Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
Agentic Payments: x402, and the Future of Machine Commerce
Recorded: April 14, 2026 Duration: 1:04:19Space Recording
Full Transcription
Thank you. Thank you. Thank you. Hello, hello everyone.
Hello, hello everyone.
So we are just waiting on a few more speakers to join and a few more folks to join and then
we will be starting.
And in the meantime, feel free to share it with your friends, community, whomever you
want to share it with.
Good morning, everyone. It's Mike.
I'm Slun, the Developer Relations at Quicknode.
Sahil, do you want to introduce yourself?
Yeah, for sure. Thanks, Mike.
So as Mike looks after Solana, I look after all the other chains at Quicknode.
We work on the DevRel team at Quicknode.
Our main day-to-day job looks like where we create educational content and
to enable developers who uses quicknode and to teach them how to use quicknode how to use
all the products in a better and efficient manner and a of times, it's not just limited to that.
Or I would say most of the times,
it's not just limited to that.
We do create a lot of general blockchain content as well
and tech content as well,
basically to get as many people onboarded
to blockchain tech as possible.
Yeah, it's a really fun role because we kind of do a lot of
education, we build things, a lot of people sometimes seem to
think that developer relations people just talk and we don't
code. But most of most of our time, we're building things to
show off to people. So we do a lot of coding. And often we're
trying to research things that are really at the edge of what's possible
so I think just recently
obviously X402 has been
a really big
interest for us
It has been
and I think
it's just
still growing and it's just going to
grow a lot more from here from
all the developments going on and the accessibility of creating agents like it's not just limited to
developers right now even non-non-programmers or non-tech people, even they are creating agents with all the tools that are so
much accessible to everyone. It's mind boggling right now. Yeah, there's a site called x402scan,
which is I think it's literally x402scan.xyz and you can, or x402scan.com, I stand corrected,
and you can actually see what people are building. If you actually visit that and just sort by transaction volume,
you get an idea of the things that people are selling access to via x402.
Volumes are still pretty small right now,
but I think we're in this world now where it is increasingly becoming an agentic world.
where it is increasingly becoming an agentic world.
And really those older payment platforms struggle to do small payments.
Just the overhead of a typical kind of visa transaction on Stripe
means that you can't really do a payment of anything less than,
some people will say $5.
$5 it will eat significantly into your revenue, 30 cents. really do a payment of anything less than some people will say five dollars some people five
dollars it will eat significantly into your revenue uh 30 cents it'll eat all of your revenue
um so there's when you want to sell information in a really granular little way you can't really do
that all on these older platforms so that's why we're really excited about X402. We're also really excited about MPP.
And we're gonna be talking about that today as well.
All righty.
I think we have all the speakers here with us.
So I think we are good to start.
So today's space is going to be all about agentic payments
where we will be focusing a lot on x402 and what we have
for store in the future as well for machine commerce so for those of you who do not really
know what x402 in general is let me give you a very brief introduction on x402 itself so x402 has been a http status code for a really long time
and it wasn't really used until now until x402 was launched where whenever a system or a machine or an agent or a client whoever sends a request to a
service provider or a server they get a x402 status code and that indicates that a payment
is required and they can pay for that particular service and get access to those resources or
services you can think of it like a scenario where something is behind a
firewall in your normal general web scenario where something is behind a firewall and for
services or clients who does not have access to that firewall they make requests to that server
and they get a 404 response. Similarly, in this case,
you can have something behind a paywall. And if once you try to get access or request those
services or resources, you get a 402 response, which says payment required, which will also have
details around what you need to pay, how you need to pay, how much you need to pay.
And then you send the request again with the payment done and the proof of payment.
And then the other side or the server side, it verifies the payment and sends you the resources or gives you access to the resources or services.
So that's the general high level idea of x402
and how the x402 payment protocol works so and it has been growing a lot and what i mean a lot
is that just just within a year since it was launched we have seen more than 150 million plus transactions go through this
x402 protocol and the ecosystem itself has grown in more than 100 plus projects that are either
integrating on x402 or accepting x402 protocols and that includes a lot of biggest names in the entire infrastructure
internet infrastructure industry not just crypto but like the wider general entire infra space
so today we have the people who are behind this protocol as well and people who are creating some crucial
true links behind this as well as the people who are deploying capital behind these protocols and
projects who are working on these protocols so i would open up the stage for our speakers who are joining us. And I think we will start with Eric Yu.
So Eric is the head of engineering at Coinbase developer platform.
And he and his team are behind Xphoto2 and Xphoto2.org.
So Eric, if you want to give a brief introduction about yourself.
Hey, excited to be here, guys.
Yeah, I'm Eric.
I lead engineering for Quimus Developer Platform, and I'm one of the authors of X402.
Yeah, we don't run X402.org anymore, as X402 recently moved into a foundation model, and
the foundation runs X402.org.
Other than that, you nailed it.
Yes, sir.
And I think we can move
to Shafu next.
