI answers on Alexa and I'm super excited to bring back one of our guests um, today. Thanks so much to Nico for joining us. Nico Pozner is the VP of uh, Product Management at Q3D Sensing and Nico's career has spanned some of the world's most innovative companies including LinkedIn, eBay, Clover and Xero. Nico's career is a masterclass in innovation with a massive impact across industries from startup growth to uh, some of the world's largest companies. Nico's been working with AI since 2018 with, with incredible results. For example, as a former VP of Product Zero, Nico led initiatives integrating cutting edge AI and predictive analytics into products used by over 100,000 small businesses worldwide. And today he's here to chat with us about a really exciting new innovative, um, launch from his company, Q3D sensing of a product called Origo, a 3D reality capture LIDAR sensor. Miko, thanks so much for coming on Today and chatting with us about this latest innovation.

Nico Posner: Yeah, thanks Polly. It's great to be back again.

Polly Allen: Yeah, wonderful. So we've talked in the past and I will link to the episode above about your background and how you got into AI. Um, I'd love to hear a little more about Q3D sensing and the kinds of business that they're in.

Nico Posner: Yeah, so Q3D sensing is an early stage uh, technology startup based here in Silicon Valley. And uh, we're really focusing on the evolution of AI, uh, ah, to develop an AI um, platform um, for capturing the real world uh, around us and um, enabling the digitization of that so that downstream uh, customers and use cases can be enhanced with the real uh, world data um, that's captured uh, through the reality sensor that we have created and then uh, transform through software into many, many different uh, solutions and use cases.

Polly Allen: Incredible. I just love the name Reality Pincer. In this age of AI when we're like you can't tell what's real anymore. To actually be like something that interacts with the 3D world and makes much like two way communication with AI between the 3D world possible. Um, it's super exciting. Um, I'd love to dive into lidar. I know a lot of people, myself included when I first started talking about this aren't necessarily aware of like can't I just capture images with my iPhone? Like what are the advantages of using a technology like lidar?

Nico Posner: Yeah, so lidar has been emerging as a technology. Lidar has existed for, for decades. Um, and some of the use cases that people are familiar with are the LIDARs that are on top of the self driving cars that are now becoming more prevalent in many. These, um, those are LIDARs that are developed for longer range and high frequency, uh, to make sure that the self driving cars don't run people over. Uh, but there are other lidars that are useful for many other use cases. Uh, Apple has integrated a lidar into their pro level, uh, iPhones and iPads and those lidars uh, can be useful for up to about 5 meters or 15ft in range, uh to capture uh, the environment around them using that technology. Those are very useful for many use cases. You could do some basic scanning of the environment around you, um, and produce some basic models. But um, our product is a lidar that uses a different base technology that gives an extended range. So um, our lidar technology can capture data as far as 70 meters or about 210ft. Um, but our focus really is between the 10 and 30 meter range. Um, optimizing, because that's where up to 100ft, which is where most of the real world business use cases happen. So um, if you think like what they call aec, architecture, engineering, construction, there's a lot of work and uh, investment in that and those industries around, um, digitization of the real world environment, uh, to enable those companies to then understand what is or uh, and then merge that through software and AI, uh to what is and what may be and then track construction projects from start to finish. Progress tracking you can do tolerance, uh, mapping around. Is this built to spec you can do with the wonders of software and compute once you have a digitized version of the reality um around us. Um, and so in order to do that there's a lot of development that are vertical specific for very specific niches for different use cases in interior design or construction or video game world development or robotics or drones. Um, but all of that data relies on uh, a data set that reflects the reality of the world around us. And in order to do that you can capture some of that using for example the iPhone lidar, or you can use it using the device that we have, which is size lidar, which attaches to a smartphone, ah, or tablet. And this can capture much greater range, much um, greater uh, depth of um, data. And to then extend that um, range and the data capture for these downstream use cases, our product really will extend um, the ability for people and businesses to capture the real world reality to then export that data either in various standard file formats or a data stream, um, to then do things with it. Whether it's insights, whether it's driving robotics or doing analytics or data processing or all the magic that you can do with the data and the insights that you can drive from that through software as ah, a downstream workflow. But someone needs to capture that up front. Yeah, building the hardware product and a software solution that accompanies uh, this device to be that front bleeding edge of the data capture into the ecosystem which will then feed a lot of that downstream magic that happens thereafter.

