Trusted Identity AI with Doug Aley

In this week's episode, I speak with Doug Aley, Chief Executive Officer of Paravision.

Paravision’s products are leveraged in digital identity verification solutions as well as government, travel, and border control programs. They also have implementations supporting stadiums, events, automotive, payments, and retail use cases.

Doug shares his personal origin story from co-founding accessibility company SSB Bart Group (now Level Access) while still in undergrad at Stanford University to his time as a category manager at Amazon.

We discuss artificial intelligence, deepfakes, National Institute of Standards and Technology (NIST) benchmarks, and what it means for Paravision's technology to be "ethically trained and conscientiously sold."

RESOURCES:

Connecting with Doug Aley

Doug Aley’s LinkedIn: https://www.linkedin.com/in/dougaley/

Paravision’s website: https://www.paravision.ai/

Companies & Resources Discussed

Paravision is a leader in trusted Vision AI, with globally deployed AI software that is ethically developed, conscientiously sold and built for the most challenging applications.

Kamil Chaudhary is the General Counsel and SVP, Data Acquisition at Paravision

Joey Pritikin is the Chief Product Officer at Paravision.

Jott Networks was a Seattle based start-up that converted voice to text technology. It was acquired by Nuance Communications in 2009

Zulily was a woman’s fashion e-commerce site that recently ceased operations earlier 2024. It was acquired in March 2024 by Beyond Inc.

Level Access (formerly SSB Bart Group) helps its customers achieve and maintain compliance with the full scope of accessible technology regulations and standards. The company’s solution ensures customers’ websites, desktop and mobile applications, embedded software, gaming software, digital products, and electronic documents are accessible to everyone.

Tim Springer is the CEO of Level Access and was a co-founder of SSB Bart Group.

Kevin Goff was a co-founder of SSB Bart Group

Ethan Kurzweil is a partner at Bessemer Venture Partners and was a co-founder of SSB Bart Group.

Atomic is a venture capital firm. It brings ideas, capital, resources, and talent together by partnering with co-founders to build the best ideas into great companies. Jack Abraham is Atomic’s managing partner and CEO

Andrew Dudam is the founder and CEO of Hims and Hers. He was previously the co-founder and general partner at Atomic.

Shiva Corporation was a company that specialized in computer networking and associated equipment, in particular remote access products. Founded in 1985, Shiva was acquired by Intel in 1998 and became part of the Intel Network Products Division.

Avatar Technologies, previously named 3R Computers, was an American computer company based in Westborough, Massachusetts, known for their Avatar series of dumb terminal-to-workstation devices. In December 1992, the company was acquired by Digital Communications Associates (DCA).

Digital Equipment Company was a major American company in the computer industry from the 1960s to the 1990s. DEC was acquired in June 1998 by Compaq in what was at that time the largest merger in the history of the computer industry. Compaq subsequently became part of Hewlett Packard in 2002.

JMI Equity is a growth equity firm focused on investing in leading software companies.

Andy Jassy is the President and CEO at Amazon. 

Jason Kilar is the former WarnerMedia (now Warner Bros. Discovery) CEO and founding CEO of Hulu.

Docker is a leader in the container platform market. It provides the only independent container platform that enables a seamless desktop to cloud experience for developing and scaling distributed applications.

Geoffrey Hinton is a British-Canadian computer scientist and cognitive psychologist, most noted for his work on artificial neural networks.

NIST (National Institute of Standards and Technology) is part of the US Department of Commerce. Its mission is to promote US innovation and industrial competitiveness by advancing measurement science, standards, and technology in ways that enhance economic security and improve our quality of life.

iBeta is a full-service software QA and testing company dedicated to providing industry-leading service to its clients.

Open AI is an AI research and deployment company. Its mission is to ensure that artificial general intelligence benefits all of humanity. Its primary product is ChatGPT.

FULL EPISODE TRANSCRIPT

Steve Craig: Welcome to the PEAK IDV EXECUTIVE SERIES video podcast, where I speak with executives, leaders, founders, and changemakers in the digital identity space. I'm your host, Steve Craig, Founder and Chief Enablement Officer for PEAK IDV. For our audience, this is a video first series, so if you're enjoying the audio version, please check out the full video recording on executiveseries.peakidv.com, where you can watch the full episode, read the transcript, and access any of the resources or links that we discussed in today's conversation. This week, I have the privilege of speaking with Doug Aley, Chief Executive Officer of Paravision. Based in the United States, Paravision builds industry leading AI building blocks that are easy to deploy and run on every major platform, from the cloud to the edge.

Paravision's products are used in digital identity verification solutions, as well as government, travel, and border control programs. They also have implementations supporting stadiums, events, automotive payments, and retail use cases. Doug joined Paravision in 2016 and was appointed CEO in 2018. His 25-plus year career includes founding, leading, and scaling technology ventures.

