Latest / The Index Podcast / Is Identity the Internet's Biggest Unsolved Problem? - Fidelity Labs
Transcript
- 0:08SPEAKER_00: One quick word from our sponsor, Triton One.
- 0:10SPEAKER_00: If you're building on Solana, your app is only as fast and as reliable as the infrastructure underneath it.
- 0:17SPEAKER_00: At Triton, we built the open source read layer for Solana ourselves.
- 0:22SPEAKER_00: The streaming engine that pushes data the moment it hits the chain, the transaction pipeline that lands your trades when everyone else is timing out, all of it running on bare metal in over 20 data centers worldwide, served by the team that wrote it.
- 0:34SPEAKER_00: We power some of the biggest names on Solana, but with pay as you go, anyone can start.
- 0:39SPEAKER_00: Sign up, get an endpoint, pay only for what you use, head to Triton.
- 0:43SPEAKER_00: Serious Infrastructure for Serious Builders.
- 0:45SPEAKER_00: I've worked with Triton for years at this point.
- 0:48SPEAKER_00: They've been around since the very beginning of Solana.
- 0:50SPEAKER_00: They've contributed open source software.
- 0:52SPEAKER_00: That's the core of the infrastructure of Solana that makes everything work since day one.
- 0:56SPEAKER_00: I couldn't recommend working with a better team.
- 0:58SPEAKER_00: Triton.
- 1:05SPEAKER_00: Hey everybody, welcome to the index.
- 1:07SPEAKER_00: I'm your host, Alex Cahaya, and today I'm joined by Evan Schneidman, head of Fidelity Labs.
- 1:13SPEAKER_00: He went from studying central banks at Harvard to founding Prattle, which sold to LiquidNet in 2019.
- 1:19SPEAKER_00: And now he's asking what it takes for institutions to trust an AI agent.
- 1:23SPEAKER_00: Evan, I'm really excited to have you here.
- 1:24SPEAKER_00: Thanks for being on the show.
- 1:26SPEAKER_01: Alex, thank you so much for having me here.
- 1:27SPEAKER_01: I appreciate it.
- 1:28SPEAKER_00: So before we get into what you're doing as head of Fidelity Labs, just give me an idea of what gets you out of bed in the morning.
- 1:35SPEAKER_00: I mean, you sold your glass company, you probably could have done anything.
- 1:38SPEAKER_00: You know, what is driving you to do this work?
- 1:41SPEAKER_01: Honestly, I just love the world of financial technology and the opportunity to build transformational fintech businesses.
- 1:49SPEAKER_01: Our technology systems have improved much faster than our financial systems have.
- 1:53SPEAKER_01: And so the question is, how do we infuse tech into our financial services ecosystem to better serve customers, to operate more effectively, more efficiently, and really just make sure that people can live their best financial lives?
- 2:04SPEAKER_00: So what is Fidelity Labs for people who don't know Fidelity very well, maybe, and also just like what a lab inside of a company like that would even do?
- 2:14SPEAKER_00: Yeah.
- 2:15SPEAKER_01: So Fidelity Labs is Fidelity's in-house innovation and incubation division.
- 2:20SPEAKER_01: So essentially, what we do is we build businesses that have the potential to be the future businesses of Fidelity or the potential to spin out and scale independently.
- 2:29SPEAKER_01: So essentially, you can think of us like a venture studio that exists inside of the financial services behemoth that is Fidelity.
- 2:36SPEAKER_01: Our job is to kind of look for what could or should exist for Fidelity existing customers and clientele.
- 2:43SPEAKER_01: And just more broadly in the financial services ecosystem, where where can we add value?
- 2:47SPEAKER_01: Where can we build a net new business?
- 2:49SPEAKER_01: We also have a sister organization that focuses on research and development.
- 2:52SPEAKER_01: So the Fidelity Center for Applied Technology actually focuses on RD.
- 2:56SPEAKER_01: And then we in Fidelity Labs focus on building out net new businesses with the opportunity to really serve our clients as best we can.
- 3:04SPEAKER_00: That's really interesting.
- 3:04SPEAKER_00: I want to talk more about that, especially as it relates to AI and agents inside of institutions like this.
- 3:10SPEAKER_00: But before we do, I want a little bit of hit more history on Prattle.
- 3:13SPEAKER_00: What led to Prattle?
- 3:14SPEAKER_00: I mean, it seems like it was somewhat related to the research you did on central banks.
- 3:18SPEAKER_00: I would love to dive into that.
- 3:19SPEAKER_01: Yeah.
- 3:20SPEAKER_01: So I actually started my career as an academic game theorist.
- 3:24SPEAKER_01: So I was studying small group decision dynamics essentially during the financial crisis.
- 3:29SPEAKER_01: And decided at that moment in time that the most interesting small group of decision makers in the world was the Federal Reserve, which circa 2008, 2009, I didn't feel like that was a bad call.
- 3:38SPEAKER_01: I uh decided to dig into how do central banks make decisions and how do those decisions affect financial markets.
- 3:45SPEAKER_01: What I figured out probably took me a little bit longer than maybe it should have, but uh what I finally figured out was that uh there were only a couple dozen academics in the world that really cared about central bank decision dynamics.
- 3:55SPEAKER_01: And in fact, markets didn't care much about their decision dynamics.
- 3:59SPEAKER_01: Markets didn't even really care about policy.
- 4:01SPEAKER_01: What markets focused on, and I think still focus on, is communication, because how the market interprets the communications coming out of central banks actually really impacts the future of policy and that roadmap and then, of course, the reaction function.