Polly Allen: Oh my gosh. We'll get into that magic in just a sec, but I'd love to see a demo of the device. I remember you saying that normally lidar that has this kind of capability is much larger and kind of unwieldy and like you have to kind of bulk carry it around to construction sites and things like that. Is that right?

Nico Posner: Yeah, there's really um, a spectrum of technologies, cost, fidelity or accuracy and precision. Um, you know, there's the lower end devices that either Use um, what's called photogrammetry, which is using just the cameras, um, and interpolates um, the uh, 3D environment and can create either a mesh or a point cloud out of that. Um, then there's the lidar that are embedded into the iPhone pros and iPad pros right now that can capture again up to about 15ft. Those can all be then processed. And then uh, on the higher end there are you know, tripod based systems or handheld based systems that use other um, lidars, uh, real lidars, um, that use uh, different technologies to scan much wider range. And the tripod based, professional based systems can capture millimeter precision up to you know, hundred hundreds of meters in, in distance. And you know, there are even ones that go on airplanes for even much farther range against specialized tools. But for the AEC, um, industry typically the LIDARs that are on tripods, um, can produce you uh, know, the millimeter level precision um, measurements. But you know, they cost between 20,000 and $150,000 per device. They're big, they're heavy, um, they require training and um, therefore they often are not available to the teams that need them or if a company I spoke to recently only has the budget to afford one of them. So they are limited in terms of the channel of the throughput of that device because they only can afford one. Um, but our device is sitting in between these. So from the cost perspective it's more expensive than an iPhone, but it can give you some of the uh, greater range and technology of a true point cloud that you would get from a professional level device but with centimeter level precision. So you know, what we've uncovered with a lot of customers is that those professional level, very expensive devices are amazing, but they're overkill for a lot of use cases where they don't need that millimeter, uh, level precision. Centimeter level precision is perfectly fine for estimation. Or if you're doing a civil engineering project and you want to scan the pipe before you put the asphalt down on the street, which is now becoming common practice for electrical utilities, et cetera. You don't need to know that down to a millimeter level precision, centimeter level is perfectly fine. So you know where to dig and where not to dig in the future. So there's this sweet spot in that spectrum of um, accuracy that's at a lower level of precision than these very expensive high end systems. For those use cases where that just isn't needed. And then where you do need those higher level precisions, then you bring in that expensive Equipment to get that millimeter level precision. So these devices often can work in concert, they can work together and the data sets can be combined on the back end to combine the data sets. But in the meantime we provide a device that is you know, only a few thousand dollars. Uh, it has the mobility, you know, you can attach it to your iPad, you know, or your iPhone, you know, just snap it right on and there you go. You can have each of your field teams out there having these devices collecting the data with the precision that's needed for those use cases. And then you reserve that specialized expensive equipment for where you need that higher level of precision and where the fewer devices may be available. So it's really filling a gap in that spectrum of where you want something that's more robust than just an iPhone. M But you really don't need that $100,000 device that may be overkill for many use cases. So it plays as part of a larger family of tools that companies will need. But uh, we think it's going to be very, going to be very exciting for many companies who are going to be interested in this tool that fill that middle gap. Something that's very affordable. Provide uh, point cloud data sets with centimeter level precision, um, and then they can reserve, you know, work on their toolkits, um, out um, around how they deploy all their tools and resources most effectively to drive, to drive business ultimately.

Polly Allen: Oh fantastic. You've been involved very, very early days since Q3D, right? So you were involved in like initial decisions about go to market, like how did you think about customer segmentation and like hey, what markers do we go through first?

Nico Posner: Well I think uh, we've been thinking about where do these technologies exist today? And again the LIDAR technologies with these more expensive and high precision devices have existed for a long time and Apple uh, has done a great job on the lower end, more the consumer level. But um, our founding team really realized that this um, there was this gap, this middle gap and that we could repurpose some of these technologies that have been in market for a long time and look for a new form factor, um, and bring out device to market which will solve that need and strike that intermediate balance. We're not trying to compete or replace these millimeter level tripod based systems. Those are very valuable and important for specific use cases. But there was this pain point of companies saying I don't either have the budget for those or I don't have enough of those or the training. You know, I'm worried about sending that into the field or they, you know, it won't fit above the ceiling, it won't go down into a manhole or I don't want to send it into the mine, you know, because it's a really expensive piece of equipment. What happens if it falls and breaks etc. That's expensive piece of equipment. So there is this opportunity that, that the team saw around doing something that will bring that professional level of quality data and provide that more um, affordable price point. We think it can really democratize access to this lidar technology to smaller firms who don't have million dollar budgets but maybe smaller medium sized firms who would love to have this technology and have access to it and really again use the teams and tools that they have more effectively and have a cost effective solution where every crew can have one of these devices out there at uh, you know, without breaking their budget and, and really drive business and create value for them.