In addition to being a serial entrepreneur, his early career included time at Amazon in product management, marketing, and business development at Jot Networks. And he helped Zulily scale from 100 million to 700 million in sales as VP of product and corporate development. At age 19, Doug had co-founded and was CEO of his first company, Level Access, which was formerly known as SSB Bar Group, a company dedicated to advancing web accessibility. Welcome, Doug. Thank you for making the time to be on the podcast. 

Doug Aley: Thanks for having me, Steve. Really appreciate it. 

Steve: Let's get started. Can you share more about Paravision? What's your typical elevator pitch when you're meeting new people and sharing what you do? 

Doug: Yeah, I feel like you actually just gave the elevator pitch so well. But yeah, we're a company that, you know, tries to and thinks we, you know, at least at this point, lead the world in providing AI building blocks for identity verification and authentication and can probably best be known for face recognition and related technologies like liveness detection, deep fake detection, age estimation.

We take a sort of an “Intel inside” type approach to our business model. Kind of letting our partners lead the way from an actual end solution provider perspective. So we're laser focused on research and making sure that all of the AI models that we create are fair and accurate. And, quite frankly, like super flexible from a deployment perspective.

And so we think of ourselves kind of in that vein, our view is that sort of face recognition as a passive biometric or a relatively passive biometric is applicable to kind of the entire world of authentication, whether digital or physical. And so you'll find our technologies and, you know, in everything from identity verification, you know, which you're very familiar with, all the way through to kind of stadiums and events, border control, even cars.

Steve: Can you share more about your team? I understand HQ for you is in San Francisco. Where are your employees based? 

Doug: Yeah. So we have a headquarters in San Francisco. I would say probably about a third of our employees are in the Bay area. But we also have a pretty substantial presence in Canada, in the sort of Toronto metropolitan area, as well as in Vancouver. And then some folks on the east coast in the United States.

And then our head of public sector and international is in the UK. And it's actually a pretty amazing group of people. I would say-- I think now the average tenure-- we've only been around for six years and the average tenure on the team is five plus years at this point.

And so we've kept the team, I think, purposefully pretty small and nimble and focused on research, but a pretty special group of people. At this point, we-- you know, we do describe ourselves as a team and not a family and yeah, primarily so that we can use sports metaphors like throwing no-look passes to each other. And that's the way that operates every day. Like, you know, work with a lot of teams and I'm incredibly blessed to say that this is the one that has just gelled the best for the longest.

Steve: You mentioned that you support global use cases, are there particular geographic markets that you have more implementations or more customers, partners?

Doug: Yeah, I mean I would say, you know, as a US company, I would say a lot of the implementations that we have are with large US multinationals, so we tend to serve a lot of rather large companies. But outside of the US, a ton of implementations in Europe, especially on the travel and border side.

And APAC, increasingly, we've had customers in Japan for years but now have customers in, you know, in Australia, Singapore, and elsewhere in APAC, in India. And I would say sort of increasingly, we're starting to see a groundswell in Latin America and in the Middle East, you know, so we serve, you know, with a relatively small team, we serve a global group of partners.

I would still say, you know, about 70 percent of our business is outside of the US from an actual deployment perspective. But probably more like 50 or 60 percent from a companies and partners we support tend to be headquartered in either the US or Europe. 

Steve: That's good. That's good background. I guess when you have edge and cloud services, it gives you a lot of flexibility where you go in the world because when the edge use cases that you're beholden to infrastructure or ‘co-los’ and things like that. I'm curious, Doug, how did you get connected to Paravision? You are not the founder of the company, but you're CRO-- you're now CEO. How did you come to join? 

Doug: Yeah, I would say all of us early folks at Paravision proper-- and I can get into this-- kind of feel like founders, but not technically founders. So a guy-- two gentlemen named Jack Abraham and Andrew Dudum, who were founders at Atomic, which is a venture studio, actually founded the corporate entity way back in 2013. And then about the time I arrived at Atomic as an operating partner and one of the companies that I was tasked to work on as a chief revenue officer amongst a bunch of other companies, was this sort of the predecessor company to Paravision.

In about probably a year and a half in this sort of the board and the management team kind of figured that that company, which is a photo and video storage business wasn't going to reach venture scale. And they asked me to to lead a pivot as a CEO of the new company, you know, Jack is still the sort of the managing director at Atomic. Andrew's gone on, he went on to found Him's and Her’s, which it's a sort of large kind of telehealth company a public company, very successful. And we still sort of have a great touch with both-- Atomic is still heavily involved. 

Steve: Before we go deeper into Paravision, this is EXECUTIVE SERIES, So I like to go into origin stories of individuals. And I went down deep into your LinkedIn profile. I see you're a Stanford undergrad, Harvard MBA. That's like a super impressive combo. I'm curious, are you from California originally? Like what's your-- what's your background? 

Doug: No actually, you know, my delicate pallor, I'm surprised I'm still here in California. I'm too pale to live here. No, I grew up in a small town in Boothbay Harbor, Maine. And kind of between Maine and New Hampshire, my dad lived in New Hampshire and that's where we kind of started and then moved over to Maine after my parents separated and kind of did junior high through high school in BoothBay. It's pretty beautiful place to grow up. 