- 4:16SPEAKER_01: And so what I realized along the way was that I was studying the wrong thing.
- 4:21SPEAKER_01: That in studying central banks, if I wanted to study something really impactful, I had to analyze market-moving language coming out of those central banks.
- 4:28SPEAKER_01: And so um I backed away from game theory, focused in on uh natural language processing, and developed a novel methodology to mathematically map linguistic patterns and tie them directly to price action.
- 4:39SPEAKER_01: Wow.
- 4:40SPEAKER_01: Long before the era of large language models and uh GPT, you know, we we were, I mean, essentially at that moment in time, this the industry standard was a back-of-words approach.
- 4:50SPEAKER_01: It was a glorified good buzzword minus bad buzzword equals score.
- 4:54SPEAKER_01: And so we took a much more advanced methodology and said if we are able to identify how every word, phrase, sentence, and paragraph interrelates to one another and ties directly to the price action in markets, what we can do is actually identify how hawkish or dovish a central bank communication is in in near real time.
- 5:14SPEAKER_01: And so that became the core, so that was my academic research, that became the core of what was Praddle.
- 5:20SPEAKER_01: So we started doing that on central banks and then realized there was a whole other application of doing that with corporations.
- 5:26SPEAKER_01: So things like earnings calls, regulatory filings, speeches by corporate officers, press releases, um, analyzing the most impactful market-moving language, not just from policymakers now, but from C-suite executives as well.
- 5:39SPEAKER_00: I mean, I find it fascinating.
- 5:40SPEAKER_00: I was sitting here thinking about my early days working in the blockchain and cryptocurrency space since like 2016.
- 5:46SPEAKER_00: And that was my first exposure to protocol design and game theory.
- 5:51SPEAKER_00: Like we had all these game theorists like on the team, you know, figuring out how people would attack the network and how they would do whatever they needed to do with like how the tokens could potentially like compromise usage of the network and all that kind of stuff.
- 6:05SPEAKER_00: And it just there's a lot of parallels between that.
- 6:08SPEAKER_00: I mean, it's basically the same thing, especially if you do launch like a successful thing, it can lead to, yeah, you have to be very careful about what you communicate to the world if you're like running a foundation.
- 6:18SPEAKER_00: So that seems very related.
- 6:21SPEAKER_00: So what did so Praddle was like a tool for helping corporations manage what they said or helping traders predict?
- 6:29SPEAKER_00: Like what was the tool actually, uh, what was the product market fit for that?
- 6:34SPEAKER_01: You're quick.
- 6:34SPEAKER_01: It took us a little longer to figure out that uh there was two sides of that coin.
- 6:37SPEAKER_01: So um we were sell selling those analytics to essentially to hedge funds, asset managers, and banks, right?
- 6:45SPEAKER_01: Essentially helping them interpret first central bank communications, then corporate communications.
- 6:50SPEAKER_01: Um, a few years into the business, I gave a presentation at an investor relations conference, which prompted uh a few hundred investor relations officers to follow up with us and say, how do we vet our communications ahead of time and make sure that we're actually getting scored the way we want to be scored, which is a really interesting InfoSec problem, too, because we then were all of a sudden dealing with material non-public information because we were getting communications in advance.
- 7:18SPEAKER_01: So we actually had to create an edge-based computing version.
- 7:21SPEAKER_01: Break out the legal bills.
- 7:22SPEAKER_01: No, thankfully, with technology solution was edge-based computing.
- 7:27SPEAKER_01: We never let those communications hit our systems.
- 7:29SPEAKER_01: Oh, nice.
- 7:30SPEAKER_01: They never actually hit our servers because we really never wanted to handle that.
- 7:33SPEAKER_00: You know, it's so wild to me that you developed this pre like AI being a like mainstream, because it's like it's not it's not a like a general purpose LLM, but it is, you know, intelligence, right?
- 7:45SPEAKER_00: I would say that that's pretty like predictive.
- 7:47SPEAKER_00: I mean, predictive analytics, I don't know what you want to call it, but it's definitely in the in a related field and super interesting.
- 7:53SPEAKER_01: And in fact, the the now famous Google Transformers paper that have that came out in late 2018, um, we took that that paper and ran with it and actually built out a large language model of our own to actually just solve a named entity recognition problem.
- 8:09SPEAKER_01: And so, you know, late 2018, early 2019, we stood up our own LLM long before they were commonplace just to solve a very specific tech problem about identifying named entities.
- 8:21SPEAKER_00: So, how is what you learned building Prattle kind of influencing the work that you're doing as head of Fidelity Labs?
- 8:29SPEAKER_00: And also I'm curious like about your journey to Fidelity Labs, like why, and I know we talked about this a little bit more uh at the beginning, but you know, going from being an entrepreneur and you're still an entrepreneur, but you're inside of a big corporation, like what was the decision making like that like for you going into that role?
- 8:47SPEAKER_00: I know I asked like a lot of questions there, so just pick which one you want to start with.
- 8:51SPEAKER_01: Yeah, let me let me start with uh look, the a lot of what I learned about uh at building Praddle was about all of the things I did wrong along the way, right?
- 9:00SPEAKER_01: Um, one of the things I I like to point out to entrepreneurs is every company I've ever been involved with eventually gets product and engineering right.
- 9:09SPEAKER_01: Right.
- 9:09SPEAKER_01: The best companies do it really efficiently.
- 9:11SPEAKER_01: Most of us take a little longer, myself included.
- 9:14SPEAKER_01: But really, where most companies stumble or struggle is distribution.