Polly Allen: Totally. Well it sounds like finding that right slice right of the exact market you wanted to, to target has been a big part is just as big as like the AI itself. Like to what degree is, is there, you know, there is AI involved? We were talking about this before. It's not necessarily the new flavor of gen AI. This is more traditional AI like you'd worked on in the, in the past for years as well, right?

Nico Posner: Yeah. So uh, there's no gen AI in this particular area but there is a lot of data processing around, um, data quality and taking the data sets and um, we use what's called, it's called a SLAM scanner, um, uh, which stitches together different frames. Um, and to do that you'd need a lot of uh, processing and math to do so intelligently and know what to discard, what to keep, um, to make sure your data sets don't blow up. Um, because you know these data, you know these, this Device sends out 130,000 laser points per second. You know that and measures that. That's a lot that can produce a large, very large data set very, very quickly. And so part of the software magic and the, and the intelligence is what to keep and what to discard to create that accurate map and what's um, you know, and create those file sizes that are manipulable that you can then um, you know, process in the cloud or in our case process it locally on the device which has a lot of key advantages um, than having to upload to the cloud which may have low connectivity or and wait uh, for a long period of time. So there's different approaches to Both data capture, data creation, the synthesis of the data is key point we take. One of our key differentiator value props is that we are doing um, processing at the edge so on the local device so you can validate have you captured everything that you wanted to capture before you leave the location. One of the key pain points that we've heard from customers is they go out and do a scan, they take it back to the office, they wait hours for the cloud processing to go through and then they realize they missed a corner, they missed an important area and they don't realize it until later and that they have to go then back. And there's a lot of cost and time uh, and effort to fill in those gaps. With our device we focus very much on that edge processing in the software. Um, you can actually do a basic rendering, a 3D rendering ah immediately on device. You can even then append to those data sets to fill in those gaps before you even finish uh the scan and before you finish uh the project. So you know with confidence that you, before you leave that you've captured everything you need to. So thinking again end to end about the customer workflows is what are some of those pain points is make it easy, affordable um you know to have one of these, have a device that is easy to use, it's plug and play and then enable the user to visualize what have they scanned to make sure they've actually completed the job that they were asked to do on site, immediately live on device before you leave the site and then with that data compression um, to then allow additional post processing in all the standard formats or um into all the other uh potential um workflows that come thereafter. So there's a lot of magic and choices that we've made around making sure we're solving for a lot of really uh key pain points for our customers. And um, there's a lot of excitement around this product launch uh, that we announced about a month ago.

Polly Allen: Oh that's so exciting. But I love seeing is that that focus on the customer and that like their pain points with their current solutions is really what's driving innovation. And it speaks to the larger question of just like you know, you've been managing innovation now for decades, generally at the forefront of this is that is the management of innovation changing is that is with these new changes with AI like how is this different things you've seen before?

Nico Posner: Yes. So I think uh, the core innovation philosophies are the same which is really understand customer problems deeply and then think Creatively and the full set of potential, um, solutions given the potential technologies that are available at that time. And so, uh, that journey can change over time. You know, the technologies that were not available two years ago or two months ago may be available today. And so part of the technology innovation strategy and leadership challenges, both staying up to breast of what is currently available today that wasn't available previously. And given the tools at our disposal, um, what is the best solutions that we could solve. And then of course mapping that to cost, quality, speed, you know, monetization, you know, opportunities, um, customer fit and appetite and demand for those solutions, you still need to do those mappings. Um, but with the evolution of AI evolving a lot of these tools so quickly, um, there are new tools that are available every week it seems. And so one of the challenges staying up to date on that, um, to make sure that you can both adopt and recognize both technology shifts that you want to take advantage of as early as possible when they are a good match for your product, customer segment and solution that you're building. The second piece is how what I see and I hear this every, you know, across the board is how are teams and individuals using AI to accelerate their own uh, m work? So across the board, so in uh, innovation development, there's a lot of work you can do with um, AI, you know, cloud code or others to uh, create synthetic uh, users and test use cases and do research and find, uh, define user segments or go find me the top thousand, uh, firms in this market who might be interested in this, help me organize um, my thinking around how are we mapping our strategic opportunities and product fit against the pain points that are out there, um, in the data sets that can be retrieved and uh, analyzed through all these tools. That process of strategy, um, uh, evolution and thinking about um, what is possible are greatly accelerated with the tools that are out there today and processing through this. Ultimately, you know, you as the human and the leader and the team, you still need to discuss which of these are the ones that we believe are correct and go validate that further. Um, and which ones do we need to dig into more and which ones, you know, are not applicable. Um, so it's not like AI can do all of this, but AI can accelerate a lot of that, um, grunt work and the research work that previously would have taken days or weeks of research and might not have been exhaustive or uncovered all kinds of data that AI tools are very, very good at surfacing data, uh, from the much larger corpus of underlying data that they have available to them.

