Ross Freiman-Mendel – Networked Identity, Reusable Personas, and the Future of KYC

Full Transcript

Transcript lightly edited for clarity.

Riley Hughes: Welcome to The Future of Identity, a show that explores the world’s most innovative digital identity ecosystems and the people building them. I’m Riley Hughes, co-founder of Trinsic, and we’re the first identity acceptance network, helping businesses verify their users up to 10 times faster through partnerships with dozens of digital IDs. Today’s episode is a great one. I sat down with Ross Freiman-Mendel from Persona to talk about Persona’s digital ID products, which they call networked identity products. We unpack Persona Connect, their latest product launch, as well as reusable personas, mobile driver’s licenses, eIDs, and more. And Ross shared some fascinating statistics about conversion rates, reduction in drop-off, and the portion of Persona’s customer base who have adopted these networked ID offerings. What really stood out to me in this conversation is how practical it is.

Riley Hughes: Ross explains why earlier attempts at reusable identity struggled, the core problem that digital ID solves, which he framed as redundant verification, and how emerging challenges driven by AI are pushing digital ID adoption even higher. So whether you are a buyer or builder of identity verification products, this episode will give you a lot to think about. So I hope you enjoy it, and if you do, please share it with somebody who will benefit. That is the best way to signal to us that we should keep producing this show. So now to my conversation with Ross. I’m here with Ross Freiman-Mendel, head of product growth at Persona. Ross, welcome to the podcast.

Ross Freiman-Mendel: Thanks for having me, Riley. I’m looking forward to the conversation.

Riley Hughes: So Persona obviously is a leading identity verification player, identity orchestration platform. I don’t know, maybe you can introduce it better, but it feels like probably our listeners will know of Persona as one of the very successful companies in the space. But how about Ross? So Persona maybe needs no intro. It’s probably very well known, but tell us a little bit about you and your role at Persona and how you got there and, yeah, start there.

Ross Freiman-Mendel: Sure. So I’ve been at Persona for almost five years at this point, and I’ve worked across a variety of our surface areas. Most recently, I’ve worked on our network area of products, which corresponds to things like our consumer identity network, which includes Reusable Personas and the Persona Wallet app, our business identity network, which includes the most recently launched product, Persona Connect, and then more forward-looking things like KYA, Know Your Agent, and digital IDs, which would be things like an mDL, a mobile driver’s license stored in Apple Wallet. Before Persona, I was at Dropbox where I worked both on our product team and on our strategic finance team. And before that, I worked in investment banking on industrials companies, which has nothing to do with digital identity.

Riley Hughes: Did not realize that. That’s quite the jump from industrials to Dropbox and then now into identity working on. Sounds like a lot of the cutting-edge stuff at Persona, so that’s pretty cool.

Ross Freiman-Mendel: It’s definitely interesting. Network-based identity and digital identity are hot topics in our space right now, and they’re solving a lot of previously intractable problems for our customers. So it’s cool to be able to work on these issues and build useful things for our customers and their end users.

Riley Hughes: Yeah. This podcast, when we started it, the goal was to highlight the product capabilities of reusable ID products or network-based ID products and really drill in on— That layer of things, as opposed to just being really lofty, thought-leadery or whatever, and be a little more concrete. So it’s great to have you here, who’s the product lead of this kind of stuff. You mentioned this can kind of solve some intractable problems. Could you maybe speak to what do you mean by that? What are some of the kind of intractable things that this kind of category of product solves?

Ross Freiman-Mendel: Definitely. So if I were to maybe give you one umbrella concept, I would call it redundant KYC. There are lots of instances where the same end user has to verify their identity repeatedly. And identity, for very good reason, is siloed. So imagine, for example, that the type of person who drives for DoorDash may also deliver for TaskRabbit. In today’s world, that end user has to verify with DoorDash, and then when they want to go drive for TaskRabbit, they have to verify for TaskRabbit. That actual process of going through KYC for an end user, it’s redundant, it’s frustrating, it can be confusing, customers don’t like it either, it can lead to drop-off. So this idea of redundant KYC and then its partner solution, transferable KYC, which I think is another way to think about network-based identity, is one that’s getting increasing currency in our industry. And then the surface areas where an individual has to verify their identity, or the number of times that they have to verify their identity, is only increasing.

Ross Freiman-Mendel: So this need around transferability and portability is only getting more acute and important.

Riley Hughes: I really like the redundant KYC versus transferable KYC. When we recently went to a conference, we had plastic water bottles that we were giving away, and we also had, like, metal water bottles, and we framed it as, look, it’s like a disposable water bottle versus a reusable water bottle. Which one do you want? Of course everybody wants the nice double-walled insulated water bottle or whatever, and we tried to use that as a thing. But I really like this redundant KYC versus transferable KYC. That’s a good kind of way to frame it. What about that was intractable for customers historically, right? Maybe speak to, like, why wasn’t this built from the beginning at Persona or any other ID verification company, basically? Why is it being built at this layer as opposed to being built outside of, like DoorDash and TaskRabbit could just get together and decide to do this separately, right? Theoretically, anyway. So, yeah.