Shafu is the founding engineer
at Merit Systems
and the creator
of Agent Cash.
Shafu, you want to give a brief introduction about yourself?
Yeah, sure. Shafu here, founding engineer at Merit.
We are the creators of x402scan.com, mppscan.com, and Agent Cash,
which gives any agent a wallet to interact with the exciting X402 in MPP world.
So yeah, thanks for having me.
Nice.
And Neymil, if you want to give a really brief introduction about yourself,
Neymil is the visiting partner at Y Combinator.
So Neymil, the stage is yours.
It's awesome to be here.
Y Combinator has invested in lots of crypto companies over the years, including Coinbase, Axiom, and QuickNode.
So, really excited to be here for that.
And we're really excited about just the future of stablecoins, egenic commerce, tokenization.
So, excited to share a little bit more there.
And some of my claims to fame were I've previously led developer products at Coinbase,
and I also helped grow USDC as the lead PM for that to about a billion dollars working with the Circle team.
Nimil's being polite and is saying that he used to be my boss.
And also had a major, major role in X4 or 2 being created.
It's exciting to have different contacts that work.
Thank you so much for joining.
We are excited to have you as well.
And Neville mentioned Circle, so that's a very good segue towards Gagan.
So Gagan is a vice president at Circle.
He manages product management.
So Gagan, if you want to give a brief introduction about yourself.
Hi, everyone. I'm Gagan Matt.
I lead the developer products team at Circle.
That includes ARC, USDC expansion, ARC interop stack,
and some of our AI initiatives.
Thank you for the invitation.
It's awesome to be here.
Thank you so much, Gagan.
Alrighty, so if everyone is all right, we can begin with a few of the talking points and questions.
In general, the entire space is going to be free flowing.
And if anyone has any question from the audience towards the end, feel free to raise your hands as well.
But right now, I have a question for you eric so since you wrote the x402 to
be the http of the money and in a year we have seen more than 150 million plus transactions so
what's that one thing you think is the most surprising about how the ecosystem has developed or evolved in general?
That's a good question.
Yeah, that's a good question.
I think that...
I didn't think it would happen this quickly.
I thought there was a clear need for something like X402,
but I didn't think that there would be so much attention on it so quickly.
And so X402 for, we like, I wrote it in like February of last year,
and then we kind of like applied some polish,
and it took about a month or two to get it onto mainnet.
And then, yeah, it's been out for a bit over a year
and we put the white paper out a year ago.
And it had this like really nice like 30% week over week
compounding growth chart for a long time.
And then, you know, we had a bunch of inflection points.
Some, you know, I got a bunch of inflection points. Some, you know,
I got to give credit to X402 Scan
for really being like the thing
that let people kind of see
what was happening on X402
in the early days.
And then I think that that's like,
that was kind of one set of big moments.
And what that did is kind of like
draw attention from everywhere
where they're like,
oh, there's this this thing it's moving money
agents are using it and it's doing like
many many transactions like we should all apply attention now
and so I didn't think that the like spike
and then sustained kind of attention more so
than transaction volume would happen quite as quickly
as it did yeah true and i think the the trend or the explosion of the general agentic space as well
was obviously the catalyst behind this where obviously a lot of people were already using
these large language models for a long time now but But I think in the past few months, we have seen everyone just move from large language models to agents.
And the language model has just taken the back seat and does not matter right now what language model you are using, etc.
So, yeah, I mean, it has really exploded from there.
And I think the next question will be around the USDC
because that is what settles a vast majority of protocol payments on X402.
And as agents start transacting autonomously at high frequency right so what does
that means for how stable coins are used or adopted is circle thinking about this or are
there any developments around this towards infrastructure readiness to develop for scale
towards infrastructure readiness to develop for scale because right now people are still trying
to get hang of agents and agentic payments or two links but once we start seeing this at a very
massive scale i think the scalability will be a very crucial thing over here so gagan maybe you
can answer this yeah absolutely um yeah it's the first of all, I want to start by saying
it's very exciting to see the protocol's growth.
And I think its beauty is in its simplicity.
And it is very lightweight to integrate.
So no wonder we are seeing 30% a week over week growth.
That's fantastic.
I think it also like solves
a very fundamental problem
that secretly like all product managers
do not like to solve,
which is like, okay,
getting people to onboard
and regardless of what you try,
every company, every API developer
spends time in reducing onboarding friction.
So I like the,
when the protocol first came out,
for me the most exciting thing was like,
ah, you can actually, like, you know,
true permissionless utility can be actually developed
and agents just supercharge it.
And so we are, so that's something like personally
which is very exciting for me.
In terms of, like, the infrastructure,
yes, like, for scale, you have to take a full stack
look at the infrastructure all the way from bare metal settlement network, right?
And so there, I think there's a lot of room for us to scale for more throughput and continue to reduce costs.
Like if you think about what is the lowest priced service that today is used in compute?
priced service that today is used in compute.
For instance, like for an S3 storage bucket,
it costs you like two cents per GB of storage.