Polly Allen: Absolutely, absolutely. So it sounds like even though your product doesn't necessarily use Genai, you've been using it a lot in your own business and planning. As a thought partner, what are your favorite tools to use when it comes to Genai and the latest wave?

Nico Posner: Yes. So internally we are a very small team and so not only um, do we use the tools, but we encourage and really talk about which tools we're using internally, which ones we find effective and we share best practices internally because we want to stay small and nimble. We want to leverage the tools that are available to us um, to their maximum potential. And so we spend a lot of time figuring out how can we use these tools and which tools have we tried and which ones work well to increase that leverage of improving our operations or improving our insights or accelerating our response time to customer inquiries and classifying, sentiment analysis and lots and lots of um, AI tools again to accelerate some of that processing and work and analysis that we can spend more time thinking about what is the results of that analysis and reacting to that uh, and doing less of the actual um, data crunching to get to that analysis to begin with. So we spend a lot of time using those tools um, and we definitely are very excited to be at the forefront of a lot of AI tools that are evolving uh, uh, in the technology uh, world that are leveraging AI um for SaaS specific disrupting new industries. Again with a lot of this 3D reality capture. There's many, many companies out there who are doing really interesting things but a lot of them, all of them really rely on some tool to go out there and capture the reality of what is to then do that magic on top of it. Whether it's incorporating a scan of your warehouse into your operations software to work out an improved workflow for your humans and or warehouse robots. You know there's all uh, you need to start with what is not what was the plan, architectural plan that when we built it, you know, two years ago or 10 years ago, but what is actually there today. The real world environment changes all the time. So there's a, there's a real need to both create. You uh, know you can look at digitization of architectural plans and build plans but then you can need to really get an up to date version of what is. And then there, what we're seeing is there's again ah, a variety of tools. You could take one of these laser scanners to scan a huge um, you know, warehouse environment that can capture that, that distance, that range, that act with a great Accuracy and then you could go through with one of our devices, maybe on a weekly or monthly basis to capture what has changed since that last deeper scan and merge those data sets. So, oh, we moved this pallet, set of pallets over here or this rack, uh, was removed or fell over and therefore we need to reflect that in our, in our data sets. That again is where then we could come in and say here's a really great tool for you to go and do those interim scans again so that you will have an accurate and up to date reflection of your reality. And then you can do all that magic of understanding what to do with that data in the many, many, many software providers that are existing today and many, many that I see that are emerging to help capture and utilize that data, the real world data, and do magic with that, with all the wonderful, powerful things that, you know, software and compute and smart engineers and business people can come up with on solving customer needs better using all the tools in the toolkit, which now includes AI, it includes ML, it includes, you know, 3D, you know, reality, uh, capture lidar devices, it includes all cloud, compute, you know, edge computing, all of that, um, to you know, really create some really, uh, incredible solutions into the real world that are going to meet real customer needs and create real businesses. And it's going to be a lot of opportunities to monetize those because they create real value for those businesses.

Polly Allen: Absolutely. It feels like the whole era is just, is just beginning. But that this, that foundational layer where you can't make AI real unless you're working with real, valid, up to date data. Right. That's just been the challenge in so many different domains when it hits the real world. Well, I want to thank you so much for joining me today, Nico. This was such an exciting deep dive. So exciting to see um, what you've been up to since the blueprint program and really uh, exciting to understand a little more too about um, 3D reality sensing and that world. So thanks so much again for joining us.

Nico Posner: Thank you Paula, for the opportunity to chat. It's been great talking to you.

Polly Allen: Of course, that's our episode for the AI product leader this week and uh, as always we'll be having our upcoming masterclass. So be sure to uh, join us soon for that masterclass on becoming an indispensable AI product leader. Um, coming in, coming up soon, the link will be in our, in our show notes. We'll see you in a few weeks. Thanks so much.