Ross Freiman-Mendel: Yeah, I think it’s a great question, and I might just reiterate that this issue is one that has been surfaced to us from our customers for years, and just digital identity vendors in general haven’t really had a great product solution. It’s not that we haven’t understood the problem; we just haven’t had quite the infrastructure in place to support it. There are a few reasons why this has been an intractable problem. Maybe the most prosaic reason is that identity is just a non-trivial product problem. And so for a DoorDash or a TaskRabbit, their focus on creating great marketplace and delivery experiences for their drivers and their end users, verifying the identity of these couriers is important for them, but it’s non-core to what they do. It’s more of a compliance checkbox thing. So that, if I were to speculate, would guess one of the reasons that vendors like DoorDash and TaskRabbit aren’t creating these connections directly.

Ross Freiman-Mendel: The other thing I would call out, both about reusable personas, which is the consumer identity network, and Persona Connect, which is the enterprise identity network, they are network-based products. And so it requires a certain density and ubiquity of usage to actually build rails that can be used by customers and end users in practice. And I just think for many years we weren’t at that level of adoption, but we’re kind of meeting the moment where there’s enough of a usage around IDV and enough overlap among end users that when you start to facilitate transferable KYC, it can actually meet a user base that’s in a position to benefit from it.

Riley Hughes: Yeah, that’s a really good way to frame it. Like maybe five years ago, the surface area that you mentioned, the, like, number of times a user had to verify their identity was—the juice wasn’t worth the squeeze, right? It’s like they’re going to onboard into a thing and then they’re not going to use it for a little while. But if a user is using multiple electric scooters or something, you need to do that, or a user is delivering for multiple platforms, we’re getting to the point where people— Can actually get benefit from it, is what I’m hearing from you, which is a great answer.

Ross Freiman-Mendel: Completely. I mean, there’s an additional, maybe I would say, fraud thrust that is encouraging this. Would it be a tech podcast if we didn’t talk about generative AI? But the scale—

Riley Hughes: of course not, right?

Ross Freiman-Mendel: The scale and sophistication of fraud is only increasing in a world where generative AI is broadly useful. And so network-based identity, both consumer network-based identity and B2B network-based identity, introduces a new set of fraud-deterring tactics that can just join the existing program that customers have in place, and it just gives our customers an additional tool to legitimately endorse good users and deter fraudsters from taking action they don’t want taken on their platforms.

Riley Hughes: Yep. So you mentioned Persona Connect and Reusable Personas as two separate products. You framed Persona Connect as the enterprise identity network and Reusable Personas as the consumer identity network. Talk to me about that distinction, because I think when most people think about a digital ID or a network-based ID product, they’ll think about it in terms of, of course, there’s the consumer side who has like a wallet, and then there’s the enterprise side, which is maybe issuing or consuming credentials to and from the wallet. I think that’s how most people listening to the podcast will think about the problem. So talk to me about how those two products map to this underlying concept.

Ross Freiman-Mendel: We think of them as two sides of the same coin, and there’s nothing that prevents a customer from enabling both Reusable Personas and Persona Connect. So maybe it would help to just more formally define what each of these products do. So Reusable Personas, you can think of it as a Persona-issued, consumer-owned identity wallet that’s backed by a passkey. It stores the PII that is collected by a customer when an end user goes through one of their flows. So for example, an end user goes through DoorDash flow, they create a reusable persona, they submit an image of their government ID, the reusable persona can store the image of the government ID. When that end user hits any other node in our identity network as an alternative to submitting the actual government ID, they can just authenticate into their reusable persona. So that’s why it’s the consumer part of the identity network.

Ross Freiman-Mendel: Connect is the other side of that coin because it allows the customer nodes within the network to create direct links with one another and transfer KYC data to one another with the permission, of course, of the end user. The way that we think about it is that it lets customers build customized sharing experiences. So a canonical example that we like to give is you have sponsor banks. They are building and providing banking infrastructure that a fintech app might consume. In a pre-Connect world, an end user may have to KYC with not only the sponsor bank, but also the fintech app that is consuming the services of the sponsor bank. Connect allows that same end user to still KYC with both entities, but to do so in a lower friction way.

Riley Hughes: So, in both scenarios, there’s a consumer consent element, the user’s in the loop, basically, right? And in both scenarios, friction is reduced, right? You get the same outcome with less friction. But it sounds like the difference is that with a reusable persona, the user has a thing that they are aware of, right? It’s like maybe an app on their phone or an account with Persona, essentially, that they authenticate to with a passkey. And in the other model, they may not. Is that a fair way to say it?

Ross Freiman-Mendel: Yeah, I think that’s a good way to think about it. So on the consumer side, reusable personas, the passkey is really what guards the wallet that the user owns. And when the user submits the reusable persona, they are authenticating into a passkey and unlocking the sharing of their data. When a user permissions the sharing of data with Persona Connect, customer A has some KYC data, which the user is permissioning customer A to share with customer B. So it’s just a different way to go about submitting the data so that the recipient organization can then perform KYC on it. And if you step back, this is all some derivative version of the same basic idea, which is that you need a way to submit PII to an organization that has an obligation to KYC an end user. In the pre-network world, it’s analog. I take a picture of my government ID. I mean, in a pre-digital ID world, it’s like literally analog. And then when we start to introduce network-based identity, we can afford tools so that consumers can more easily submit this data, or customers can share this data with consumer permission.