- 9:18SPEAKER_01: And I really didn't understand that.
- 9:20SPEAKER_01: Coming out of academia, uh, I had this whole product-oriented mentality.
- 9:24SPEAKER_01: I was very focused on the idea that I had built the perfect mousetrap, right?
- 9:29SPEAKER_01: The perfect solution.
- 9:30SPEAKER_01: And one, there's no such thing as perfection.
- 9:32SPEAKER_01: And two, perhaps much more importantly, I didn't understand my market dynamics.
- 9:36SPEAKER_01: I didn't understand market sizing, pricing, my clientele, who the target customer was versus versus who the user was, right?
- 9:43SPEAKER_01: So who's the buyer versus who's the user is a really interesting distinction.
- 9:47SPEAKER_01: So I was a, you know, if I build it, they will come entrepreneur.
- 9:51SPEAKER_01: And that's that's the wrong way to build a business.
- 9:54SPEAKER_01: And it took me probably two years to figure that out.
- 9:56SPEAKER_01: And I was very fortunate that I got enough early traction, had a little bit of funding to survive that period.
- 10:01SPEAKER_01: But there were a lot of mistakes made along the way.
- 10:04SPEAKER_01: And frankly, you know, went through the process of acquiring another company, getting acquired, um, dealing with the acquisition integration, and then ultimately leaving the company that acquired us, and then going off and going into consulting, helping stand up several other businesses along the way, working with dozens of early and growth stage companies, and just getting more and more pattern recognition over that period between really 2020 to 2025, really understanding the details of what works and what doesn't consistently across businesses.
- 10:35SPEAKER_01: And a lot of that was just understanding market dynamics, understanding what is the customer really asking for, what are they actually willing to pay for, not just building the thing I want or I think is cool or is the best new use case for this technology, but what actually solves that pain point that the customer's living with.
- 10:51SPEAKER_01: And so that was a big part of my sort of lesson in building Pratdle and sort of my takeaway in how I've advised now dozens of other companies along the way, which really led that was what led me to Fidelity.
- 11:04SPEAKER_01: I was running a fractional executive consulting business, uh, Outrigger Group, working with dozens of early and gross stage companies across FinTech data and AI.
- 11:13SPEAKER_01: And I had a lot of pattern recognition, a lot of successful businesses along the way that I helped advise and work with, and got the opportunity to come into Fidelity and do that at frankly infinite scale, right?
- 11:26SPEAKER_01: Um, when you have trillions of dollars in AUA and built-in distribution rails to tens of millions of Americans, we have the opportunity to really solve the hardest problem in building a fintech business, which is distribution.
- 11:41SPEAKER_01: So having a ready-made distribution engine where we get to tap into that core clientele and help people lead better financial lives.
- 11:49SPEAKER_01: That's what really really brought me here and you know, happy to kind of go into detail on what labs is and kind of what we're working on as well, if that's helpful.
- 11:56SPEAKER_00: Yeah, I would love to.
- 11:57SPEAKER_00: So I'm curious, listening to, you know, I obviously you're hugely valuable to Fidelity in the ability to like incubate these new concepts.
- 12:06SPEAKER_00: But do you also still work with external teams through Fidelity Labs?
- 12:09SPEAKER_00: Is it like a is there ever like a partnership model there where you try to like get access to like early stage companies that are doing something interesting?
- 12:17SPEAKER_00: You know, obviously there's an acquisition path down the road for things like that.
- 12:21SPEAKER_00: Short answer is sort of.
- 12:22SPEAKER_01: Fidelity is definitely a build-first ethos.
- 12:25SPEAKER_01: But the last two acquisitions Fidelity did do live in labs.
- 12:29SPEAKER_01: You know, we've got we've got a couple of those businesses, uh, notably Shoebox, which is now Fidelity Private Shares, which is a captable management and 409A evaluation data room document storage business.
- 12:39SPEAKER_01: That was an acquisition by Fidelity a couple of years ago.
- 12:41SPEAKER_01: And so we've been building that business out in labs.
- 12:44SPEAKER_01: And so there is a pathway to that.
- 12:46SPEAKER_01: Certainly, Fidelity, you know, certainly I think has a long history of homegrown technology building things in-house.
- 12:52SPEAKER_01: But as the technology ecosystem has gotten more diverse and as AI has accelerated the pace of innovation, partnering with external firms is just part and parcel of what any large corporation has to do.
- 13:03SPEAKER_01: And I think that we have to think about these things as truly as partnerships, not just as vendor relationships.
- 13:09SPEAKER_01: And that's that's where labs can be really helpful is not just thinking about a partnership in terms of, okay, we can we can ingest your product as it sits today, but we can work with you to co-develop product that might fit into the fidelity ecosystem and actually be distributed to Fidelity Clientele.
- 13:28SPEAKER_00: And what it in your mind, like you know, one of the cool things I think about the position you said is you've got this like broad view of the market.
- 13:34SPEAKER_00: You know, you're and and so I'm curious at the intersection of AI, the agenc web, and fintech, like what is the state of the union?
- 13:43SPEAKER_00: I mean, I'm both interested in terms of markets and adoption and also technology.
- 13:48SPEAKER_01: Yeah.
- 13:49SPEAKER_01: So I, you know, when I think about kind of the the adoption of agentic commerce or or how agents are going to change the way we we interact with one another, I actually think it really changes the atomic unit by which we are transacting.
- 14:05SPEAKER_01: So when we think about things like data transactions, right, selling data has historically only happened on a data set by data set basis.
