Don Gossen Podcast Transcript

167
Don Gossen Podcast Transcript headshot

Don Gossen Podcast Transcript

Don Gossen joins host Brian Thomas on The Digital Executive Podcast.

Brian Thomas: Welcome to the Digital Executive. Today’s guest is Don Gossen. Don Gossen, the visionary co-founder of Nevermined, the PayPal of AI commerce. Nevermined is presenting a new era where autonomous AI agents can conduct transactions entirely without human intervention. They’ve just launched a new collaboration with Olaas, Peak, and Combinder that introduces the first live implementation of autonomous AI payments for smart energy systems.

Well, good afternoon, Don. Welcome to the show.

Don Gossen: Thanks for having me.

Brian Thomas: Awesome. I appreciate you making the time. I know we’ve got kind of a time distance travel between us. You being in Lisbon, Portugal, and I’m in Kansas City, center of the United States. So I know sometimes making times and traversing time zones can be challenging, but I do appreciate the time.

Love doing international podcasts. So Don, let’s jump into your first question here. Reflecting on your entrepreneurial journey from co-founding Ocean Protocol to leading Nevermined. What have been the most rewarding aspects of entrepreneurship and what advice would you offer to entrepreneurs venturing into the AI and blockchain space?

Don Gossen: I mean, I think the most satisfying piece is basically starting from scratch and building something that people find valuable, you know, the proverbial 0 to 1 is quite challenging, but also extremely rewarding if you can accomplish it with respect to venturing into a I. And the cross section with blockchain, I mean, it’s an interesting proposition, right?

Because you’re talking about 2 technologies that are inherently divergent from 1 another, right? Blockchain technology is. Innately decentralized versus AI, which is innately centralized. So trying to merge these two technologies in and of themselves has its own challenges beyond just the technical novelty in particular of, I would call it new age AI tech.

It’s an interesting cross section to be working in and trying to deal with these sort of inherent orthogonalities or divergences that are baked into the technology sets themselves.

Brian Thomas: Thank you. I appreciate you sharing a little bit of that and two things that I highlight out of that your answer There is obviously that zero to one, you know starting something from the ground up There’s a lot of challenges with that.

But looking back when you get to one It’s certainly an accomplishment and something that feels really good But yeah, we talk about AI and blockchain here all the time and in blockchain is something that’s near and dear to my heart We do a lot of that here. In fact, the platform is built on blockchain So we Love talking about this stuff, and I love how you’re trying to bring these two technologies together, so I appreciate that.

Don, Nevermined has been dubbed PayPal of AI commerce. Could you share the inspiration behind this vision and how it has shaped the company’s trajectory?

Don Gossen: Yeah, I mean, it’s a bit cliched, but I would say this is a, it’s a labor of love. So I’ve been in the ML quote unquote AI space for my entire career, so 20 years.

I started in Like the applied side of it, I was in commodities trading on the risk side, so doing statistical modeling on like credit scores and stuff like that for our customers and have just sort of maintained the focus on this space for my entire career. And then in 2016, I caught the crypto bug. And in 2017, as you mentioned, I co-founded a project called Ocean Protocol, which was one of the first projects to make this foray into the AI and Web3 cross section.

And one of the, sort of, primary drivers for this is on the provenance and attribution side. Obviously, blockchains are great provenance machines. They’re very elegant solutions to this problem. Particular type of problem so the ability to sort of track and trace your algorithms and the data that goes along with it is blockchains just facilitate that in a much more elegant manner than other contemporary solutions and so that was that was an inspiration for bringing these 2 technologies together and then there’s been this sort of distillation.

From going on a decade’s worth of work, almost in this space where there was a realization as it relates to AI and more specifically to AI agents, that payments are a real pain point. And so we recognize this pretty early on. I mean, there’s, there’s more projects now talking about AI agent payments, you know, Coinbase has a solution.

Um, there’s a few others in the market on the web 3 side. Stripe. Has entered the fray into this space as well, but there are real challenges with respect to agents paying and getting paid. And more importantly, being able to cover sort of the operational costs of servicing a request. And we found that the infrastructure that’s available via blockchain synced up to the way that an AI.

Is going to operate and then be able to translate that operation and service into a cost for a user. There’s real synergies there. So this focus of becoming the PayPal of AI really involves enabling the metering of the service that’s being provided based on a request from an AI agent and translating that into a settlement cost for the requester.

With our primary focus being. AI to AI transactions, as opposed to like human to AI.

Brian Thomas: Thank you. Appreciate that. And I love having the conversation around this. We talked about this recently on the podcast about how people are now going to be accustomed to having, you know, like a virtual assistant, right?

But it’s AI. And. And AI can take care of those transactions again, in your space, in your business, same exact thing where AI can handle those requests, which I think is amazing. Obviously everybody’s a little bit skeptical about it, but really do love this technology and what you’re all doing out there.

Don, your platform enables autonomous AI agents to conduct transactions without human intervention. What are the key challenges in facilitating these AI to AI interactions and how does Nevermined address them?

Don Gossen: It’s a good question, right? So there’s real nuance to the proposition. So what you’ll see right now, and personally, where I see the skepticism lying, and where I personally have skepticism, is just sort of setting a one price fits all component to these AI.

So I’m going to charge you a flat fee of X for these services. That works now because a lot of these AI projects in particular are well funded, so they’re underwriting the cost of that operation. But as more agentic AI sort of infiltrates day to day life. And takes on a larger role in the overall commercial landscape, there’s going to be a lot more competition and margins are obviously from that going to a road.

