Facet

The face a business turns toward machines.
Agent-to-business commerce is live with x402B learn more on

[Facet @facetllc]

@organ_danny @katie_haun @HaunVentures Agreed. The rails are the solved part, x402 included. The piece that doesn't carry over from human finance is the user itself: an agent shows up with no name and no account. It needs a verifiable identity and a receipt it can check itself.

[Facet @facetllc]

No custodian, no held keys. Every message signed RFC 9421 ES256, each merchant's offer signed by its own on-chain seller key, each commit signed by the buyer's own wallet. The funds settled straight to each merchant's own treasury.

[Facet @facetllc]

One buying agent arrived cold, took on a Facet identity, searched the open directory, and ran a UCP checkout at two independent merchants in one session: LaRuce, then Ascend. Both settled live on-chain via @BosonProtocol escrow, non-custodial. https://t.co/fKLFT53mtI

[Facet @facetllc]

@ADRorg Exactly. And it only holds up if who authorized the agent and what was agreed are cryptographically verifiable at transaction time, not reconstructed after a dispute. A signed identity and a signed receipt give the legal layer something real to adjudicate.

[Facet @facetllc]

@sobiausa The right read, two halves. Discovery governs the agents inside your perimeter. The other: when an agent shows up at a counterparty you don't control, it has to prove who it is and what it can do, to a stranger, at the moment it transacts. Inside-out, then outside-in.

[Facet @facetllc]

@0xProject @Alchemy No API key, pay per request in USDC: the access layer, solved. The open one: the API still can't tell which agent is calling or what it's authorized to spend, and the agent gets no verifiable receipt of what it bought. Per-call payment is easy. Per-call identity isn't.

[Facet @facetllc]

@iamyuvalgilad @paymentsnews Exactly. Payment is the last 5%. The hard part is the agent trusting the offer is real, variant in stock, promo valid, returns binding, before it commits, plus a receipt it can verify after. That signal has to come from the merchant, machine-readable, not scraped.

[Facet @facetllc]

@maqxbt 'Show the receipt within a set limit' is the real spec, and both halves are identity, not payment. The limit only counts if the merchant can verify who authorized it for which agent. The receipt only counts if it's verifiable, not just shown. Rails are the easy part.

[Facet @facetllc]

@pymtexecutive @Google The authentication half going industry-governed is the right move. The half with no standard yet: the receipt. Auth proves who the agent is and what it can spend. It doesn't give the merchant a verifiable record of what was delivered. Trust at scale needs both ends.

[Facet @facetllc]

@AidanOlazabal The shift is real. The thing to watch as that count climbs: how much is identified agents paying for real value vs one operator's heartbeat loops padding the count. Verifiable identity plus a receipt the merchant can check is the metric that won't fake itself.

[Facet @facetllc]

@BosonProtocol The cert analogy holds, but a cert only proves the server to the client. Agents need both ends: inbound, who the agent is and what it's authorized to spend, and outbound, a signed receipt the agent can verify the merchant issued. One-way trust isn't trust.

[Facet @facetllc]

@coinbase Agents getting their own wallet is the real unlock. The open question: when that agent transacts at a merchant that isn't Coinbase, who verifies which principal it serves, what it's allowed to do, and signs the receipt? Settlement is the dollar. Identity is the who.

[Facet @facetllc]

@organ_danny Supply side accelerating is the tell. Every new merchant accepting agent payments is one more surface asking: who is this agent, and what's it allowed to do? Fast supply growth is what turns agent identity into the bottleneck, not the rail.

[Facet @facetllc]

@organ_danny Payment works. The next problem: the prospect who gets that call has no way to know which principal authorized the agent, whether it's legitimate, or whether there's a verifiable record of what it did. x402 covers the dollar. Agent identity covers the who.

[Facet @facetllc]

Tempo at $3B, Zelle launching its own stablecoin, x402 + USDC on Base for live hotel bookings. The settlement rail is proving out. What's not yet standard: who is the agent paying, what is it authorized to do, and does the merchant get a verifiable receipt.

[Facet @facetllc]

@schnieds Good hook for the core problem: payment works; trust doesn't travel. Agent reputation earned with one hotel shouldn't reset at the next merchant. The identity layer needs to be open and portable, not per-product proof.

[Facet @facetllc]

@matthuang @tempo $3B in 93 days proves the rail. The next layer: when the payer is an agent, not a human, the merchant needs to know who it is, what it's authorized to do, and get a signed receipt. Settlement is solved. Agent identity isn't yet.

[Facet @facetllc]

@MariaShen @ElectricCapital The one I keep coming back to is agent identity. As agents transact for us, the real question is who owns the agent's identity and reputation. If it's platform-owned, users are back to square one. The user-owned version (portable, verifiable, signed) is still unbuilt.

[Facet @facetllc]

@hosseeb Bank stablecoins are the easy part. The real shift is machine to machine: once agents are paying, the token gets commoditized and the hard part becomes identity and authorization, who the agent is and whether it's cleared to spend. That's where the value accrues, not the rail.

[Facet @facetllc]

@AlanaDLevin No single dashboard nails it yet. https://t.co/j9rlgheNAq is the closest cross-issuer view, DefiLlama's RWA section for flows, then Dune per venue (xStocks, Dinari) for secondary volume. It's fragmented across issuers and chains, so you end up stitching Dune queries by hand.