- 14:14SPEAKER_01: By having an agentic commerce, if an agent's talking to an agent, you can actually go down to the granular level of buying an individual data point, right?
- 14:22SPEAKER_01: When you when you start talking about sort of these microtransactions that I think the cryptocurrency community and digital asset community has really embraced the idea of for a long time, we haven't been able to functionally achieve that because essentially we haven't been able to break down the information into small enough bytes.
- 14:40SPEAKER_01: Now we have that opportunity, and I think agents really facilitate that with these sort of packets of information that can be really, really, really small to the point where having a human being in the loop is unlikely to be necessary and frankly probably a waste.
- 14:57SPEAKER_01: And so that's where I think agentic commerce gets really interesting really quickly.
- 15:01SPEAKER_01: That's not the only use case, but I think that as we look at sort of where does the world go from here, it's it's paying for the things you actually want and need.
- 15:11SPEAKER_01: And and agent-to-agent communication allows for that to happen in a seamless way.
- 15:16SPEAKER_00: Yeah, I agree that's where it's going.
- 15:18SPEAKER_00: I mean, we started uh so I had Namil from Coinbase on my podcast.
- 15:22SPEAKER_00: He was the head of developer tooling there back in May over a year ago.
- 15:27SPEAKER_00: And he told me about X402.
- 15:29SPEAKER_00: And like to make a really long story short, we ended up building a framework for agentic commerce called Fairmeter, which is like uh agnostic in terms of standards and in terms of network that it operates on.
- 15:41SPEAKER_00: So it supports X402 and MPP.
- 15:43SPEAKER_00: And we ran that for over we ran that in production for a year.
- 15:47SPEAKER_00: We're like one of the first teams to market, and um, it did a lot of volume.
- 15:51SPEAKER_00: Like we were probably 60%, 50, 60% of the market, mostly on Solana, but only on the open source side, like a lot of people installing the client, a lot of people like running the facilitator themselves.
- 16:02SPEAKER_00: And a lot of this stuff we were seeing that we could monetize just wasn't big enough, and or it was like very speculative, you know.
- 16:09SPEAKER_00: And I do think the data play is like one area where there's gonna be something real there.
- 16:14SPEAKER_00: We're just not there quite yet.
- 16:15SPEAKER_00: Like I'm not sure if it's 12 months away or if it's five years away.
- 16:20SPEAKER_00: I can't really tell.
- 16:21SPEAKER_00: And so, you know, we ended up building that's still open source and being used by a bunch of people, but we're now focused on an agentic operating system at our company that's open source for like guardrails, you know, permissions, like identity, that kind of stuff, because that's the kind of thing that I think needs people need like today.
- 16:38SPEAKER_00: But I I am very fascinated by that market, and I think that in some near-term thing, like five years from now, it's gonna be you know, billions of agents transacting at the speed of light, and the market cap, like the volumes are gonna be huge.
- 16:51SPEAKER_00: Like it could, it's gonna go parabolic at some point, but we're still like in that it's like when you first heard about Bitcoin or something, right?
- 16:56SPEAKER_00: Like it's like no one knows about it still, and there's not really a use case yet.
- 17:00SPEAKER_01: I also think you you highlight the question of when does that happen, right?
- 17:05SPEAKER_01: I I don't think it's an if.
- 17:06SPEAKER_01: I think it's a when do we see more age more agent-agent commerce?
- 17:11SPEAKER_01: And what's gonna enable that is really database architecture, right?
- 17:14SPEAKER_01: Yeah.
- 17:15SPEAKER_01: Right now, what's happened is most organizations are operating on legacy or uh organizational systems, l legacy architecture.
- 17:23SPEAKER_01: Frankly, most of that data is commingled in ways that it probably can't be atomized, right?
- 17:27SPEAKER_01: Most of those transactions are happening in large packets of information, right?
- 17:31SPEAKER_01: We're still only a decade removed from everything being being transferred via FTP.
- 17:36SPEAKER_01: So, like, you know, yes, APIs were a step in the right direction, but like, you know, MCP to MCP transactions is like that's a brand new paradigm.
- 17:45SPEAKER_01: And it's gonna take a while for the data infrastructure that underpins that to catch up.
- 17:50SPEAKER_00: You know, a market opportunity I actually think somebody could move on and do really well is real estate data.
- 17:56SPEAKER_00: I so I was just hacking the other day.
- 17:58SPEAKER_00: I've been getting into multifamily real estate investment, and I built an agent team to like scour the internet and do build a pipeline, do the analysis, get me my cash and cash return, all that stuff.
- 18:07SPEAKER_00: And I have a friend who's an agent.
- 18:08SPEAKER_00: I got them to get like an actual real estate agent, not an AI agent.
- 18:12SPEAKER_00: Um, and she gave me, I got access to MLS keys that way.
- 18:16SPEAKER_00: And because you can get them as long as you're not reselling it.
- 18:19SPEAKER_00: And so the agents were trying to ingest this data by just reading the API, and they really couldn't.
- 18:25SPEAKER_00: Like the JSON was formatted in an unexpected way, and it like, you know, it kept messing up.
- 18:30SPEAKER_00: It was like there's no address.
- 18:32SPEAKER_00: And I was like, look again.
- 18:33SPEAKER_00: And so I think there's like a, you know, even just I don't think the MLSs are gonna add MCPs, you know, in front of their APIs.
- 18:41SPEAKER_00: Like, I don't think that's happening anytime soon.