Riley Hughes: If we add the concept of third-party digital IDs, I know this is on my list to get to a little bit later to talk about Apple Wallet and mDLs and things like that. But maybe for the, like, data protection nerds listening, right, it’s like reusable personas. It’s the, like, first business, business A, that is the custodian or guardian or controller of the data for the user. And then in the third-party world, it’s like Apple is your custodian or guardian of the data because it’s in the Apple Wallet or something. And in all cases, users consent to share the data with some entity that needs to KYC them. Do I have that right?

Ross Freiman-Mendel: Privacy was a very intentional architectural choice that we made in developing reusable personas. It’s important to everything that we do, but obviously, like, the nature of the data that we’re dealing with there is particularly sensitive. And so there were a lot of decisions made to make sure that the owner of the data and the only person who could see that data and the only person who could permission the release of that data is the consumer because it’s their data after all.

Riley Hughes: That is good context. I wonder, we’ve identified this pattern, right? There’s a business that needs to verify the user. You’ve got to submit PII to them somehow so they can also some trust signals and whatever so that they can do their job. And there’s some ways that you’ve developed to make that a little easier. And I wonder if you could speak to the how much easier question on the reusable persona side or on the Persona Connect side, are there Metrics or outcomes or anything that Persona shares publicly or that you could share even at a high level to give us a sense of the magnitude of what does this really end up meaning for customers.

Ross Freiman-Mendel: Definitely. The way that we think about trying to quantify this are fraud metrics and conversion metrics. So I’ll start with conversion metrics would obviously vary slightly between reusable personas and Connect, but I’ll just talk about them in a general way. For our network-based identity products, the way that we think about it is through cohorts. So you have the cohort of users that verifies normally, and then the cohort of users that verifies with a network-based product. For the cohorts of users that verify with a network-based product, we tend to see huge increases in completion time and conversion. You know, on average, users complete flows 4x as quickly and they convert at twice the rate. If you were to go through a reusable persona or Persona Connect enabled experience, I think it would be self-evidently obvious why that happens. We’re removing steps from an end user. The other thing that’s, I think, quite interesting to highlight is the fraud implications of enabling network-based identity.

Ross Freiman-Mendel: So in addition to the conversion-based improvements, there are a series of fraud benefits that flow from relying on network-based identity. One of the things that surprised us with reusable personas in particular is that they emerged as a trust signal. And if you think about why, it actually makes a lot of sense. Every time a user takes any action, it contributes to your notion of their trustworthiness. And this has implications both for good actors and for fraudulent actors. So a good actor, and we see this with reusable personas, can reinforce their trustworthiness via expected utilization. This actually matters for Connect as well. If a user completes a flow, for example, with a reusable persona that has an NFC or an mDL verification in it, that is a very high assurance form of verification. And your trustworthiness of that end user is going to be comparatively higher for a user had they just gone through a government ID with a plastic government ID flow.

Ross Freiman-Mendel: The flip side of that is that fraudsters face an adverse scenario because, in general, they don’t want to share information that they don’t have to. And so network-based identity gives you another way to understand and observe anomalous activity, which could indicate something about the fraudulent or whether behavior was fraudulent or not fraudulent.

Riley Hughes: I want to follow up on both the fraud and conversion things. This is really good stuff. On the fraud side, am I understanding you that fraudsters may be more likely to revert back to the old ways of doing things, right, or uploading a document or things like that, and that the network-based products are primarily signals of good users? Is that the right way to think about it?

Ross Freiman-Mendel: I think so. If I could refine it a little bit, the fact that somebody uses a reusable persona, as an example, is in and of itself a trust signal for the reason that you just mentioned, which is that a fraudster doesn’t have to use it, and there are ways to provide less information by not using a reusable persona. Not only that, but the idea of assurance is one that’s quite important in the identity landscape. And so as you start to have network-based identity products, you can get a slightly more precise notion of the assurance of an end user. I think the most intuitive way to think about it is if a reusable persona, for example, was created off of a flow with a high assurance form of verification, knowing that information might tell you something the next time the user attempts to verify their identity.

Riley Hughes: And is the reusable persona a higher signal from a fraud point of view because there’s some feedback mechanism, right? Because as a user, if I reuse this in three places and I commit first-party fraud in one of those places, that somehow that feedback makes its way… back into the rest of things, and now that reusable persona is invalidated somehow, or is, like, flagged or whatever. Is that, like, a reputational type of thing, or not quite, right?

Ross Freiman-Mendel: Not at this stage. So I should be clear, because this is a very common question that comes up. We are never in the position of creating, like, global block lists or global allow lists. And so sometimes customers will ask us—I’ll just continue to extend the DoorDash TaskRabbit metaphor—if somebody was deemed fraudulent on DoorDash, can DoorDash just tell that to TaskRabbit? Reusable personas in Persona Connect do not do that. But there are signals about the behavior of that reusable persona across our identity network that might be additionally considered as part of a very broad menu of signals that our customers ingest that tell you something about the trustworthiness or not trustworthiness of the user. So, for example, if a reusable persona has verified five times in the last 30 days—I’m making this up—that may or may not be anomalous compared to the behavior of reusable personas that we see across our identity network. We can make that information available to a customer who can then decide whether or not to act on that signal.