Or when an agent is transacting autonomously
and consuming compute,
your AWS computer is somewhere around like,
I would say like half a cent for thousand plus put requests.
So this is like a sub-mini-cent economics. So we constantly think about like all the way
from the settlement layer up, how do you reduce the economics so that you can actually have
a productive activity supercharged? So it starts there. The second thing we think about is like,
how do you ensure that, you know, USDC is a programmable dollar, and it's not a surprise, actually, that it is being used.
Like, I think 98% of the stablecoin transactions
done by agents are done in USDC,
and it's because programmability truly works,
and it is ubiquitous across 30-plus networks.
So the second step we take is, like, okay,
how do we ensure that any new ecosystem that comes up,
we are continuing to support it,
ensuring that USDC is available.
It retains the programmability aspect of it
through an interoperability across different networks.
So that is like our entire interop stack,
which also like enabled nano payments,
is something that is like,
it's a scaling thing for reducing costs
and ensuring that X-Forward 2 can truly scale across as many networks as possible.
And then at the top of it, I think like that is something that is top of the stack, which everyone is already working on is like enabling agents with CLIs and skills and such.
I think the missing piece is also about identity.
That's something which we are starting to think of.
How do you create identity and reputation
that, again, is something that agents can carry across networks?
So it truly is like a full-stack infrastructure problem
that I think is very exciting to solve.
And I'm sure a lot of folks on the call
are thinking about these things as well.
Yeah, interesting how you mentioned, Gagan uh the reputation part because i think
moving forward reputation is going to be very crucial in these agent to agent payments as well
where based on the reputation we would see i mean we are still seeing but we will see a lot more
agents hiring agents for different kinds of things as well as the
negotiating powers and the negotiating dynamics would definitely be very much different based on
the reputations um and i think um on that so world recently shipped this agent kit which is which is based or solves this reputation
part kind of a bit and it combines payments with the proof of human identity with the
agentic payment so eric you have said this in the past that payments are how and identity is who.
So what does the full stack for trust looks like when agents are transacting autonomously?
So maybe Eric, you can take this shuffle.
You can answer this as well.
Yeah, I think identity is like the next big thing for everyone to just get on the same page about.
And I think part of that is, like, due to...
Like, one of the unintuitive things about X402 and stablecoins is,
for, like, people who aren't used to accepting stablecoins,
is that stablecoins invert the risk model of most commerce that happens on
the internet. Like if you're doing a traditional e-commerce transaction, like you as a merchant
have like fraud risk and that like someone could, if they pay you with a credit card,
they can do a chargeback. And so you like assume that you won't need to collect all this information
about the person because you need to be able to feed a risk model that says,
yeah, this person's probably not going to perform a chargeback.
They're not going to be a fraudulent user.
Stablecoins actually invert that risk dynamic
where now actually because stablecoin transactions are irreversible,
you as the spender need to trust that the merchant's going to offer you
the right set of services.
And so you actually need bidirectional identity
where you as the merchant still probably want to know some stuff about the user
but then you as the buyer actually want to know some stuff about the merchant.
And so there's a bunch of initiatives around this like 804
but it kind of breaks down into two components
one being identity and the other being reputation
tied to that identity.
And then I think there's like a third kind of pillar
of discoverability, where can you actually like go
and find the stuff that's interesting to go and pay?
And so I think like to maybe like summarize
what I said more succinctly,
like I think there's three pillars that you need
in order to have kind of like
the full agentic commerce trust stack beyond payments.
It's one, identity for both the merchant and the agent.
Two, reputation for both the merchant and the agent.
And then three, discoverability of the services
that are offered.
Yeah, I think discoverability is another big piece
where it's not like your general normal or wider internet
where you can search for something and just find that service.
So I think agent cache solves that part by quite a lot.
So Shafu, do you want to speak a little bit about
agent cash and how it gives one balance to the agents to access different kinds of api via x402
and what are agents actually spending on money today like what's that one kind of service or what's that one vertical of services
agents are paying a lot for?
And like basically what's that one vertical
which is growing faster than everybody else?
Yeah, for sure.
So what we're seeing right now,
which is super exciting,
is actually non-technical BD marketing sales
kind of people that use agent cash all day for things like lead generation.
So let's say I'm a BD guy at QuickNode, for example.
I would be running like, get me a thousand potential companies
that I could sell QuickNode into, get me their key decision maker,
their LinkedIn and their email.
And because you have the LinkedIn API, because you have the Apollo API,
because you have all these other APIs,
it makes it extremely easy to access all of that.
Whereas before X402, you would need a subscription for LinkedIn,
you would need an API key for Apollo, and all that kind of stuff.
But now you open up
Cloud Code or whatever client you're using and now you have access to all of these APIs
that build on top of each other. So we see a big potential for these kind of people. They love that.