And so what you’re, you’re going to need to do is have really accurate accounting or measurement on the actual usage required. To instantiate a response based on a given request and from that accurate metering, be able to translate that over into an accurate price so that you can extract as much of these diminishing margins or returns as possible.

So kind of a general example, just to give your user base sort of a sense of what I’m talking about. You know, right now you go in and you pay a subscription fee of whatever, 20 bucks a month, and you get access to these services as these services become more and more complex and as competition enters the fray and begins to kind of erode the margin that those services are based on, you’re going to need to get more and more accurate on the pricing as an AI agent service provider.

So the example I give is, you know, Brian, you and I, we can use the same agent. Right. Even though we’re completely different user profiles and may have completely divergent or orthogonal requests. Right. And in this case, I’m a simple user. You’re a power user. And we want to use a particular third party agent.

You know, if you’re looking at agentic AI, Right. The understanding here is that these agents are no longer just, they’re not just models, right? Or a single model. It’s a, it’s a compilation of different tools for one, and that represents itself in probably, you know, a composition of different models that that AI agent can call upon based on the request that’s given to it.

Right? So you and I, we can make these divergent requests to these, to these, Agents, the agent can respond to those requests by instantiating different models in response. So the, the simplest example, if you’ve used like 01 or 40 or whatever, is that when I make a simple request to that agent, that request is getting decomposed and rooted to a model that can handle the simplicity of my request.

Let’s say in this case, GPT 3. Right? And that has an associated cost to it. Whereas you make a more complex request, a multimodal request to the same agent. Well, it’s got to invoke a more complicated model in order to handle the multimodality of that request. In this case, maybe GPT 4 or 4. 5 for that matter.

The spread and cost that the cost difference between operating these 2 models can be an order of magnitude or more indifference. And so, as these margins road agents are going to have to have more accurate accountability on that metering. So that they can translate that into more accurate pricing, both from a competitive point of view, but also from a margin standpoint, so that they can extract the most margin out of that transaction as possible, because what we’ll see over time is that these services will effectively get commoditized down to some sort of standard market accepted rate.

And so the ability to translate the metered cost to do what an AI is called, it’s called observability, translate that into accurate pricing and settlement. We’re banking that this is going to be a massive opportunity. And so that’s what we’re addressing.

Brian Thomas: That’s amazing. Thank you for unpacking all that.

That’s very helpful for myself and our audience. Of course, that’s very helpful. You know, you’re in my mind, kind of a pioneer in this space. And I think it’s really exciting time to be in. So thank you for sharing that. Don, last question of the day, transitioning AI driven payments from niche initiatives to mainstream practice requires overcoming certain barriers.

What steps is Nevermined taking to facilitate this shift and encourage broader adoption?

Don Gossen: Yeah, it’s a good question. I mean, this kind of goes back to the first question and the entrepreneurship side. You know, what do you need to do to get this to fly? And a lot of that effort is You know, sort of boots on the ground manual organic work, right?

I think advocating and amplifying the message that we have, right? When we go out and talk to those in the space, whether it’s Web 3 or Web 2, those that are building AI. What we’re building sort of this mission and the overarching narrative really lands with these builders, but a lot of them aren’t thinking about it explicitly at the moment, right?

It’s, um, it’s coming. But part of this we believe is it’s narrative driven. We fundamentally believe that we’re in a stage of categorical definition. So we’re defining a new category. Our label for this category is AI Commerce, and that, you know, these agents are emerging, they’re going to interact with you and I, they’re going to perform functions like assistance, et cetera, but they’re also going to interact with one another, and that these agents are going to emerge, and our belief is we’ll be standalone economic actors.

Right? And so this concept of AI commerce is intended to encapsulate that category and this overarching, I think, evolution of, of effectively e commerce, right? It’s no longer just humans operating and buying things from one another via the internet. It’s also going to be computers. Basically or robots however you want to frame it and so kind of espousing that message and amplifying it.

I think it’s quite critical so that people become aware that they sort of think about it in the back of their heads when they’re building this, but it’s not necessarily front of mind and so bringing that to the forefront that this is, uh, a new opportunity from an economic and financial standpoint.

Both in terms of those that are building these agents now, and then in the future agents to kind of, you know, build themselves that this is sort of the new frontier and emergent. It’s very mid nineties ask, right? That the new agentic economy. Is going to emerge and change commerce forever. And in the same light that the internet did beginning in, in the early nineties through the mid nineties, and then we obviously saw what happened in the 2000s and 2010s with the internet.

So same type of evolution and definitely the same type of impact. So right now, a lot of what we’re doing is educational. Getting the message out and getting people to sort of understand the impact that’s going to be made and accelerate that adoption. So we hit a steeper curve faster.

Brian Thomas: Thank you again. I talked about this just a second ago.

A great time that we’re living in in the adoption is key. You know, you kind of use that analogy the Internet in the nineties and how that has changed things. But the Web 3 adoption and the use of AI is so important these days. Obviously, we got to keep ethics there, but Hopefully we can integrate more DeFi blockchain technology with the AI.

So we can kind of keep things in check as, as I’d like to say. So I appreciate that. Don, it was such a pleasure having you on today and I look forward to speaking with you real soon.

Don Gossen: Thank you for having me. It was great to be on.

Brian Thomas: Bye for now.

Don Gossen Podcast Transcript. Listen to the audio on the guest’s Podcast Page.

Subscribe

* indicates required