And then my brother, one of my brothers, I have seven siblings between the sort of the four that are direct siblings and then three that are step siblings-- one of my brothers moved to the West Coast kind of the early dot com, actually even pre dot com days. So, like-- back-- like, kind of a Shiva and Avatar and Digital Equipment Company type days. And he moved to the West Coast to be an entrepreneur. His twin stayed on the East Coast, also an entrepreneur. And now they're-- they've co-founded a company together for the second or third time. And then my sister Kara, who is a division one athlete. At one point, I think, ranked 21st in the world or sorry, 21st in the nation in the kind of three mile cruiser cross country-- got a full ride to Stanford. She, back in the day, kind of pre digital she would send me photographs of the campus, and I'm like, that seems pretty nice. So I applied kind of sight unseen and early decision. And I-- you know, I like to say geographical distribution got me in. I don't think there were that many kids applying from Maine. So yeah, very fortunate to end up there because that kind of jump started my career. You know, kind of from that point forth, the people that I met and some of the experiences that I had, and actually quite frankly, just being in, like in the heart of Silicon Valley.

Steve: It's a great back story. And once you get to California, sometimes it is hard to get to other places, it's very, very beautiful. I haven't been to the area you speak of in Maine, but that's a great back story. But then I see kind of reading through the background you have in some of the bios out on the internet that you're at Stanford.

You must have been like a sophomore or a junior, you decide to drop out and you start SSB Bart Group with Tim Springer. Like, can you share more about deciding to-- maybe you got the dot com fever. You're like, “Hey, I'm going to start a company and follow in Bill Gates shoes and drop out.” How did all that come together?

Doug: Yeah, I wish it was quite the same sort of like end point that Bill and Mark got to. But yeah, you know, it's kind of like, you know, the thing to do at the time. We had several friends that had stopped out of school, they call it ‘stopping out’ at Stanford because, you know, they do want to, you know, get you back and get your money, but in the best way.

But it's really super flexible program and I had told my parents, I'm like, “listen, I'm going to, I'm going to stop out now. I promise I'll get my degree.” And, you know, I think at the time it was me, Tim Springer, Kevin Goff, who went on to Google and actually incidentally Okta, sort of very relevant to our sort of conversation here. He ran I think he was director of product marketing and Okta. And then a guy named Ethan Kurzweil who went to Bessemer Venture Partners and all of us are still great friends to this day. Actually, Kevin just moved like basically two miles from me up here and Tim's down in Southern California now.

So yeah, so like great group of friends that kind of started that. And we-- Tim and I were the first to-- I think I was the first to drop out at the kind of middle of our junior year. And then Tim dropped out within like, you know, by the-- at the end of that year. And then I think Ethan kind of joined us for kind of six months and then decided, “hey, this isn't for me. I'm going to go back to school.” But you know, in the meantime, I think closed the deal with Adobe in those first six months. And then Kevin stayed with us for-- I think actually Kevin even outlasted me there, I think he was there for close to four years before he went to Google. 

Steve: For those that might not know SSB Bart Group are now called Level Access. So what was the problem space that you were focused on? I mentioned it was web accessibility, but what was-- how was your-- what was your business doing at that time when you dropped out? 

Doug: Our goal there-- and it's actually kind of-- there's some interesting parallels with where we are today with Paravision was to make the web and software more accessible to differently abled people. And we, Tim and Kevin and I and Ethan kind of early days all, you know, had friends that had had difficulty accessing the internet. We thought we could do something about it. There were some open source tools at the time, but we felt that through some of our early work that there was an opportunity. We actually started-- as actually consulting with a bunch of groups. That company started as a tool that was trying to provide e-commerce and other kind of community services to people with disabilities. And then what we found is through the sort of process of partnering with different people is that there was an opportunity to really help have a major impact in terms of making websites accessible because most of them weren't at the time, especially to screen reading software.

But so, yeah, you know, Tim-- I lasted, I think, probably three and a half years before burning myself out in that role, kind of off and on. And I went back to school after I think about three years, we had a couple of different CEOs that were not great fits, I'll leave it there, before Tim asked to, and it was only sort of myself and Tim on the board for gosh, like 16 or 17 years before Tim said, “hey, you mind if I take the reins as CEO?” And, you know, he's now built it into kind of a juggernaut, definitely the leader in the industry. We sold collectively to a private equity group called JMI back in I think it's like kind of 2017ish. And I dropped off the board at that time, but I think Tim's grown the business four or 5X since then and just had really-- just a great example of having a double bottom line.

And you know, we use that in how we sort of think about-- we look at sort of face recognition and a lot of the bad name that it got early on for not treating all people equally. It's like in the early days of voice recognition, if you were a white male, things worked out just fine for you. But in the early days of face recognition, if you weren't a white male, the technology was really biased. And, but for what it's worth, like a lot of vendors still maintain that bias today. We've spent a tremendous amount of time and focus on making sure that the technology is as fair as it can possibly be, and we'll never really be done there because we can always improve. 