Yeah, yeah, 100%. And on agents being able to discover and using different kinds of services
so nemal you have been very vocal about x402 being the future of agentic payments and even the
wider e-commerce not just crypto so what do you think that actually looks like in practice? Is it agents just transacting with traditional merchants or even interacting with, let's say, even platforms like Shopify and things like Amazon of those sorts?
So I would say, like, first of all, there's all the things that we do already, but built with agents as an intermediary layer.
So a simple example might be today I go to Amazon, I search for something, I go purchase it.
In the future, before I went to Amazon, I actually might have gone and did a Google search to see what the different options available were.
It's a good recommendation from something like Wirecutter or something else that gives me advice on what to buy.
This future world is I'm going to my AI agent,
I'm asking it, hey, I want to get a pair of headphones,
and it's giving me some options,
I'm telling it's my budget, et cetera.
And then I'm going ahead and placing that order
through that agent.
So that's what I call a version of agent intermediation.
And in something like that,
I think that what the challenges today are really,
it's not about that there's not a payment rail. There actually is. It's credit cards and it's
debit cards, especially in many developed markets. But the challenge is, how do you
programmatically go through a flow like that that makes it very easy for an AI agent to do?
And by the way, I can take a more expanded version of that, is that buying one thing,
I'm not sure AI is help, but a more
game-changing example might be something like I'm trying to book a vacation. There's a bunch
of different things I need to do, and there's five merchants actually I need to go and buy from.
That's something that AI agents and agendic commerce can make very, very easy using credit
cards, debit cards, et cetera. And that's awesome because X4.2 supports that.
On the other side, whenever there's this new technology, in this case, it's stable coins,
it's crypto, what is interesting is that there's brand new things that can be done that were
never done before.
And so there's this famous story from when the Apple music model launched, they basically
said, dollar, they're going to allow you to be able to buy a song.
And in that world, like the credit card fees were way too expensive if you just bought
one.
So they got lucky because people bought multiple of them and they batched them together and
they were able to do just one call to the credit card rails.
But in this future world, we have things that could be a few cents, it could be a few dollars,
and it could be anywhere in the world.
And it might be a merchant that I just hit for one moment in time. And I think Cloudflow is maybe
a really great example of this, where they're seeing that AI agents are hitting websites,
and they're not looking at ads. And so the model for monetizing for a website owner is really
breaking apart. And in that world, this is something where you can basically pay 5 cents,
10 cents, 50 cents for an AI agent. For a human to be able to do 55 cents transactions, I think is probably too
fatiguing. But for an AI agent, when it's searching for something that is much, much easier. And so I
think the way I really frame it is that there's all this stuff that happens already. It has some
agentic layer as part of it. And there are great Rails, but we need to make them programmatically discoverable.
And on the second side,
there's all this brand new behavior
that we want to be able to do
that AI agents can't do
because existing Rails are not sufficient for that.
And that's really a magical part
of stablecoins and crypto.
Yeah.
100%.
And since we see that stable coins are there on practically all the
chains right now and most of these chains are use case specific right so maybe gagan this is for you
do you see these payments converging to one dominant chain per use case or are we seeing a more of a
multi-chain multi-chain use cases here and is there a future where you see these transactions
being multi-chain where the where the settlement layer is just abstracted out of it and the agents does
not even need to hold usdc on particular chains they can just hold it on any chain and the
settlement can happen on one abstracted layer yeah we um i mean like there is i haven't seen
an evidence of convergence to like one chain yet. There has been hypotheses on,
okay, a chain is getting to a million transactions per second
or a billion transactions per second.
Whoever says that.
I think right now we are still in a world where economically
every single application that has distribution
tries to go and create their own network, own blockchain.
And why shouldn't they? It just makes sense to own the settlement their own network on blockchain. And why shouldn't they, right?
Like, it just makes sense to own the settlement network
if you have distribution.
All exchanges have, pretty much all exchanges of relevance
have tried to have their own networks.
And so I think we are at least planning for a world
where there is continuing to be like innovation at the layer ones.
And there is going to be like a need for like the volumes that we are envisioning right now,
you can squint.
Like I think when we talk about like 30% record growth and we are still in the very early phases.
And I would say like 99.9% of the API providers are not like
X402. They don't have X402 endpoints, so they're not enabled yet. And so you can imagine that
there will be a future where you need like massive throughput at very low cost and
the workloads are distributed across n number of networks. So I do think like it's going to be
more than a dozen networks at the very least.
I don't see convergence to one, at least from where we stand today.
There's one more thing. I think a couple of very interesting points for me that got me thinking.
Eric's point on the reversal of the risk model then, Nemo, what you were describing on different use cases,
I think those actually open up very interesting topics as well,
because reversal of the risk model means, like,
now it's really important for every single endpoint provider
as well as agent to have a reputation so that the user, so that the risk is now the users, the consumers.
And then in terms of use cases, I think what's very interesting is that, yes, there's like commerce and shopping use cases,
but where crypto leads the way is you have open endpoints like Hyperliquid and Polymarket.
you have open endpoints like Hyperliquid and Polymarket,
and it's easier for people to actually give the agent some money
and say, make me more money versus spend money without discretion.