- 18:43SPEAKER_00: I think there's somebody else can like some other engineer or somebody's gonna get really creative on how they aggregate the data and organize it in an agent-first manner, and then they can paywall it, you know, however they want.
- 18:55SPEAKER_00: And I think that's where microtransactions, pay per call, pay, not even pay-per-call, but pay per bite, you know, make make a lot of sense.
- 19:02SPEAKER_01: Yeah, and I think and I think like that's a great great example.
- 19:05SPEAKER_01: Like what, you know, what what's an older, more you know, established industry than literally the ownership of property, right?
- 19:12SPEAKER_01: Yeah, like the data, of course, reflects the fact that it is, you know, it's an older industry.
- 19:19SPEAKER_01: And so until that data is broken down in ways we can actually see AI systems using it, like it's going to take a while for us to get through this transition phase.
- 19:29SPEAKER_00: So when you're talking about institutions, trust an AI agent, how are you thinking about that?
- 19:35SPEAKER_00: Like this is the thing, one of the things you read in the intro that I was like, okay, what does that look like for you?
- 19:41SPEAKER_01: Yeah, and I think, I mean, look, you know, Fidelity's core brand is trust, right?
- 19:45SPEAKER_01: We're extremely careful about our legal risk and compliance framework and making sure that anything we're putting in front of clients is not only regulatory compliant, but has gone through an enormous number of checks to make sure that our systems are stable and reliable.
- 20:00SPEAKER_01: And so, you know, for us, you know, I think we're we're a ways away from these types of agentic transactions occurring.
- 20:07SPEAKER_01: We're a ways away from actually leveraging these technologies at scale.
- 20:12SPEAKER_01: But that's where labs comes in, right?
- 20:14SPEAKER_01: Like that's where we have the opportunity in labs to test and probe and figure out what the future of this industry will look like and build things that are subscale, right?
- 20:24SPEAKER_01: Test them out with pilot products, with with new ideas that we can we can sandbox off and make sure that before we roll them out to tens of millions of clients, we've got this real opportunity to validate that the technology is going to work in every scenario.
- 20:41SPEAKER_00: Yeah.
- 20:42SPEAKER_00: You know what's interesting too is that the you know, a lot of the ideas that are in FairMeter, like we have a payments gateway, right, that supports like streaming payments.
- 20:49SPEAKER_00: These are not new concepts.
- 20:51SPEAKER_00: Like MasterCard and Visa have solved these problems.
- 20:54SPEAKER_00: And you just need to kind of it's a different slightly different architecture and different payment rails, but the ideas are basically the same.
- 21:01SPEAKER_00: And so that's kind of the good the good news, right?
- 21:04SPEAKER_00: Is you can you don't have to reinvent the wheel.
- 21:06SPEAKER_00: The tougher part is the control of the agents and knowing what they're doing and who they're tied to, right?
- 21:13SPEAKER_00: Like that, and that's I think the the place where there's like governance and identity and permissioning systems and treating them like humans with like authorization and access to stuff.
- 21:24SPEAKER_00: Like this is where the market is today, right?
- 21:26SPEAKER_00: There's like I I can't tell you how many executives at from you know Siri seed startups to 5,000-person automotive company I've talked to, they're like, how do we wrangle these things?
- 21:37SPEAKER_00: How do we how do we orchestrate the agents across a company, like agent operations and like H A, it's almost like agent HR and agent security, you know, like how do we deal with that?
- 21:46SPEAKER_00: I'm and I'm I'm curious, like, you know, what once you get those things in place, then you can start to transact, you know, and you start small and they kind of like grow.
- 21:55SPEAKER_00: And I'm I'm wondering how you're thinking about that, what you're exploring in this like operating system.
- 22:00SPEAKER_00: Space because you need those those tools to sit before you can send it to the 10 million users.
- 22:06SPEAKER_01: I think, I mean, the the the crux of your question is identity.
- 22:10SPEAKER_01: Right?
- 22:12SPEAKER_01: What is the identity of the human being that is behind those agents?
- 22:17SPEAKER_01: What is the identity of the agent?
- 22:19SPEAKER_01: Like understanding how that how that data is going to be transacted, how it's going to be stored, what it's going to be used for, right?
- 22:26SPEAKER_01: Ultimately, the hardest problem in technology today is verifying identity.
- 22:32SPEAKER_01: And that's true across all human-based systems, right?
- 22:35SPEAKER_01: Human beings have a lot of variables going on.
- 22:37SPEAKER_01: That's true with agentic systems, identifying what is the identity of that agent, who's the person who programmed it, who's operating it, how's the how's the information being used?
- 22:46SPEAKER_01: These all boil down to identity problems.
- 22:48SPEAKER_01: And ultimately, once we solve for those, right, once the once we figure out the identity of that individual, that agent, that that organization, right, then the internet becomes a much safer place.
- 23:02SPEAKER_01: And I think you know, what's interesting in this moment in time is the extent to which AI systems have exposed vulnerabilities across the entire technology ecosystem.
- 23:13SPEAKER_01: I mean, like that's like, and that all boils down to the fact that our systems have, you know, it's it's a 50-year patchwork of tech, right?
- 23:21SPEAKER_01: It goes all the way back to ARPANET.
- 23:23SPEAKER_01: And so the question becomes like, okay, if we're going to modernize the entire internet, essentially, right, what's the fundamental building block of that?
- 23:33SPEAKER_01: It's identity.
- 23:34SPEAKER_01: It's identifying who is the person who that is powering those agents and what is the network of whether it's five agents or five million, right?