Steve: It's a really good point. When people think about bias and algorithms, they think about how the data is trained, but even for SSB Bart Group there's bias in how you build user interfaces and how you build web experiences. And when I was researching your background for this episode, I made the connection to SSB Bart Group, and I recalled that in 2015, I used to service. I was working at one of my first identity verification jobs, and we used it for our software development kit to improve accessibility. It was a vision oriented experience. So the talk to text feature and supporting those that might have tremors or other physical ailments that made it difficult to use, so that's very cool. 

So you ended up going back to school, you graduate. At some point you find a very little known company that was just selling books online. You take another career gamble and you join Amazon. I'd love to just touch there briefly before you jump back into Paravision about joining that company and what you're working on and how it might've influenced how you lead as an executive.

Doug: Yeah. So Amazon was kind of-- it was interesting and I don't know how much of a gamble it was. It was 2003, so it was past the Amazon dot bomb era. And but what I knew of the company I knew-- I basically said like, listen, the next step that I want to take, I want to learn how to do this, right? We made so many mistakes. I mean, it's like 19 to 22 you make all the mistakes and they're great mistakes. They-- they're great-- kind of lead you to so many awesome stories from that period of time that are sort of like deeply embedded in my DNA now. But you know, Amazon taught me how important it was to have a very, very sort of clearly defined culture.

And you can kind of say what you will. And lots of people have opined about Amazon's culture, especially in that kind of like, early days through the last kind of five or 10 years. But one thing that I think I learned, I think one thing that Charlie our CTO takes, you know, the things that we take with us are that it's super important to have clearly defined culture and to build your team around that culture, around what your kind of core beliefs are. And then when you do that, you can create magic and they've, they've certainly captured lightning in a bottle. 

And I was super fortunate to have some awesome mentors early on back in that day, I did. I was there for six months prior to going to business school. And then basically worked all the way through business school while at business school to kind of fund my way.

At the time, they sort of paired you up with different mentors. If you were a business school, they did the same thing for the computer science folks that were going through their master's programs there. And I got paired up with Andy Jassy and Jason Kylar. Those are my two mentors.

And obviously now Andy's CEO and Jason went on to be CEO of Hulu and Warner Brothers and a bunch of other stuff. So I was-- you know, kind of really-- kind of a fun time and really fortunate to get to bend their ear as a, you know, as a young guy who didn't know anything and they were in the midst of kind of figuring it all out, you know, for Amazon. 

And I worked on the early kind of merchant platform there debugging flat files for the sort of early kind of third-party merchant, third-party seller business. Went on to launch what is now Prime Video, but at the time was Amazon Video Downloads as the lead PM on that and worked for a bunch of great leaders there.

Then my mom passed away, needed a little catharsis primarily because I, you know, wasn't smart enough to deal with grief in the right way at the time. So I just buried myself in work and moved over to help run their sporting goods business, which was super fun. Incidentally met the guy that invented the perfect pushup, who is now a parishioner at my church and so I see him all the time. He was one of my first merchants and so helped them launch fulfillment by Amazon there. Actually Perfect Push Up was one of the first products on that. And so, yeah, just super fun. A lot of hard lessons learned, but the biggest one was definitely, yeah, sort of clarity of culture. And hiring just absolutely the best people. Jeff Bezos used to say, he's like, “A players hire A players, B players hire C players.” And so if you can keep it to A players then you have something really-- the potential to have something really special. 

Steve: Can you share as we move into what Paravision does today, what the core pillars of your stack are? And then you have this concept of identity AI. Can you share more about that? 

Doug: Yeah, sure. So we think of identity AI in the long term is as the-- in our job in that ecosystem is really to provide the sort of the core AI building blocks for that. So today, that's the core pillars are really face recognition, liveness detection, deepfake detection, and age estimation.

And those are the things that we provide to our customers via a super flexible network of APIs and SDKs. We don't host any of that. We provide those in a you know, it's sort of typical implementation as either via, but kind of like either a mobile SDK or an SDK that you want to sit on on your servers or via Docker containers, probably the most popular way of deploying that. And you know, our-- you know, our focus in that ecosystem is making sure that those models are ethically built, hyper accurate, and very fair and easy to deploy. And I think we've done that. You know, we've we now--, you know, we're up over a hundred major customers at this point.

And we've really been able to kind of choose our customers, which is equally important to us is to make sure that our customers and our partners are similarly aligned on the sort of ethical deployment 

Steve: And looking through Paravision’s press releases and your posts on LinkedIn. I understand you work with identity verification providers as well. Can you share what your approach is to partnering and integration and just how you operate within the ecosystem of these identity providers? 

Doug: Yeah, sure. So you kind of trace this back to one of our core values, which is thrive as a team. And we consider thriving as a team that our partners are part of that team.