Riley Hughes: That makes sense, and that aligns with—you attended a little panel that I was on at a conference recently, and it was a panel with someone in the identity verification, document verification world, and basically the consensus was, like, I made the point that mDLs specifically are useful for signaling good users or whatever, and his point was, like, the fraudsters are still going to fall back to the weakest link or whatever, which is either going to be the phone number stuff or the document stuff or whatever it may be. And so you still need good document verification or still need good phone number, like, data and things like that, because that will still be a vector. But I wonder if there’s—do you see Scenarios where people, like, what happens when a fraudster gets a reusable persona? And does that just mean they can just, like, use it everywhere and they’re all set? I’m sure not. So I guess, how do you think about that? And also fraud is not like a binary, right? It’s not just either fraudster or perfect user, right?

Riley Hughes: There’s different signals and different behaviors and different contexts that probably come into play. Yeah, how do you think about that?

Ross Freiman-Mendel: So there’s maybe a few questions buried in there, and I’ll try to tease them out. The first is, so often in digital identity, the conversation is about how do we block fraudsters. It’s just as important to enable legitimate use, and it’s useful to our customers to make it easier for their end users who are taking activity that the company is willing to sanction to go ahead and do it and not block those users. We have a random statistic that we often mention where flows that collect government ID and selfie, the drop-off can be as high as 25 to 35 percent. Surely those are not all bad users. Those are people who are just hitting a high-friction flow and not able to proceed through it. So endorsing legitimate use is a very important part of why we want to build network-based identity products. In terms of the question of can a fraudster get access to a reusable persona, they can. And what we tell our customers is reusable personas are not a cure-all. They are part of what ought to be a much more comprehensive platform-based approach to deterring fraud.

Ross Freiman-Mendel: And so I would think about reusable personas as exposing a set of signals. about a user that a customer can ingest. And a lot of the work that we do at Persona is not necessarily on the reusable persona itself, but the tooling that we enable to ingest that data and then empower our customers to take action on the data in a way that respects their risk tolerances. I think maybe the final part of your question is that obviously, like, risk is unequal across customers and across use cases. So the type of risk that you might be willing to assume for somebody who’s going through a low-risk use case is very different than the type of risk you might be willing to assume for a customer who’s operating in a highly regulated space or administering, like, a high-risk use case like an ATO, for example. And the way that we think about working with them is not to create, like, a definitive posture: this is how you should ingest reusable personas, but allow them to basically be granular about how they manipulate the data.

Ross Freiman-Mendel: In the former example, the low-risk use case, you can choose to accept all the information off the reusable persona. In the high-risk use case, a customer may elect to only accept the government ID and step up to other higher-assurance forms of verification. That also exists with Persona Connect, by the way. There’s obviously not one-to-one mapping of KYC data between customer A and customer B. And so a lot of the product work that we’re doing is thinking about how we can make the handoff to step-ups, if and when they happen, seamless and then differential based on the risk assessment that you’re applying to a given user and a given use case.

Riley Hughes: That makes sense. One of the ways that I think about this is talking about when people think about reusable ID products or digital IDs generally, a lot of times people assume the thing that is reusable is the decision from the first enrollment, right? So, is this trusted or not is like the thing that’s reusable. But it sounds like what I’m hearing from you is that these products are more granular in the sense that it’s almost the inputs that are reusable, right? If a user uploaded an ID document in scenario number one, they can reuse that input in scenario number two, but that the decision criteria and the overall fraud posture for company number one and two are different. And so they may get to different decisions on the same set of input data or something. Is that the right way to think about it?

Ross Freiman-Mendel: Exactly right. I can translate it a little bit into persona parlance, which is often our customers ask us, are we— persisting verification results across our identity ecosystem? And the answer is no. What we are trying to do is basically enable the same end user to submit the same data to different customers and then build tools that allow different customers to treat that same data differently in a way that respects their risk tolerance. And I can tell you, across the thousands of customers that we have, it really is quite varied.

Riley Hughes: That is really helpful on the fraud point. I want to go back to the conversion point that you mentioned, which is that you said four times faster and convert at two times the rate. I think you said in more analog forms of verification, the drop-off can be between 25 and 30 percent or whatever it was. And so does this two times the conversion rate mean drop-off is in the 12 to 15 range, or are these two different numbers?

Ross Freiman-Mendel: I would say two different numbers that are possibly spiritually related. In the same way, you get this drop-off because these flows are high friction, and we observe these completion and conversion time improvements because we are removing friction. And I should say, and we’ll probably discuss this when we discuss mDLs, it’s very rare in our world that you can simultaneously improve the user experience and produce a more secure outcome for customers. I think that the trade-off forever has kind of been you could literally block all fraud if you block all users, and you can let in a lot of fraud if you basically have a very lightweight identity program in place. And the challenge is contending with that trade-off. But portable identity, network-based identity, digital identity allows you to have more of a win-win type scenario where you can enhance the end user experience and simultaneously ensure less fraud on your platforms.

Riley Hughes: I totally hear you. I think one of the, maybe to speak to the challenges here and how you’ve overcome those, I know that in— My experience in the space, there’s different challenges, one of them being this chicken and egg problem. And you spoke to that at the beginning, which is maybe platforms like Persona and the concept of verifying a user’s identity are widespread enough now that you can overcome that just by sheer size of the customer base and the overlap between different users interacting with different platforms. So that makes sense. But I’ve also heard contention from businesses as it relates to reusable ID that goes a little bit something like: in order to enroll in a digital ID, there’s at least one extra step, right? There’s at least one extra button to save a passkey or something, or whatever the case may be, download a mobile app, right? That’s a lot more than just drop a passkey. But there’s at least some extra friction at that enrollment step. And I’ve never seen anybody get away from it.