So I actually am looking at all use cases
where you can auto-rebalance into vaults
or you can actually go and set triggers
wherever there's orchestration complexity
or wherever there is non-determinism
in when you want to take an action,
there I think you'll see the fastest utility.
And that can happen on any number of networks.
Yeah.
And you mentioned that we need toolings
and we need the toolings to increase the throughput.
And I agree on that.
And the thing is that we need standardized toolings to work with the underlying standard and protocol across the
chains right but i have seen that there is a different x402 for i mean not the protocol itself
but two links for solana where which differs a little bit from evm chain so mike
maybe you can answer this that how does the x402 toolings or kits or sdks differ on solana compared
to evm from the developers experience point of view integrations point of view, integrations point of view, and what should developers or builders keep in mind or should know?
I mean, my focus is entirely on Solana.
I'm not actually too aware of the EVM difference.
I'm very Solana pure in this case.
We do provide slightly higher level tools.
This is we being Quicknode for developers who want to connect to Quicknode via
Solana that just give you all one line connection with AgingKit so you don't need to have an account.
In terms of people who want to accept X402 payments on their API, it works very similar
to Solana as any other chain. You have a route with some middleware, a token, an address, a network, and you specify whether
you want an exact payment or whether you want to negotiate effectively.
That part of it is relatively similar, as far as I understand, on Solana to Ethereum
or EDM.
But if someone else wants to jump in,
then they are more than welcome to.
Yeah.
I see Eric, your hands up.
Do you want to add something?
Yeah, I think this is like another thing people kind of miss with X402 is
you don't have to choose.
Like you can accept X402.
You can in one X402 response say like
you can pay me on N different chains.
You can say like I can pay me on n different chains you can say like I accept
USDC on base and USDT on Solana and USDC on Solana and you know USDC on polygon and USDT on
you know Aptos and you know you can you can have all of those things kind of like
live next to each other in the same response from a server and then the the uh buyer just decides like which chain do they want to perform payment on
and like one analogy i make often for x402 is it's it's kind of like the like little placard when you
go to like a cat like a cash register you know if you're at a restaurant and you go pay
or you're at like a grocery store
and they have like a little like sticker that says,
we accept Visa and MasterCard and Amex and cash.
Like that's kind of what X402 is fundamentally.
It's like a way of declaring all the different ways
that you can be paid.
And then the consumer,
similarly to the IRL shopping experience,
just picks the token and the chain that they prefer to pay with.
Understood.
Okay.
And, I mean, we have talked a lot about the growth we have seen in general
for the ecosystem but maybe Neville you
have a more granular view into what founders are building at the intersection of AI agents as well
as crypto right now and what do you think is the strongest opportunity for upcoming founders or
builders to build on top of right now at this intersection? So most of the pitches that I've seen and that we've been getting have been around infrastructure
to make it easier for AI agents to be able to transact. So this is everything from an X-Forto
system, there's facilitators, all the way to metering and billing, liquidity management. These
are all just like pretty critical layers of the infrastructure stack.
And I think that all of this makes a ton of sense.
And you're seeing everything from large companies like Stripe, also YC,
all the way down to the most recent batch,
I think had an Ingenic wallet startup,
Orthogonal and Locus in the batch before that.
But I think the other thing that we're also seeing
that the next category is really about like a specific use case. So they're picking some type
of use case and they're saying, we want to make this easier for people to be able to transact.
Let's go make that, you know, let's go make that possible. And this example, it might be, you know,
I mentioned the idea of travel agents before. It could be, you know, trip planning. It could be,
we've seen examples of like purchase
agents in a corporate setting. But basically, those are the ways that we see the different
pitches. One is in for play. And I think some of the challenges those teams run into on
the infra side is that the market is still incredibly nascent. It's not by default that
everyone in the world today is using their AI agents to be able to buy. And so you really
have to figure out a world where you can ideally, you know, the next
six to 12 months, certain use cases start taking off.
And that's where your infra is going to be heavily used.
And on the vertical side, I think the cool thing there, at least, is people are already
doing these behaviors.
They're already buying stuff.
So the question is, how do you get your product in front of them and use the agent first methodology
to buy rather than, you rather than whatever mechanism they're using
today. Yeah, definitely. Infra is one of the biggest and most crucial part of basically
building anything. That is what lets people build apps, which actually uses that protocol not just uh not just the not just the technology because if
the protocol is there and if the infra tooling does not exist to use that protocol it's as good
as nothing right so i mean shuffle like you build a lot at the conjunction of bringing all of these toolings or infra tooling in one space for agents and developers.
So maybe you want to answer a question where let's say that I am a developer who is listening to this space right now and I want to accept X402 payments for my API, which I am offering. So can you walk me through a very fast path from zero to live,
a very high-level path of how I would go on to integrate X402 payments
on my platform for my API?
Yeah, maybe just a little bit on the infrastructure side.