- 23:43SPEAKER_01: Like who's running those and where does the data sit?
- 23:45SPEAKER_01: And that that to me is ultimately, you know, that's that's the InfoSec question of this decade.
- 23:52SPEAKER_00: Are you seeing any solutions to that that you're excited about?
- 23:55SPEAKER_01: I am, yes.
- 23:56SPEAKER_01: And I think there's there's a whole patchwork of solutions.
- 23:58SPEAKER_01: Like this is not, there's no one silver bullet that I've seen.
- 24:02SPEAKER_01: There's a lot of the human-based identity solutions have gotten really sophisticated.
- 24:06SPEAKER_01: I was just playing with a voice-based solution on Friday of last week, just because it's really interesting that technology's come a really long way.
- 24:14SPEAKER_00: It like fingerprints your voice, it's like a signature, and you say, yes, this is me.
- 24:19SPEAKER_01: Correct.
- 24:19SPEAKER_01: And in fact, what's there's sort of ancillary benefits to some of that technology, things like better noise canceling on uh on calls, right?
- 24:26SPEAKER_01: So if you're in like a you know airport, train station, noisy environment, whatever it is, instead of it just doing generalized background cancellation, if it can isolate in only on the tonality of your voice, like there's some interesting uh um benefits to that type of technology.
- 24:39SPEAKER_01: It's not like that.
- 24:40SPEAKER_00: Yeah, like my kid's not hijacking my Alexa.
- 24:43SPEAKER_00: Exactly.
- 24:44SPEAKER_01: There you go.
- 24:44SPEAKER_01: Perfect.
- 24:45SPEAKER_01: That's a great example.
- 24:46SPEAKER_01: And so I think it's one of those interesting things, though.
- 24:48SPEAKER_01: Like you look at the the landscape of uh of what's possible with that type of tech, and yeah, we all hear stories about people's voices being spoofed, but now there's there's now tech technology that takes that one step further and says, you know, you can you can verify certain vocal patterns.
- 25:04SPEAKER_00: And so I I couldn't agree more with you.
- 25:07SPEAKER_00: You said that like that that hits a nerve strikes a nerve for me because I had this guy, I think his account was hacked, so I don't know who it actually was on X, who was impersonating as as if he worked for the index, uh the index studios, and going out and trying to set up meetings and stuff and uh get people on the show, and then he was trying to steal money from them.
- 25:26SPEAKER_00: And luckily, no one that I know of like fell for it, but I was getting like hundreds of DMs.
- 25:30SPEAKER_00: This guy was working so hard.
- 25:31SPEAKER_00: I was like, this is another thing.
- 25:32SPEAKER_00: I'm like, man, this guy, if you just put that energy towards something meaningful, like good, you would do like a lot of good.
- 25:38SPEAKER_00: But like he's just gonna try and scam people all day.
- 25:41SPEAKER_00: But anyway, there's no identity.
- 25:43SPEAKER_00: I'm like, like they there's no, you know, on X, like he he literally posted a selfie of himself in an elevator with me, and he stole my likeness from the show.
- 25:53SPEAKER_00: Like he screenshotted a screen grab of one of my episodes, and so it looks pretty damn real.
- 25:57SPEAKER_00: Like it's it's pretty close.
- 25:59SPEAKER_00: And I think that that's gonna become more and more of a problem, you know.
- 26:02SPEAKER_00: And I talked about this with Ryan Carson.
- 26:04SPEAKER_00: I don't know if you've seen his stuff on X, but he's um the founder of Treehouse, which he exited back in 2021, and then now he's building an a an agentic first um company called Untangled.
- 26:13SPEAKER_00: Really cool story.
- 26:14SPEAKER_00: But he was saying how you know the open AI thing, where the agent like got out of its sandbox and started uh the way it got out of its sandbox was it found a whole, a zero-day vulnerability that gave it access to a messaging board for this specific software.
- 26:29SPEAKER_00: It was like a messaging board about this kind of software, and it started leaving messages, basically asking for help getting internet access.
- 26:36SPEAKER_00: And another agent at OpenAI found it that was not a part of this whole experiment and started helping it and communicating with it.
- 26:44SPEAKER_00: And it did it twice.
- 26:45SPEAKER_00: They figured it out the first time and blocked it, and then the second time it got in there.
- 26:48SPEAKER_00: And so what does this mean?
- 26:49SPEAKER_00: It means that your InfoSec and your opsec have to be at the speed of light, they have to be at the speed of agents, right?
- 26:56SPEAKER_00: Like that's a whole nother thing for institutions is security.
- 26:59SPEAKER_00: Like, how do you in real time combat, you know, agents that are trying to hack a user's account, right?
- 27:05SPEAKER_00: Like when it can happen, you know what I mean?
- 27:06SPEAKER_01: Like, do you have the equivalent of the OFAC list for agents, right?
- 27:10SPEAKER_01: Yeah.
- 27:10SPEAKER_01: Where you know it's a sanctioned agent, right?
- 27:13SPEAKER_01: And I think ultimately that's where we'll land.
- 27:15SPEAKER_01: There's gonna be a lot of a lot of steps between here and there.
- 27:17SPEAKER_01: And that's that's why we have to be so careful about what goes into production.
- 27:21SPEAKER_01: It's also why it's so important to be testing these things in in a truly sandboxed off, high hyped off way to make sure that that we don't expose those vulnerabilities to clients.
- 27:32SPEAKER_00: So are you are you guys building an identity solution at labs or what what are you actually making?
- 27:38SPEAKER_00: Are you just like looking for that as a part of a solution that you want to deliver?