Riley Hughes: It just is what it is, especially you need the user to consent and be aware of it and whatever, and so it just, it is what it is. And sometimes that may mean agreeing to the—I’ll just use Persona in this example. I don’t know how your product is set up in this way, but just the user agrees to the Persona terms of service or something when they get save the passkey or whatever, which is happening inside of the DoorDash or TaskRabbit. And again, just using these as examples. I don’t know if those are customers, but just conceptually, right? And a lot of times, like a DoorDash or something might say, I don’t want you to add one extra step of save a passkey and have the user agree to your terms of service inside of my UI, when the real company that’s going to benefit from the fact that was just saved here is TaskRabbit, not me. It’s like in the upfront, there’s a little extra friction at least, and the party that benefits from that little extra friction is not the entity that endured that extra friction or allowed it to happen, right? It’s somebody else downstream.

Riley Hughes: Would you correct anything that I’m saying here? And if you think I’m representing this properly, then I guess speak to me about how you address that or how your customers think about that.

Ross Freiman-Mendel: I think you’re representing a conversation that we have frequently, accurately. And actually, what I would say, because I think your question is basically asking, how do we think about scaling the size of our network, especially when the size started at zero? The answer is very different for reusable personas and Persona Connect. So maybe I’ll take each in turn. Reusable personas, we faced the classic challenge that you’re describing, which is to create the credential introduces friction. How do you convince a customer to introduce that friction when the size of the network itself is maybe not dense enough to produce the conversion and the completion time benefits that I just mentioned? What I would say is we were very fortunate to have a large built-in user base, and so we started our efforts by trying to find those customers who just have a higher appetite to try new products and are more interested in operating at the identity cutting edge, or customers who just had less conversion sensitivity around their flows.

Ross Freiman-Mendel: And getting the first 10 customers is so much harder than getting the next 10 customers, and success begets success. So it was a lot of just being kind of surgical and prescriptive around finding the right candidate customers to begin pollinating the network, and then we grew it from there. Persona Connect is slightly different. We have actually made an explicit choice to focus on growing the network within crypto, and the reason for that is twofold, actually. One is practical, one is philosophic. The practical reason that we decided to focus on crypto is when you think about, like, the different types of players—on-ramps, off-ramps, exchanges, wallets—the same end user can end up hitting multiple KYCs almost immediately. And so the issue of redundant KYC is so acute that being able to have any answer for that was interesting and alleviating to those customers.

Ross Freiman-Mendel: Also, maybe just as like a side tangent, the philosophic reason that we decided to start with crypto is they’re not specifically interested in decentralized identity, which is just a byproduct of, I think, the ethos of crypto, which is trying to decentralize finance. So that has been a huge focus on Persona Connect. In terms of jump-starting the network, there are a lot of core nodes in the network, like these infrastructural layers on top of which many applications are built. And so we were quite surgical about targeting companies like a Coinflow or a Bridge, because we knew that if we could activate a Coinflow or a Bridge, the value that redounds to all of the recipient partner orgs would be outsized. As we move into trying to get, for example, sponsor banks and fintechs in 2026, it’ll be the same focus on growing the network. And so this is a very long way of saying there’s lots of different ways to grow the network, but it just requires creativity, persistence, and understanding the dynamics of the customers and industries in question.

Riley Hughes: Yeah, that is interesting. I mean, I think the question I asked was maybe twofold. I think the one part of it you’ve already covered. The other part of it is even now, imagine your network is really, really big. You’ve got half the U.S. population in it or something, right? And so it’s like, obviously any U.S. company is going to, if they accepted these reusable personas, they would see the benefits that we’re talking about here if the network is that big. But then I could imagine a scenario where a customer says, Yeah, I’ll accept it. I’ll accept these IDs if they’re there, but I don’t want people— Saving new reusable IDs and getting a Persona passkey in my user experience, right? Like, you can grow the network with these other people. I’ll accept IDs from the network, but I don’t want to be, like, helping to enroll in the network. Maybe because I’m ride-sharing company A and I don’t want ride-sharing company B to, like, benefit from the users that I’m onboarding or something, that sort of thing. So contend with that concept for me a little bit.

Ross Freiman-Mendel: It actually— so we were concerned about this. Maybe concerned is the wrong word. We were aware of this dynamic. It actually doesn’t come up as much as you think it would. I would say one of the things that we often share with customers is that when you are turning on reusable Personas, you are not only allowing your end users to create these credentials, but you are inheriting a network that now covers over 90% of our customer base, for which millions of credentials have been issued and accepted. And generally that’s pretty convincing to our customers. The other thing that I would say is the cohort improvements that you would see, like the conversion and the completion time increases—it’s, you know, one of these, like, something goes up for a set of users, something goes down for a set of users, but what you care about is the net effect. In general, our customers find that the conversion and the fraud improvements outweigh any slight drop-off that you might see for the sets of end users who drop off when they proceed to create a reusable Persona.

Ross Freiman-Mendel: The final thing that I would call out is we do spend a lot of time thinking about the right way to craft the creation and the submission experience. That too is actually a non-trivial product problem. I just call out a couple of examples. What copy do you use? What is the binding mechanism? It doesn’t have to be a passkey. It could be phone, email, selfie. Where in the flow is it placed? Is it an embedded step in the flow? All of these questions we think about quite a bit, and we are flexible with customers to insert that step. The way that it’s appropriate for a customer, depending on their use case and their need.