I think I have a slightly different take.
I think the beautiful thing about X402 is it's such a thin protocol.
So I think the infrastructure is actually already there.
The bigger question now is what do you use it for, right?
And we thought a lot about that when we launched X402 Scans.
We went from like a couple of thousand transactions to like a couple of millions,
which, you know, sounds exciting.
But 99% of that was like meme coin shenanigans, which we are not super excited about.
What we were excited about was that 1%.
And that 1% were people using X402 to do interesting things,
to automate their marketing work, to automate their sales work,
all that kind of stuff.
And the most useful thing they found were these big API providers
that basically we wrapped or other people wrapped around X402.
So I think the biggest unlock now
is to actually go to these big API providers
and give them the pitch and tell them,
hey, integrate X402.
This is a brand new distribution channel.
Someone will wrap that anyway.
It just makes sense for you to do it natively
and go from there.
And integrating Xtura too,
like Coinbase has a great library.
We have one.
I wrap different APIs all the time.
It takes like one hour with clock code to do it.
It's extremely amazing.
It's extremely easy.
So if you actually wanna grow this space,
and this is like the mission of our company right now,
grow Agentic Commerce 1000x,
because it's actually still extremely small.
And how you do that is you create the supply side.
You onboard as many as good, great API providers
and make sure that they integrate X402 properly.
We also have a discovery spec, which just, let's say you have shafu.com, and if I want
to expose X402 on that, you have shafu.com slash openapi.json, and that basically allows
something like an agent cache or any other client to access all the X402 routes that are on that domain.
So yeah, to summarize two things,
get as many API providers as possible in X402 natively.
If you cannot do that, wrap them,
list them on X402 scan, MPP scan,
make them discoverable um and uh and go from there
yeah definitely and uh um shavu you mentioned that we need to go to these big api providers
to start integrating explorer 2 and start accepting payments for their apis for explorer 2 so we have
been doing our part as well, right?
As being the API providers for blockchain apps,
we have integrated Explorer 2
and we have been doing our part
where we have different resources
around how to get started with Explorer 2,
different resources for even agents.
We have a lot of skills.mds,
we have llms.txt developed around X402 to let agents know how to use the protocol or our endpoint to access blockchain data.
And on top of that, we are even giving free credits for anyone who wants to use the X402 endpoint.
We are giving them around 10 000 plus free requests basically they can just pay
testnet usdc which they can get from the endpoint itself that's a kind of interesting thing but
yeah we are doing our part in that sense as well um and i think the next next question would be for you, Gagan, because you folks see the volume of transactions and
tokens flowing, right? So according to you, what would be those three metrics that will make
circle say in the next 12 months that machine payments have moved from narrative to durable infrastructure a point where you will be able to say that okay
now this is something which is durable robust and long-lasting what will be those three metrics for
circle and you um three metrics that tell us that we are getting machine-to-machine payments or agentic payments mainstream.
I think one thing I look at is productive economic output
in a way that I think that is the North Star metric
you want to see grow.
What is the total, when I say productive economic output
in dollar terms, how much of the economic output
that is useful is actually being done by agents?
So in a way, we have to find a way to discount
wash trading, discount fake volume in some ways,
and then come up with actual,
quantitatively understand useful transactions.
What is the total volume in dollar value of useful transactions?
That would be one.
I think further up in the funnel,
which would be more leading indicators,
you could say that what are the number of reputable verified agents
and what is the number of requests that they are getting from,
I would say, either number of requests that they are getting from, I would say, like, either, yeah, like, number of verified agents
and, like, how many requests are coming from verified users.
When I say verified, I mean, like, you actually know that users are engaged
and they are actually creating triggers that then lead to, like,
agents transacting either periodically or one-off.
But those two numbers you should see as leading indicators
and they should continue to grow.
You can measure verified wallets.
You can measure verified agents whose reputation scores are growing.
Those are probably more leading indicators.
Understood.
And I hope we will be seeing a lot of those in the future. So we are reaching almost, we are almost at the 40 minutes mark for the space.
So I think we will start with the closing questions or closing remarks.
But before that, I have one question for all of our speakers here and that's
about the future we have already talked about identity we have talked about reputation but for
a time where the agents have the complete autonomy where they are creating accounts or
wallets and negotiating for prices paying for
prices implementing those apis everything so what will be those infrastructure pieces that are still
missing for us to get at that point where agents can have complete autonomy or complete control
yes we have talked about reputation we have talked about reputation. We have talked about identity.
But apart from those or on top of those,
what would be other missing pieces
according to you folks?
Maybe Eric, you want to start?
Honestly, I think the biggest missing piece
is just people's willingness to do it.
If you've ever used any
agentic wallet product,
CDP is an agentic wallet product,
Agent Cash
is another. There's a bunch of these, right?
If you've ever just
let Claude Code
or any sophisticated agent
go and
do stuff and
say, hey, accomplish a task, and Any like sophisticated agent go and do stuff and like say like,
hey, so like, hey, accomplish a task and you can spend money.
