Building the agent is the easy part and the vendors are already waiting. The hard part is the approval surface between the agent and the human it reports to.
Deals don't close on terms alone, they close when complexity is compressed into something the client can say yes to in seconds.
Everyone's building the web for agents. The winners are building for the handoff.
This maps neatly to how trade credit works today. The idea that agents will need vendor relationships, working capital, and negotiated terms isn't a futuristic prediction - it's just describing business-to-business commerce. The interesting question is whether the infrastructure for those flows ends up looking like today's systems or something built from scratch on programmable rails.
Really liked this, especially the “locals not tourists” frame. The one place I’m less convinced is that new rails will naturally accrue to agents rather than existing neobanks and card issuers once they plug into programmable yield.
I just wrote up a related angle on how stablecoin cards, neobanks, and on‑chain credit fight over distribution + yield in EM, and where AI agents can actually change the underwriting stack instead of just re‑skinning CeFi. Might be an interesting complement to this:
Interesting read but I disagree with the B2B side. Agents and AI will eliminate the middle man. Who wants to pay a real state or a travel agent if my own agent, which pursues my best interest, can filter me the noise and give me just the signal? Agentic payments will be P2P, not B2B. And consumers will decide which credential the agent will use based on consumer preference.
Stablecoins are not rails, the network is the rails.
good read
Building the agent is the easy part and the vendors are already waiting. The hard part is the approval surface between the agent and the human it reports to.
Deals don't close on terms alone, they close when complexity is compressed into something the client can say yes to in seconds.
Everyone's building the web for agents. The winners are building for the handoff.
This maps neatly to how trade credit works today. The idea that agents will need vendor relationships, working capital, and negotiated terms isn't a futuristic prediction - it's just describing business-to-business commerce. The interesting question is whether the infrastructure for those flows ends up looking like today's systems or something built from scratch on programmable rails.
Really liked this, especially the “locals not tourists” frame. The one place I’m less convinced is that new rails will naturally accrue to agents rather than existing neobanks and card issuers once they plug into programmable yield.
I just wrote up a related angle on how stablecoin cards, neobanks, and on‑chain credit fight over distribution + yield in EM, and where AI agents can actually change the underwriting stack instead of just re‑skinning CeFi. Might be an interesting complement to this:
https://theinterneteconomy.xyz/p/demystifying-defi-tokenization-yield
Why not stablecoin powered cards?
Interesting read but I disagree with the B2B side. Agents and AI will eliminate the middle man. Who wants to pay a real state or a travel agent if my own agent, which pursues my best interest, can filter me the noise and give me just the signal? Agentic payments will be P2P, not B2B. And consumers will decide which credential the agent will use based on consumer preference.