And so we take our partnerships very seriously. They don't stop with the deployment or just handing over the software to folks who make sure that it gets deployed in a responsible manner. And you know, ultimately our goal for them is to be able to provide a seamless and secure experience for their customers.

We think face recognition is kind of uniquely positioned to provide that in the world now that you've had this sort of generational kind of shift in compute and the availability of compute, which not only leads to better models, but leads to faster user experience. 

We focus on long term partnerships. Our typical deal is you know, kind of three to five years long. And I'm super fortunate to work with a bunch of amazing partners in IDV. 

Steve: When it comes to digital identity and identity verification, where are you seeing the most growth and adoption around face and liveness? Is it regulated KYC, or are you seeing more non-regulated scenarios?

Doug: I think it's really-- especially in the-- I would say it started with highly regulated stuff. I would say financial institutions, banking as well as sort of major government programs. You know, electronic travel or authorization programs, stay tuned for some pretty major announcements there.

But what I would say is it's definitely migrated and our partners kind of lead the way there. It's migrated to you know, everything from kind of gig economy, onboarding and enrolling people in the gig economy, to online gaming other kind of either age or identity restricted areas.

So I think what we've been… a health identity actually is a huge one that's coming on board, payments. It's really interesting to see that the kind of real shift in the last year and a half from more kind of places where facial recognition was more traditionally accepted and there's been this kind of now that all the building blocks are in place, you kind of have this opportunity for sort of a an explosion of of new use cases.

And I think that we're at that place in the same way that AI broadly has, like, now that all the compute and infrastructure is in place now that sort of people are able to take advantage of you know, sort of building on the backs of, you know, kind of Jeff Hinton's early papers and deep learning 

Steve: When I meet practitioners or buyers of these technologies, one of the first things they start to focus on is match rates on the verification process, face recognition itself. But one of the pillars you mentioned, the liveness detection, deepfake detection, really ensuring that the person who's in front of the camera is a real person, you know, beyond just do they match. But, is it a person? Is it a presentation attack? Is it a spoof? In the market, there are active techniques; there are passive techniques. I'm curious for your stack, where does Paravision fit in that? Like, how do you tackle the liveness problem? 

Doug: Yeah, I mean, we've always had the perspective of-- well, we've had a few different perspectives that kind of drive our development.

One is that great user experience will always win the day. And that really means getting out of the user's way. So for us, our focus on liveness and deepfake detection is single frame, either video or photo, passive liveness detection. We think ultimately that will win the day versus more active models, which are sort of equally successful at this point where you're either blinking lights at somebody or you're asking them to move their face in a particular way or say something or those types of challenge based approaches which again, they have their place. But we've gotten to the point where the technology from a single frame perspective is, we believe, equally as good.

And we think the promise of it is much better from a user experience perspective that, you know, we're all kind of product folks by trade. And so Joey Pritikin, our Chief Product Officer leads the way there and kind of determining what we're going to do. 

Steve: Do you think that when people are implementing these technologies, they often take the path of least resistance of like, “hey, let's just ask the user to upload a picture from their camera roll” versus going through like an SDK experience. So do you, do you feel like from what you've seen in the market that that happens frequently?

Doug: No, I think now-- I think people are taking it super seriously Steve. Like, I mean, and they should I think you have this-- we're not quite at the point, I think the-- kind of there's a bunch of people that are with the sort of the ease of building deepfakes. Now, we're like, “oh, there was a 3,000 percent increase in deepfakes year over year.” It's like, “yeah, that's because there weren't that many deepfakes last year.” The tools weren't quite there, yeah, kind of ready yet. That being said, I don't think the fear is right because the technology is improving so quickly that that making sure that you're putting robust sort of barriers in place, the right amount of friction and the right barriers in place to people to kind of dissuade them from doing that.

The thing we know from history, right, is that hackers are going to hack. And they're going to try to find a way in, always. And, you know, companies like us, you know, our job is to make sure that that doesn't happen and our customers demand it and they're being, for the most part, really smart about it. If they're not, they're not our customers. 

Steve: I want to shift a little bit into accuracy and earlier in our conversation, you talked about the NIST organization. I feel like in the field of facial biometrics, we're pretty fortunate that we have the National Institute of Standards and Technology in the US that has benchmarks; there's independent testing. What do you think are some of the key benefits of just having this within the industry? Just having an industry benchmark? 

Doug: God, don't we wish that NIST was there to be a benchmark across artificial intelligence, right? You know, I think the reason NIST exists not only for consumer protection, but also to serve its customers in the federal government.

And so. You know, kind of post 9/11 where face recognition and other biometrics were coming onto the scene. They started the process of putting kind of benchmarks in place for all biometrics, iris, fingerprint, et cetera. And, face recognition has definitely been the beneficiary of, as an industry, those public benchmarks. And so they're just invaluable. The team, they're so incredibly smart, they're-- I would say the smartest thing that they do is being completely objective, no favoritism by country or by company. And they produce these, you know, beautiful 300 page reports that not only, you know, act as a kind of a way to publicly benchmark for customers that are trying to choose a vendor, but really they're an invaluable resource for us as a, you know, development team.