- 27:41SPEAKER_01: We're not building a dedicated identity solution as a standalone.
- 27:46SPEAKER_01: This is part and parcel of building any new business that has to operate in a secure environment these days, is making sure we know who the users are.
- 27:55SPEAKER_01: And so within labs, we've we've got nine businesses, some of which are publicly out there and known and commercialized, some of which are nascent new ideas, things we've been working on internally.
- 28:06SPEAKER_01: And so, you know, we've got a variety of things.
- 28:09SPEAKER_01: Almost all of them involve having to verify who the client is, how they're touching these systems.
- 28:15SPEAKER_01: And so, in in some sense, everything we build touches the identity question.
- 28:20SPEAKER_00: Oh, very cool.
- 28:21SPEAKER_00: Got it.
- 28:21SPEAKER_00: Um, is there any one of these public ones that you want to talk about?
- 28:24SPEAKER_00: Like share with people, like what it the business is?
- 28:27SPEAKER_01: Yeah, I think I mean I think the one that that probably uh bears highlighting in this moment in time is Fidelity Stock Transfer, which the first fully modern transfer agent launched in over three decades.
- 28:38SPEAKER_01: Oh, cool.
- 28:39SPEAKER_01: The industry standard on this for for you know decades has still been paper stock certificates.
- 28:45SPEAKER_01: And so we built a fully modern transfer agent that operates fully electronically, seamless record keeping, and and much more efficient way of essentially when a company goes public, they have an IPO.
- 28:58SPEAKER_01: Think about how frustrating it is to wait two weeks for your paper stock certificates to show up before you actually like have any verification that your net worth is what you think it is, right?
- 29:07SPEAKER_01: Well, having that instantly when the company goes public, having that fully electronic and dropping directly into your Fidelity account.
- 29:13SPEAKER_01: Yeah, that's what we solve.
- 29:14SPEAKER_00: That's amazing.
- 29:15SPEAKER_00: Is that only for Fidelity, or do you guys let other people use that transfer agent and they and charge them for it or something?
- 29:21SPEAKER_01: Yeah.
- 29:21SPEAKER_01: So any, I mean, any company that is going public and can use our transfer agent.
- 29:26SPEAKER_01: We're building out additional capabilities to work with existing public companies right now.
- 29:31SPEAKER_01: Um, we've processed more IPO volume than any other transfer agent.
- 29:34SPEAKER_00: Cool.
- 29:34SPEAKER_00: That's super cool.
- 29:35SPEAKER_00: I mean, the only reason I know what this is is because I'm a huge fan of another company called Super State.
- 29:39SPEAKER_00: Have you heard of Superstate?
- 29:41SPEAKER_00: I haven't.
- 29:42SPEAKER_00: I think they own a transfer agent, like or they became a transfer agent, and that's how they're able to like bring stocks on chain.
- 29:48SPEAKER_00: But that was like my how I learned I didn't even know what a transfer agent was, you know.
- 29:51SPEAKER_01: And it's an esoteric corner of the market for sure.
- 29:54SPEAKER_00: Yeah, and well, and it's you know, it's funny.
- 29:56SPEAKER_00: I I didn't know that's that's what it was, but I did.
- 29:58SPEAKER_00: I wrote my first angel check when I was like 22 into this company called Outlet Baby Care, and they like ended up IPOing for like a billion dollars, and then they got wrecked on the SPAC.
- 30:07SPEAKER_00: It was a SPAC, so they got wrecked on the and I like round-tripped it.
- 30:10SPEAKER_00: It was great though.
- 30:11SPEAKER_00: And they did they saved like hundreds of babies because they put like a uh O2 sat sock on people's fits.
- 30:17SPEAKER_00: So, like really, really cool product and team and company, and I love those guys.
- 30:22SPEAKER_00: But it took me forever to get my shares, partly because it was like I'm busy and it did, it wasn't important enough, you know, and and partly like it was just hard.
- 30:31SPEAKER_00: Like I had to get this piece of paper and this number, and then the it was like all over email.
- 30:35SPEAKER_00: And I was like, why is that why can't they just put shares in my Schwab account?
- 30:38SPEAKER_00: You know?
- 30:39SPEAKER_00: So I have definitely felt this pain of the old way of doing it, and it sounds like a cool product.
- 30:44SPEAKER_01: Yeah, it's I mean, so it's we've you know, we we solved for this thinking uh honestly, like this just shouldn't be a pain point in the industry anymore.
- 30:52SPEAKER_01: It's it's ridiculous that that we're still operating on outdated technology.
- 30:56SPEAKER_01: Um, turns out that really resonated with the market.
- 30:58SPEAKER_01: I mean, we we processed our first IPO less than a year ago.
- 31:01SPEAKER_01: And as I said, we know we're already the number one transfer agent in the world this year in terms of IPO volume and number of IPOs and dollar volume.
- 31:09SPEAKER_01: And so, you know, I think we're we're very we're very keen to grow future products like that where it's an industry utility, it's an inefficiency that shouldn't exist.
- 31:19SPEAKER_01: And it makes it one of these things where it just makes clients' lives easier when a company goes public.
- 31:25SPEAKER_01: You know, you the you have the wonderful experience of being an angel investor.
- 31:27SPEAKER_01: I'm gonna remember that.
- 31:28SPEAKER_00: I'm gonna go open a fidelity account right now if I don't already have one.
- 31:31SPEAKER_00: I've got to find out.
- 31:32SPEAKER_00: Talk to me, call my financial advisor, you know.
