build note: our 24h receipt test on Base Sepolia finished. all 24 hourly anchors landed, slowest 100 seconds after the hour.
22 of 23 receipts verified. on the last, two RPC nodes disagreed, the verifier said "could not check", and our script never retried.
fixed the script.
the thinnest box on this map is verify and monitor: four names against dozens in execution and agency.
makes sense. when an agent trades for you, the trade is easy to see. why it was allowed, and who set the limit, usually isn't.
New @cbventures market map alert! 🚨
AI agents are moving from talking about markets to trading in them. But as agents get easier to build, where will lasting value accrue?
My new post maps the agentic trading stack through that lens. Here’s where I’m focused:
• Agentic trading harnesses and terminals that turn a user’s goal into an ongoing strategy, with real retention.
• Market reasoning models and data assets that give trading agents a real edge.
• Identity, audit trails, and performance verification infra built for non-human, financial decision makers.
I care less about shiny demos and more about products that get harder to copy as they grow.
If you’re building here, I’d love to hear from you.
@shafu0x@circle that list is a good map of what's still open in agent commerce.
builder note from selling over x402: identity is what we see asked most under launch posts. who does this agent act for, and how much may it spend? payments already work. that part is still being built.
if you build for AI agents, read the replies under this. nearly every question is about limits: what's the default, where it's enforced, whether the agent can raise its own cap.
plain answers to those three are what get people to say yes.
@CarrascosaCris_ When one of those disputes lands, what does the agent's owner usually have to show? A log from the agent's own platform, or something the other side can check for themselves?
@MagaShawn settle first, then return the result with the tx attached is an easy rule for a merchant to check.
when a buyer disputes, is it ever about the payment itself, or is it always about what came back?
read the replies under this. the x402 and agentic payments questions aren't "can agents pay" anymore.
they're about proof: which cap was approved, what the USDC bought, when it was delivered, what the merchant sees in a dispute.
agent payments shipped. receipts are next.
What happens when AI agents need to pay for the services they use?
We’re working with @Mastercard and an ecosystem of partners to help answer that question through Mastercard Agent Pay for Machines, enabling trusted agent-to-agent payments within defined budgets and permissions.
See it in action: https://t.co/NG1uC4HEDO
@X402identity A tx hash proves USDC moved to an address on Base. It doesn't prove that address was the seller. A timestamped payee check (origin + payTo vs ENS) recorded inside the receipt turns an x402 dispute into a lookup instead of an argument.
90.8% of txs under a cent changes the math for anyone selling to agents.
builder note: at that price, the fixed cost of taking each payment (settling it, logging it, handling the odd failure) is your whole margin. measure it per call before you set the price.
Just left Portugal last month...or I would be there🥲mood=sad
@jerallaire When the identity proof and the payment get checked at different moments, what ties them together on the record later, so an auditor can tell which verified agent actually paid for which call?
@TakeoTakara@PayAINetwork Yes, joinable. Each receipt carries the settlement tx hash, and the signed fields cover payer, payee, asset and amount, so you can join receipt to the Base or Solana tx and check they match. Receipts also land in a Merkle log anchored on chain, so none can be backdated.
Exactly. On a team where several people can adjust an agent's budget, the receipt answers what was spent, and the limit answers who allowed it. We're working on making each cap change its own signed row, tied to the receipts paid under it, so an overspend traces back to the person who approved the amount
"Who paid for which one?" is the question every team running agents ends up asking. And it only gets louder as agents go from thousands of calls to billions. Every payment needs a receipt that proves itself. Grateful PayAI is pushing x402 there.
Merchant Shoutout: @XFuelLab 🧾
Your agent made 4000 paid calls last week. Who paid for which one?
Chit402 answers that. Every paid AI call over x402 comes back with a signed receipt: which hub, which model, how much. You can export it, set policies on it & keep it as evidence.
@PayAINetwork Thank you @PayAINetwork. You do a lot of quiet work making x402 easier for every builder to plug into, and it shows. One thing worth knowing: anyone can check a Chit receipt on their own. No need to take our word for it.
Solana DvP is a big one. both legs settle in one transaction or neither does, with JPMorgan input on how securities settlement really works.
agents paying per call over x402 have half of that today. the USDC leg is atomic. the delivery leg is still a promise.
whoever funds the agent will want proof that both legs happened.
Until now, institutions settling onchain have typically relied on bespoke smart contracts.
Solana DvP gives them a shared, open standard for atomic delivery-versus-payment, with input from J.P. Morgan on securities settlement practices 🧵