We pour over those reports to find where we're-- where there are opportunities to improve and we've basically just said you know, that NIST is, and the accuracy that it helps us enable, is kind of our north star. 

Steve: Are there any areas of independent testing where you feel like we're still falling short whether it be in efficacy or performance, inclusivity, any specifics where we've still got some work to do?

Doug: Yeah. I mean, the sort of PAD-- so we still don't have a deepfake benchmark, that's again, objective third party deepfake benchmark. We have PAD benchmarks both at NIST and iBeta and other places. We think those are super valuable. Always, you know, our focus internally is always on across the board demographic performance.

We recently launched a white paper kind of showcasing the results of some of that focus. And so I would say, like, the places where I would love to see continued improvement and investment is number one, always in demographic performance. Demographic performance, by the way, often gets shorthanded to race, but it's really the intersection of, you know, age, skin tone, gender, you know, those tuples across a range of different environmental factors and capture factors. And so when we think about demographic performance, we think about it across that entire spectrum and have an internal philosophy. “If it doesn't work for everyone, it doesn't work for anyone.” And that keeps our team going.

I would also say that, like you know, people-- the way that NIST is used as a marketing vehicle sometimes it can be a little disconcerting when people aren't providing all of the data for people to make independent decisions. But they just focus on the kind of like one area where they did well in and then there's some kind of like subtext there.

So I think anytime you enable something like that, you've got the people that are going to try to take advantage of it in different ways. And I think NIST has properly stayed out of the fray there. I do think that there probably should be minimum bars for deployment. I would love to see NIST or some other organization say, “hey, listen, this is the bar you have to hit anything below that you really don't apply.” That I think would push the industry to have more people kind of performing at the top before you see the kind of dramatic drop off in performance. 

Steve: You've mentioned a few times AI and artificial intelligence and the work we have to do there. I want to segue a little bit deeper into the technicals. And when I think about this space, computer vision techniques, especially the classical ones, where they're like geographic matching or geometric matching, those have been around for decades, right? And some of those early algorithms were why we had a bad rap sometimes on face verification or face recognition, because they weren't very accurate.

But we've seen explosive advancements and machine learning and data set availability. There's potential for all this synthetic data creation. But then just in the last few years, there's been this tech industry land grab on data. There's been a lot of press about Open AI and what they may have used in YouTube or on the web that they shouldn't have.

And one of the really cool things in looking at your website that I found is you have a statement around AI principles and I'm going to read one. “We believe in AI that is both ethically trained and consciously sold.” Can you share more about this particular set of principles and how they influence the company culture that you've referenced earlier.

Doug: Yeah. I mean, they're everything to us. So we first of all, unlike-- you know actually I want to sort of mention sort of other folks that have similar principles. We make our employees sign those principles every year. It's a sort of continued commitment to sort of stepping into the fray.

We've operationalized those principles across the board. So when you think about ethical training first and foremost, you want to make sure that the technology that you're creating is fair. So making sure that it works for the entirety of the world. But you also want to make sure that the way that the data that you collect in order to train your models is fair and with proper consents in place.

And I think we're one of the few companies globally, we're not the only one, that really believes in ethical data collection. And we do that by, you know, we have a bunch of partners globally, photographers and videographers that we send scripts. They go and recruit people to come in and take their photos, they're full biometric consent there. People understand what they're signing up for. That they're signing up to improve on face recognition models through training. We don't save any of those identities related to it. So it's effectively like anonymous data, but individuals you know, in our systems.

And so I think that's been really a focus for the last five years. Or you're going to be in a place where, you know, open source data is no longer acceptable to use and people are going to have to sign up and commit to that. And so we think we've set a great foundation in place to be kind of one of the leaders there.

And then we talk about sort of conscientious sale which is equally as important, which is how do we make sure that this technology gets used properly and doesn't get deployed by actors that could use it for discriminatory purposes or to deny access to certain groups of people. And so we heavily monitor and we actually have a weekly use case review where any new use case that sort of comes in that hasn't been previously defined in the use cases, really what is it being used for and where is it being used and by whom? You know, and kind of, you know, kind of for what purpose.

And we sort of walk through that as a team. Our general counsel is also purposefully our SVP of data acquisition, Kamal Chowdhury. So we've separated that from engineering, which provides a nice check and balance. And he has, you know, he and I are the two kind of leaders of that use case review. He has veto power over me and that's the way that it should be. So again, we've-- we sort of focus-- it's all nice to put some stuff on the site, but how do you actually operationalize it so that your customers can count on the you practice what you preach? 

Steve: There was a Netflix show that I watched recently about an adult dating site, which I won't go into details on, but they had a bunch of stuff on their website about security and protection of data. And later was found out, well, that was just website copyright. They weren't doing any of those things. And so it's really important that you have this guiding light. You have this ethos, you have this way of operating, but it needs to permeate through all layers of the organization.