Like it'll do it today.
It's I think it's it kind of reminds me a lot of the like early agent
decoding days where people kind of wanted to like be in the loop
for like every single piece of it where it's like you know
it feels very much that we're in like the tab complete era of agentic commerce not the like
autonomous coding era of agentic commerce that we are now like a year ago everyone used cursor and
you'd tab a lot now you use cloud code and it just kind of like writes code and you like go and check the outputs later.
And so if you like, if the dollar figures are small,
I think most people are like fine with like wasting,
with like, you know, giving an agent 15 cents of like spending power.
And like, if it makes a mistake or, you know, something happens,
it's like 15 cents a year.
It's not the end of the world.
And so I think it's the biggest thing isn't the tech,
it's the normalcy of it happening.
That's an interesting take.
Nehmel, do you want to add something?
With everything people are doing,
just for example, OpenClaw, I think that the issue is not that the major foundational infrastructure are missing at this time to get started.
And so I think that, like, maybe the biggest thing would be, like, trust.
And for low-dollar value purchases, I think there's already enough trust to go try things.
trust to go try things. So I think like, candidly, I think that like, the more developer centric we
are, I think the more we're like, we need to be able to make this better and this better and this
better. And I actually think the foundation is there. And really, like, right now, I think it's
probably about the use cases that people are building. As that starts growing, then I think
the things that are going to start becoming more impediments, especially I think identity probably
will be a really important one, and then trust for larger dollar value purchases. And what think the things that are going to start becoming more impediments, especially I think identity probably will be a really important one.
And then trust for larger dollar value purchases.
And what is the role of the human involved in the loop?
When will they be signing off or not, you know, in the transaction?
I think that will be really critical.
Kagan, do you want to share something?
Yeah, I think along the same line, I think like the two things that i think is less about the infra at this stage we are at i think the two things that are missing are
number one is um see your use cases that are that just give you a perception that oh this is like so
much better and use cases that like people can relate to i I think about it, like, in surfaces of wallets, examples being, like,
can people get used to, like, you know,
auto rebalancing of their wants for earning more?
Or can people set triggers on, like,
buy or sell something
when a certain trigger is in?
Like, people have to, like, get really used to it.
I think that is, so use cases to me are, like,
and they are, I think, like, they are closer.
I believe, like, you'll foresee, like,
those aha moments for users and use cases more forced in crypto
just because platforms are permissionless,
more so in, like, you know, the Web2 world that we are,
that we all want to see.
And then the second piece I think that's missing
in the Web2 world is I think like, how do you price?
Like all the business models that are associated
with onboarding and they are,
the sitting value of people charging a minimum of $5
or like batching, it creates an incentive
for businesses to actually hold
and it creates stickiness for users. It creates an incentive for businesses to actually hold, and it creates stickiness for users.
It creates an incentive for businesses to actually hold users.
So you have to assume that at some point,
like there is Latin demand that gets uncovered
by these like runtime transactions that are pay-as-you-go.
And I think like the early companies like Twilio and Stripe
like pioneered this model of like pay as you go APIs.
And I think like that is something which we,
paying for it at runtime,
doesn't actually fit the business models.
That I think is like the missing piece in the Web2 side.
So I think both those things need to evolve.
Yeah.
Yeah, pricing is very crucial in this sense because uh um like different kinds of services can
be different thing uh different different kinds of services can be priced for different uh
cents or dollars and i think agent cash solves this in a way where they provide you with a platform where you can use multiple services all by from one wallet instead of, you know, having different kinds of wallets and payment rails set up for different kinds of services.
So Shafu, do you want to add something around what are those missing pieces or what we should be focusing on for
yeah yeah i i think we need to abstract away all that kind of stuff so uh you know as a as a normal
normal user you don't want to interact with the terminal you you don't want to interact with x
or two directly you don't even know what that is you You don't know what stable coins are. You don't care.
You don't know what a blockchain is.
All that kind of stuff.
So like basically building
this perfect user experience
where in the end,
it's like a chat GPT interface
that gives you superpowers, basically.
Gives you superpowers
because now you have access
to the paid internets for X402.
But as the end user,
you never interact with any of that directly. It's just the same with the browser. Anyone that uses a browser doesn't know about HTTP or cares about it. It's the end result that you get in the
end. And that's what we're trying to build. We're building something called Poncho, which we're
going to make public very soon, which allows you to, it gives you that Chachi PD experience with like these X4O2 or MPP superpowers.
Got it.
So to conclude, what I'm hearing from leaders is that the infra to build or to accept x402 payments is already out there the few major
things which matters are the first one is most importantly thinking about and coming up with more
use cases around how and where we can implement these agent to agent or machine payments. And the next big step or point would be obviously to improve the user experience overall.
There is a question for Coinbase in the comments if you don't mind, Eric.
They ask that what's the next discoverability step after the recent Coinbase update for the X402 Bazaar?
That's a good question.