Riley Hughes: So that helps a lot. I think what you’re saying is reusable personas is something you can turn on. You could choose not to have it at all, or you could choose to turn it on and you get both the benefit and you get the acceptance and the enrollment if it’s turned on, and that the benefits of acceptance outweigh the costs of enrollment, basically. And so that’s a pretty convincing argument if you’ve got the numbers to back it up, which sounds like you do. And speaking of numbers, did I hear you say that 90% of Persona’s customers use reusable personas?

Ross Freiman-Mendel: Yep, over 90% of our customers are activated on the network.

Riley Hughes: Wow. I’m surprised by that. That is not because of the value prop not being there, but because it’s hard to get 90% of anything. It’s like if you polled Americans, who likes chocolate, right? You’d probably get 85% of people or something. That is a lot. So yeah, congrats. That signals quite a bit of network success. And I think, to your point on being surgical and prescriptive about those first customers and building the network out methodically from there, it sounds like that approach has worked. You’ve been involved in this specifically, reusable personas, from the beginning. I’m sure you had some success metrics lined out at the beginning, right? Has this been more successful than you were expecting, or about as successful as you were expecting, or how has it played out relative to expectations?

Ross Freiman-Mendel: I know you won’t believe me when I say this, but our ambition for this at the start was very quaint. We understood the problem, and we thought that the solution to the problem was useful. But when we conceived of reusable personas, we conceived of it with success metrics along the lines of, it would be great if we could convince five people by the end of the year to use this, or if we can get, you know, a bunch of customers to verify. It’s probably just derivative of Persona’s product development culture, which comes from our co-founders, which is, you know, we take it day by day, and as long as we’re building useful things, we’re not really goaling ourselves in terms of how big we’re trying to make something by the end of the year. Reusable personas have never had an annual goal. We’re going into 2026 product planning. There is no annual goal. The focus is much more on the sets of things that we want to build to make it useful for our customers. It’s a long way of saying that we’ve been pleasantly surprised.

Riley Hughes: I want to touch on kind of the third. third-party digital IDs a little bit as well. I don’t know if that’s the way that the Persona parlance necessarily, right? But specifically, I saw on your website mobile driver’s license and Apple Wallet listed there. I’m sure that you are looking at Google Wallet and looking at the broader landscape as well, but Apple Wallet is what I saw on the website, so that’s what I bring up here. Tell me about your journey there. The support for Apple has been there for quite a while. I think the blog post that announced support for it is a few years old at this point. Curious, what is different and what is the same relative to what we’ve already talked about, Persona’s network-based products? What is similar and different about these specifically, maybe Apple Wallet and mDLs versus the Persona network-based products?

Ross Freiman-Mendel: So what’s similar about—well, maybe actually, I don’t know if your listeners are familiar, but just, you know, mDL stands for mobile driver’s license. State DMVs, and really globally, people are issuing digital forms of their driver’s licenses. The United States is obviously decentralized and super fragmented. So unlike our counterparts, for example, in Europe, the way that this seems to be manifesting in the United States is state DMVs are in the process of digitizing their driver’s licenses, and then the two primary ways that people will be able to store these driver’s licenses are either in an OEM wallet like Apple, Google, or Samsung, or in a wallet issued by the state DMV. You’re right, Riley, that really one of the first and most public partnerships we have around this is enabling acceptance of an mDL stored in the Apple Wallet. So the end user experience would correspond to basically a user authenticating into their Apple Wallet with Face ID to submit the mDL as an alternative to taking a picture of their government-issued ID.

Ross Freiman-Mendel: The ways in which it’s the same, I would say that mDLs are also one of those rare solutions — maybe it’s not so rare if I’ve mentioned it twice in this conversation — that increases security and fraud prevention and simultaneously produces a smoother user experience. Some of the ways, though, I would highlight that it’s different is the education, application, or burden has been much lower with customers on mDLs. I would say network-based identity products, that requires a lot of explanation. It’s not necessarily intuitive. mDLs are self-evidently, obviously a thing that all of our customers want, and there hasn’t really been much explanation of them about why they would want to use these things. It’s really been actually them coming to us, putting pressure on us to be able to announce support with the various OEMs and other third-party identity providers. So those are probably two of the biggest similarities and differences that I would highlight.

Ross Freiman-Mendel: I mean, I can go into detail on the ways in which adoption has been different or coverage has been different, but I would say spiritually, under the same umbrella.

Riley Hughes: Yes, for sure. And I think what you’re saying makes total sense. If you’re accepting a picture of a plastic one, you probably are interested in accepting the digitally native one. One is a picture of the thing, one is the thing itself. So that makes sense. I can see why people would want it. We see a similar thing, of course, obviously at Trinsic. But I guess, how do you contend with somebody shows up in California, right? There are four wallets that a California mobile driver’s license could live in. Let me actually — Samsung’s not live yet, but it might be by the time this podcast is out. But if somebody shows up in Arizona, there are four different— Wallets that they might hold their mobile driver’s license in, right? There are Apple, Google, Samsung, and a state wallet. And each of those has different degrees of adoption. You can’t get your mDL in two of them in Arizona, at least, and so it’s only in one of them. And you don’t know which one it’s in, and you don’t even know if this user is one of the people who has the mDL in that state.