- 31:35SPEAKER_00: That's great.
- 31:36SPEAKER_01: So no, it's so yeah, that's I mean that and that's the that's probably emblematic of the kinds of things we like to build in labs, right?
- 31:42SPEAKER_01: Where it's an industry utility, it taps into Fidelity's existing distribution rails, and it then it improves a client's life, right?
- 31:49SPEAKER_01: It's it's just really eliminates friction and gives them access to their wealth more readily.
- 31:54SPEAKER_00: So we're we're at the top of the show here, and I always ask this question, but what have I not asked you that I should have asked?
- 32:00SPEAKER_01: What's the future of FinTech?
- 32:02SPEAKER_00: I I'll let yeah, fill me in.
- 32:04SPEAKER_01: It's my favorite topic to opine on is where is the world going from here?
- 32:09SPEAKER_01: And and I think there's a couple of different angles uh on that.
- 32:12SPEAKER_01: One you've alluded to a couple of times, which is I do think the digital asset space is going to underpin more and more of our infrastructure.
- 32:19SPEAKER_01: And I think you've seen Fidelity make enormous investments in the digital asset ecosystem, the crypto ecosystem now for years, because we've we firmly believe that it's a huge part of how the financial services world is gonna operate for the for the the next generation at least.
- 32:35SPEAKER_01: Right.
- 32:36SPEAKER_01: The second piece of that is I actually think that there's a we've seen this rise in alts, alternative investments, over the last real dec, really the last decade.
- 32:45SPEAKER_01: Yeah.
- 32:45SPEAKER_01: And I think that we're gonna see a convergence between those two things.
- 32:49SPEAKER_01: That as alts and crypto start to come together and some of the crypto infrastructure facilitates alts transactions and meta record keeping, we'll actually see those things more readily available to investors as well.
- 33:02SPEAKER_01: And we'll actually probably have an environment where maybe maybe alternative investments, maybe private investments, they are they're less opaque than they have been.
- 33:12SPEAKER_01: And so they look a little bit more like public markets, which allows for people to transact them more readily.
- 33:17SPEAKER_01: It allows for people to have more liquidity and allows the markets to function more efficiently so people can actually utilize their wealth in reasonable ways rather than having it locked up for you know seven, 10, 14 years, whatever the duration.
- 33:32SPEAKER_00: Well, I think it's important too to bring up access, right?
- 33:35SPEAKER_00: Like um, even even for credit investors, like I don't there's tons of opportunities that I just don't see because they're like only in some deck from some fund I'm never gonna talk to, you know, and uh those opportunities could be tokenized and on-chain and like easy to buy and easy to trade.
- 33:52SPEAKER_00: And they can launch just like a you know, any other token and or or whatever, like be issued to the market and have some kind of event, and then they're tradable.
- 33:59SPEAKER_00: And the technology is finally getting to the point, like on Solana, for example, you know, there are token standards that allow you to like only allow a certain subset of wallets to transact.
- 34:09SPEAKER_00: And so those wallets can be like proven KYC against the identity, go back to identity.
- 34:13SPEAKER_00: Like, how do you do that?
- 34:14SPEAKER_00: And then but what's cool, and this is something that Superstate was working on with all the major wallet providers.
- 34:19SPEAKER_00: What's cool about that is when when you know like 100 million wallets are all like approved for this kind of token to be purchased and traded, then it's all composable across DeFi and you can take advantage of all those innovations, right?
- 34:34SPEAKER_00: That that otherwise you wouldn't uh get access to.
- 34:37SPEAKER_00: Another company that I'm really excited about is Ground, which is um actually the former COO and co-founder of um Superstate just left to start Ground.
- 34:45SPEAKER_00: And Ground is a product for institutions who want to leverage, uh who want to be able to get yield from DeFi, but like in a stable way.
- 34:52SPEAKER_00: We're not talking about like gambling degeneracy, but like stable yield on-chain and actually benefiting from the rails, streaming payments.
- 35:00SPEAKER_00: At least that's my understanding.
- 35:02SPEAKER_00: And a lot of this stuff, like the regulatory environment didn't exist, the the rails weren't there, the the ability to like ID a wallet or 100 million wallets as KYC'd and like okay to participate in this system didn't exist.
- 35:15SPEAKER_00: I mean, so much has happened and it's taken us like 10 years to get there.
- 35:18SPEAKER_00: But we're finally like almost over the hump.
- 35:21SPEAKER_00: I don't think we're quite there yet, but we're pretty damn close.
- 35:23SPEAKER_01: I I'd like to point out your last point there.
- 35:25SPEAKER_01: It was about identity, right?
- 35:26SPEAKER_01: Like this it always goes back to the same thing, which is that which is that once you can verify that, you can add there's so many ways for people to transact, access liquidity, really build out in an a, you know, a whole pipeline of new financial products, but it all it all rests on on the ability to confirm that you're transacting with who you think you're transacting with.
- 35:48SPEAKER_00: Well, Evan, this has been super fun.
- 35:50SPEAKER_00: Um, you're welcome back anytime.
- 35:52SPEAKER_00: You guys want to talk about cool stuff that's coming out of labs.
- 35:55SPEAKER_00: I would love to hear about it and share it with people.
- 35:56SPEAKER_00: Uh, thanks for taking the time today.
- 35:58SPEAKER_00: Appreciate you.
- 35:58SPEAKER_00: Awesome.
- 35:59SPEAKER_01: Thank you so much for having me on, Alex.
- 36:00SPEAKER_01: Appreciate it.