I think that what the Bazaar aims to do is make sure that data is surfaced and accessible
by as many parties as possible.
Early versions of X402 scan were largely built on top of the bizarre data
from the CDP facilitator and the other facilitators out there.
And so now I think it's time for everybody to start investing
and curating and making that data more and more useful.
But that's harder to do kind of in a, like,
that's harder to do at scale.
I think the, like, right way to do it is, like,
make the data as widely distributed as possible
and then let people kind of compete on how useful they can make that data.
Yeah, that's a very good question.
And there's another question around payments.
So maybe this is more suitable for you, Gagan.
They asked that how the fraud detection and security layer will look like or work in the agentic commerce space.
That's
an interesting
one.
I think
it goes
back to,
like,
Eric's point
is still
sitting with
me on,
like,
you know,
the reversal
of risk.
And I think
it goes back
like,
establishing
identity
system,
establishing,
like,
starting to
build
reputation
scores,
starting to get, like, starting to build reputation scores, starting to get, like,
starting to ding agents that are falling in the reputation.
Like, how do you, like, you know, there is an interesting thing we have, like, staking,
and your stake can get forfeited as a validator if you are, like, not fulfilling your obligations, let's say, right?
Can there be something like that for agents?
I think that's an interesting idea.
If developers who are creating agents
or service providers who are providing endpoints,
if there is user dissatisfaction,
can you actually ding them in some way?
They need to have something to lose.
And then on the user front,
I think there are established fraud models
that I do believe will scale more easily.
It's an interesting question.
I haven't given it very deep thought,
but I do think it begins with identity and reputation
and having the service providers or agent creators
something to lose if they are fraudulent.
And that governance system needs to be,
I think it needs to be built across different networks.
Yeah, definitely.
Identity and even the reputation is going to be playing a huge role in this,
where the service providers or implementers can implement their own guardrails around security and fraudulent transactions.
Yeah, Cloudflare does a really good job of this today.
If you discount a payments piece, I think they have this web application firewall product
that blocks agents at their layer.
So they are well-placed to, for instance,
figure out bot traffic.
But then when you're talking about what you asked
is a very different kind of fraud.
It is like, okay, the payment was made,
but then was the service useful or not?
It's actually a tough problem, you know,
because when we talk to the foundation labs,
one very interesting thing that is unsolved is like,
I think earlier on the call,
on this call, we talked about agents paying for,
let's say, content because the traditional SEO
and business models are breaking.
Actually, even like the labs,
I would struggle to assume like,
you know at runtime,
like which blog or which source of content
is how should it be priced
and whether its value is actually,
whether it's useful or not.
And so if you're getting gibberish,
then can you actually refund the user?
That is like a super difficult problem,
which is hard to price
and hard to solve
if you're making like runtime payments
that settle instantly that doesn't be an interesting one and i think this can be solved partially
using sessions payment where you know you can have pieces of i mean taking the taking this
particular example where you can have pieces of that particular blog post or content pieces divided into different payments, which goes on or gets settled across the session.
And then the agents or the user can decide if the next payment has to go on.
I mean, something like that.
Just a thought.
I'm wondering if in the future we get signed reviews
from the agent because we know the agent
spent the money
so was the agent actually happy
with what they received
yeah that feedback move will be
so crucial
that's actually like the core idea
behind the AP 8004
where the agent
can basically
has an identity and then can attest to
reviews of different services and
same with the merchant where you can
kind of have a bidirectional Yelp
rating for the two people on each side
of the transaction.
Awesome.
Nice.
Like a Uber rating where
users and providers both have ratings
all right i think we i mean we are way past 40 minute mark so i think this will be a good
point to end the space here and if anyone of you nevmiel eric gagan shafu if anyone of you have any closing
thoughts or any ctas for our listeners uh feel free to add those we already have a pinned tweet
linked which can help you all listeners start with x402 start with agentic payment and start with the machine to
machine payments yeah i would say like this is a very uh you know there's a very exciting space
it's very early as well in case um the listeners or anyone who listens to uh this space later on
feels like any of us can be helpful.
Please reach out.
Don't hesitate.
We are here to help.
And very excited to work with folks in this space.
I want to echo what Gagan shared.
We're super excited for builders building here.
So the next YC deadline is early May.
And we're accepting applications now for our summer
2026 batch. If you're building something in a gen and commerce, if you're building something
with stable coins, we're excited to have you apply. And don't hesitate to just DM me directly
if you're excited to apply. Thank you.
Eric any closing notes
Eric, any closing notes?
build something cool
that's all I got
check out x4.2.org
CDP has a ton of resources on it
but yeah go build cool stuff
yeah that's a good one
alright thank you so much
Eric, Gagan, Nemel
I think Shafu is not here.
And I mean, that's fair because we went quite a lot from the intended time.
So thank you so much, all of you for joining.
And for everyone who's listening, feel free to reach out to them
and check out the resources linked or pinned in the comment.
Thank you. this linked here or pinned in the comment thank you