One of the challenges I see for the digital identity space is we're fighting some of the evils of the world. It's not just about opening accounts, it’s combating money laundering, which leads to organized crime and human trafficking and terrorism and things like that. Fraudsters, criminals, they don't have ethics. They use generative media and AI technology for bad purposes. And so us on the good side, trying to create technology that is not biased, that complies with regulation, that gets proper consent is challenging. As an ecosystem, what can be done for us to fight back? Like, how do you think about the bigger picture world problems that you're solving?

Doug: Yeah, I mean, I'd say a couple of things. So first of all, fraudsters and hackers have existed since the dawn of humanity. Like we're not getting away from those people existing, right? And at the same time you have to do everything in your power to keep them out, to not let that be an excuse, right. 

And so from our standpoint you have to do a couple of things. Number one, create really, really robust technology out of the gates that covers all of the things that you know, right? And then the real question is like, okay, what about the things you don't know? The approaches that people aren't going to take. And I would say there's two things there that are really, really helpful as strategies. One is red teaming. So having teams outside your organization that you pay to attack your systems. And I think that's something that frequently people don't spend enough time on.

So it's like, you've got benchmarking, internal benchmarking before you release something, external benchmarking before you actually release it to production you know, sort of, that's, that's what the role that NIST plays. And then you've got red teaming where people are consistently attacking trying to come up with new ways to beat your system and your technologies.

And it's something that I think the industry needs to do a better job at and then the other thing, and probably the most important thing, and I think the thing that we really pride ourselves on is like, listen, at the end of the day, let's say you've got 1,000 different tools like software packages that can create deepfakes at this point. You want to be red teaming with all those different tools because those are likely the tools that the hackers are going to use. But you also want to take it back a layer from a sophistication perspective and actually look at the research that's being used to generate those tools, right? And if you start at the research layer, the great thing about the AI, you know, industry or community is that there's a lot of sharing of research and that research gets ultimately embedded in these tools.

And so if you start at the research layer. And say what's going to be used at that level to attack the system, then you can effectively thwart off-- you don't have to deal with 1,000 things, you're dealing with the, like, you know, a more confined group of kind of papers that you need to look at. 

We do both. But I would say the integration of not only our primary research, but also secondary research that comes out in the community, your ability to sort of ingest that and make it a part of your products as quickly as possible is kind of the order of the day. We're fortunate to also have, in addition to the stuff that we provide you, our partners that are consistently-- you back in my e-commerce days, we used to call it secret shopper, right? You're going around to your sort of competitors and seeing what they're-- it just takes a really curious mind to continue to attack. That's why we have another thing internally. One of our core values is celebrate curiosity and productive failure. And you know, the thing that we try to do over and over again is just not assume that we're the best.

And I think if you start with that-- whatever you say from a marketing perspective is one thing-- but if you have enough humility internally to say, “hey, I don't know everything, I'm going to keep attacking, keep asking questions,” then you're going to end up in a really good spot.

Steve: Well, the concept of a secret shopper, we're in a free market economy. There's competitors in this space. Fraudsters, they don't often think of themselves as competitors. They freely share ways that they've hacked into services or how they've exploited an identity workflow. And so I think it's important that there be that sharing in that community, and it's part of why I do PEAK IDV to help share knowledge.

I want to switch on to the positive side because there's a lot of doom and gloom about deepfakes and these attacks. What are some of the things you're seeing from AI that are creating truly magical experiences? And if you've got like a Paravision spin on it or how you're enabling it, that would be extra interesting.

Doug: Yeah, I got to say, broadly, AI, we use a lot of AI internally as well. It's like the concept of minimum viable product I now think of it as a minimum viable team. How can you create a group of people that you trust inherently and give them the tools to be 10-- every one of them be a 10Xer. And so we use-- experimenting with a lot of artificial intelligence internally just to make ourselves more productive.

And you know, I think we're on the cusp of a real productivity boom that's driven by smart use and calculated use of some of the generative models that exist now. And we're seeing that productivity on our own team. So I can only imagine what it will do for much larger companies and teams.

And obviously the flip side of that is, okay, that ultimately means that companies aren't necessarily growing headcount as quickly and-- or they're shedding headcount. And that has a real, like-- we need to do a lot more thinking about how do we retrain people? And, how do we create more opportunity for people to land in great jobs?

But I do think it's going to be a massive boon for productivity in every sector. The way that Paravision thinks about it, like I just-- we just think about it as more from like a nerd perspective. Like, how can we create the coolest experiences for customers, the most seamless experiences whether it's like walking up-- and by the way, I'll sort of preface all of this by saying like everything opt in, right? If you don't want to have the experience, you shouldn't be forced into having the experience. And you know, several companies have kind of learned that the hard way. But it's something that we take a lot of pride in. So it's like, you know, if I'm walking into a-- walking into a basketball game, do I want to have to stand in a long line to actually just even get into the stadium? Do I want to have to further stand in a line to get a beer? Like the teams that succeed are going to have butts in seats, right? Like cheering their teams on. And so we see those experiences coming. We see pretty magical payment experiences coming ultimately that will be integrated with CRM.