Riley Hughes: Maybe they’re in Arizona, but they’re from another state, and they’re visiting because they’re on vacation or golfing in Scottsdale or something. I don’t know. Like, how do you productize that in a way that makes it useful for customers to accept an mDL, given the, like, fragmentation and adoption challenges?

Ross Freiman-Mendel: I should say that the fragmentation is worse than you describe, because that’s just the fragmentation that exists among digital IDs, but you need to build flows that work not only for digital IDs and the various places that they can be stored, but also the different types of physical analog IDs that you can provide, which is not just driver’s licenses. I would say that, in general, Persona is trying to be platform agnostic, and so we’re not betting on a single winner, and we’re certainly not trying to force customers into one model at all. What we’re doing from the platform side is basically trying to build end-user experiences that seamlessly give users the choice to submit the right identifying document, and then useful fallbacks in the event that they’re either unable to submit the document or something is messed up about the submission. And then on the customer side, basically, as we’ve been discussing with network-based identity, the tooling to make these granular configurations in a way that is specific to the use cases that they need.

Ross Freiman-Mendel: So we only expect that this proliferation of digital IDs to increase and this fragmentation to increase, and our general platform vision is to be able to support whatever ID a customer wants to accept in any circumstances, and we’re making no determination around what the right ones or the wrong ones to accept should be.

Riley Hughes: I think that makes total sense, and I feel a similar way about the fragmentation and what I would bet on. I guess, how if a customer came to you and asked, like, how should I incorporate the Arizona mobile ID wallet into my flow? Is it, like you mentioned, like any product change has some benefits and some drawbacks, right? Like you change one thing, you’re going to get an uplift in conversion over here, but a reduction in conversion over here. I guess what I’m wondering is if you add the verify with your Arizona mobile ID wallet into your flow as, like, one of the options, are there enough users where showing them that option is worth it? And how do you know when to show them that option? And I’m thinking about the user experience and kind of the conversion and invocation of this digital ID step and how you think about that from a customer point of view.

Ross Freiman-Mendel: So it’s a really good question. And actually what I would say is you rightly point out that we released this blog post on Apple mDL a few years ago, but the conversation with customers became a lot more urgent and interesting after California announced support. The reason is that California obviously is the largest state, and it suddenly made it real for our customers because there was a possibility that a large proportion of their end users may actually have the ability to submit these mDLs. And so I think you are right to point out that for any given customer conversation, there’s always this question of, like, here are the benefits. Am I willing to incur the cost of receiving those benefits? And in the topic of U.S. mDLs specifically, that conversation usually centers on coverage. But all of our customers are asking us now is when will the next large states launch?

Ross Freiman-Mendel: I mean, we want to add support for all 50 states, but obviously when a New York or a Florida or a Texas goes live, that suddenly step function increases the coverage available and the likely impact that a customer would see from both the conversion and a fraud perspective.

Riley Hughes: Yeah, I think right now in the U.S., there are 17 different wallet apps that a user’s mDL could be in.

Ross Freiman-Mendel: That doesn’t surprise me.

Riley Hughes: Yeah, it’s much more fragmented than even people imagine, right? There’s 24 states issuing mDLs. It’s 17 wallets. It’s a lot. Ross, as you mentioned, this would not be a tech podcast without bringing up AI. Really enjoyed the conversation we had so far and probably dug in pretty deep on the first pieces because I thought we were just on a roll there. So maybe we could do a part two of this someday, but to go deeper here also. But you mentioned KYA and know your agent and agentic AI flows at the beginning of the conversation as a piece of kind of your scope. Tell me a little bit about what that means, right? What is real here? What does that end up looking like from your point of view? And yeah.

Ross Freiman-Mendel: Sure. KYA, know your agent. It’s definitely a newer part of the identity landscape, and one that I would say is mostly conceptual. But like mDLs or like network identity, you kind of see the future. We know it’s going in this direction. The question is not if, but rather when. Obviously, in this world of agents, people want to authorize agents to take action on their behalf. And there’s this basic question of identity that comes up, which is, who is the verified human that this agent is acting on behalf of, and is the usage legitimate? So KYA is basically trying to answer that question. And actually, interestingly, agentic AI, it basically breaks the fundamental assumption of the internet, which was that we do not want bot activity happening on the internet. When you go through a CAPTCHA, that’s basically trying to say, are you not a bot? Suddenly agents explode onto the scene, and now we want bots to take action on our behalf.

Ross Freiman-Mendel: So there’s a lot of product thinking that we’re going to have to do to operationalize and empower these agents to take useful, legitimate action on behalf of users. What I would say that we’re seeing in partnership with customers today is that the actual companies that are implementing agentic action are starting with low-risk use cases that don’t really need to touch PII and/or complete regulated action. I’ll just give you a specific example. We’re working pretty closely with one of these agentic browser companies, and we were talking about building a shopping cart. And the way that this browser company looked at building a shopping cart was primarily as assistive and collaborative. So for them, it would be a successful outcome if the agent could do some research, build your cart, get you to checkout, but then hand back to the end user to actually go through the process of checking out. The reason for that is to avoid, understandably, The questions around identity and liability complexity.

Ross Freiman-Mendel: I would say that as the usefulness of agents increase, and it’s going to be a prerequisite for their increased usefulness that they can start completing higher-risk transactions like doing checkout and payment on that cart, that’s where KYA is going to become more important. And for us, it’s like the same basic canonical questions of identity: is this use legitimate? Is the person, in this case, is the agent acting on behalf of someone who is who they say they are?

Riley Hughes: It will be interesting to see how KYA plays out because a lot of network-based identity products and solutions are proprietary at some level, right? So even for a standards-based solution like an mDL in your Apple Wallet, Apple still has a proprietary kind of approval and vetting and KYB process that they conduct on anybody who wants to request the mDL from the Apple Wallet. And obviously, in the reusable personas, I got to be a customer of Persona in order to access that network. But in KYA, it really seems like a quintessential example of where standards are necessary because, man, like, if every single company that my agent could interact with needs to be a Persona customer, that’d be pretty challenging, right? On the other hand, if there are a few big agent companies, OpenAI and Anthropic and whatever, a few big ones, and they dominate the agentic market at some level, then maybe it does go this route where there are a few big players and it doesn’t need to necessarily be fully standards-based. How do you think about that?

Riley Hughes: Like, just if you were just total speculation here for a moment, right? How do you think this will play out? Do you think there will need to be new standards that come into the picture to help companies or to enable KYA and everybody builds to these standards? Or do you think it’ll end up a little more—we see the federated login world where it’s like we’ve got a few big login players and IDPs and it goes that route?

Ross Freiman-Mendel: Maybe I’ll answer your question by not answering your question, which is that, you know, the three most important words in the English language are I don’t know. And I think it would be most honest for me to say that I don’t know. We’re certainly in the phase right now of basically trying to understand how our customers think about implementing agentic action in high-risk use cases. And I view understanding that as kind of prerequisite to opining on, like, what are going to be the right standards to support that. And I think even our customers are trying to figure out right now what does it mean to facilitate these use cases. So I don’t know, but I know that the way that we’ll answer it is by staying close to these customers and trying to co-develop and be thought partners with them, which is really how we develop all product at Persona.

Riley Hughes: Yep. I think that’s very smart. That is obviously the right answer. I guess one thought that I’ve had on this is, like, a lot of agent builders are building for low-stakes use cases, like you mentioned, and the things required for high-stakes use cases are different than the things required for low-stakes use cases. So on the one hand, we want to pick the low-hanging fruit, that’s good, but also… We don’t want the ruts in the road to form around the paradigms and patterns that work for the low-stakes things, and then we try to bring the high-stakes things into the and run them on the same rails, and then find that things don’t work or that it’s not as secure or whatever as it could be. And so I’m really interested to see how this space pans out because I haven’t seen agents doing anything really high stakes, like you mentioned, yet. And I’m really curious the path to get there. What will that look like? Yeah, I guess we’ve wandered into speculative territory here, but just curious if you happen to have anything interesting there.

Riley Hughes: But it sounds like we’ll need to do another one of these in a year or something.

Ross Freiman-Mendel: More to discuss in part two.

Riley Hughes: Yeah. Good. Ross, maybe zoom out. Anything we didn’t cover that you wanted to?

Ross Freiman-Mendel: No, I think that it was a useful opportunity to just discuss the different ways that we’re thinking about identity. I’m sure you discuss this with your customers and listeners all the time, but we definitely are in interesting times with all the different types of topics that digital identity is covering and the ways in which it’s evolving. So thanks for the opportunity to speak.

Riley Hughes: Yeah. We close things out by asking everybody the same question, which is, tell me what the future of identity looks like to you. Why does that matter for the world? Why do you get up in the morning trying to make the future of identity at Persona?

Ross Freiman-Mendel: Maybe I would just highlight two things that I think is going to happen in the future of identity and I think is important to happen in the future of identity. The first is it’s a future that ultimately is privacy preserving. I just think that’s, like, super important that we never lose sight of that, like the importance of maintaining privacy. We want to make sure that people are KYC compliant, blocking bad actors, things like that. But there’s kind of a good way to do that, in my opinion, and a bad way to do that. And I think and hope that we can tend towards a world that kind of respects users’ privacy. The second is just convenient. I think for anybody who goes through any of these flows, even the most well-designed flows are difficult, they’re frustrating. And for anybody who has the magic of, you know, basically using Apple Pay to pay, that is like a future that we foresee enabling with identity as well. Payments happen to be ahead of identity, but it’s one that we’re marching towards.

Riley Hughes: Great. Well, that is certainly a future I want to live in. So thanks for helping to build it.

Ross Freiman-Mendel: Amazing. Thank you for helping to build it.

Riley Hughes: If people want to learn more about reusable personas or Persona Connect or anything like that, how can they get in touch or reach out?

Ross Freiman-Mendel: If you go to withpersona.com, we have all this information on our website. I’ll specifically plug Persona Connect as one of our more recent product launches under the network umbrella, and it’s one that we’re quite excited about.

Riley Hughes: Good. Okay. Thanks a lot, Ross, and we’ll talk to you soon.

Ross Freiman-Mendel: Cool. Thanks, Riley.

Riley Hughes: Thanks so much for listening. If you enjoyed this content, the best way to signal to us that the content is valuable is to share it with others who will benefit from it. Meanwhile, if your organization is interested in accepting digital IDs, you can find me or Trinsic on LinkedIn or X or on our website at trinsic.id. And if you haven’t already, visit trinsic.id/podcast to subscribe to the Future of Identity newsletter and listen to any of our prior episodes. Thanks so much for listening.