So when I walk into, the same exact Starbucks that I go to every day. They know the drink that I want. And it's already like in the process of being rung up. Well, you know, when I get there and they're like, you know, “hi, Doug, do you want your, you know, your tea today?” I'll be like, “yes, I do.” That's like, great. Already paid for, you know, it's just done. There's a pretty-- once you experience them, they're pretty magical experiences. 

And then the flip side also, I would say, you know, some magical experiences and how do we protect our kids, right? The ability, you know, to have age estimation technology that keeps age gated content away from kids. If the, you know, state of the art is, are you 18, yes or no? I'm pretty sure my 14 year old can get past that. But if state of the art turns into actual age estimation, that is, you know, employs privacy by design where you actually don't need a full IDV on that transaction. That's pretty interesting to us in a way to just do a lot of good.

So yeah, I think it's-- for us, we get really, really excited about the good that we can do in the world and the making services kind of more freely accessible to anybody. I'm not sure if that answers your question. I know it's just like kind of an explosion but that's what we're seeing. And that's kind of who we serve.

Steve: I feel like each one of those use cases, we could go into and explore. And I think about what you said about consent and opting in, like people should have choice and how they do those experiences. And if they're wary of having a biometric or a face experience, like there's a more traditional route. I think choice is really key, but we're running up close on time.

I'm going to wrap up here in just a moment. If you've seen any of these episodes of EXECUTIVE SERIES, I like to go just a few steps beyond the profile. So I went from your profile on LinkedIn to your X profile, and I see you're a lover of travel and music. So I'm curious about places you like to travel to and what are some of your favorite musical genres?

Doug: God, first of all, I don't think I've actually really contributed meaningfully to X in years, so I'm surprised that all those things remain true. On the travel side it's a little bit actually kind of boring. I'll say like the places where I've spent a lot of time, where I live currently, Marin County, Seattle, Washington, where I spent 10 years kind of raising my kids and growing my career. And then where I grew up in BoothBay Harbor are still kind of three of the most magical places, you know.  My dad gave me some advice a long time ago, which is like, find the place that you want to live and make a life there that everything else will follow. And I truly believe that. 

And then outside of this country in those places, Stockholm is probably one of my favorite cities in the world; do a really great job of blending modernity with their history. And I proposed to my wife at the top of Mount Kilimanjaro in Tanzania. And so that holds a kind of special place in my heart too. So, and it's not one place, I don't think I could pick one place, but those are the, those are the top spots.

Steve: Very cool. And musical preferences for those spots? 

Doug: I probably similarly have a hard time choosing. If I had to choose, most people who know me would probably say like, I spend way too much time listening to like kind of singer-songwriter stuff from the seventies. James Taylor is probably my favorite artist bar none. You know, more on the rock side, Dire Straits. And then kind of more recently folks like Leif Vohlebeck and Ben Rector and Francis and the Lights. And I kind of, I love all genres of music, probably with the exception of heavy metal, sorry to any heavy metal fans out there, but that has its place too. Yeah. Usually in like kind of warm up for your teams when you're playing basketball or something like that, but yeah, I love music. Actually think I frequently describe our team as a jazz ensemble as well. Not, not just a, using a sports analogy. And I love jazz for the reason that everybody shines at their point in time and they come together to create pretty beautiful music holistically. But you kind of pass the ball in a similar way. 

Steve: The musical genre question is always tough because it depends on the mood. It depends on the place, what you're doing. Are you driving or are you at an event, et cetera? So that's great background. 

Well we're going to wrap up for today. I'm curious for those that are watching this or listening to this, what type of conversations would you be interested in having from the audience or the market?

Doug: I think your question about data is probably the, it's probably the most important question right now. Which is kind of who's ready for that, you know, who's been consistently collecting data in the right way with proper consents in place. I think that's going to be the kind of like next big phase that we go through. And, you know, fortunately or unfortunately, depending on if you're a company that's been collecting data that has not received proper consents for a period of time, you'll have a bunch of litigation. But you're also probably in some cases big enough just to pay off that litigation.

And I think that's probably the bet that a lot of the larger generative companies have made. For those of us that are a little bit more at, you know, have a niche. I think it's super important for our customers and potential customers to really dive deep on how that AI is created because they're going to be the ones that are held accountable for that.

So I think that's super important. I think we don't talk about it enough in the industry. It's kind of like, you know, kind of thought of as a dirty little secret. People don't talk about it because they don't want to know. But people should ask a lot more questions. 

Steve: Yeah, that's great. That's great. Well, Doug, thank you so much for taking the time to speak with me.

I learned a ton about Paravision today and your background. I really wish you much continued success. 

Doug: Yeah, thank you so much for having me you know, on the show and thanks for having this forum, by the way. It's a super cool forum. So I wish you a ton of success too.