Q&A With a Composite Senior Protocol Engineer on Agent Commerce
Published 21 April 2026 · 8 min read
Label: This is a composite Q&A. The answers below are synthesised from conversations with several protocol engineers in payments, identity and messaging standards. No single person is represented.
Q: Where do protocols like this usually fail?
On governance, not on syntax. Any protocol author can draft a reasonable envelope. What kills adoption is whether the governance model lets competitors trust each other. GeraNexus has to get that right by v1.
Q: What’s the single biggest mistake you see?
Over-specifying the first version. The temptation is to ship a complete and beautiful RFC. The winning move is to ship a minimal verb set that two independent implementations can clear end-to-end, and iterate.
Q: Signed receipts — on-chain or off?
Content-addressed and cryptographically signed is the non-negotiable part. Where you store them is a deployment choice. Don’t bake on-chain into v1.
Q: How do you handle cross-jurisdiction disputes?
You don’t, in the protocol. You let the protocol carry a jurisdiction hint and you let the marketplaces bind their own arbitration. The protocol is transport; justice is business logic.
Q: What would make you integrate on day one?
Two things. One, a reference implementation that actually works end-to-end, including refunds. Two, a partner program that covers the first three months of reviewer costs while volumes are low.
Q: What should GeraNexus not try to do?
Don’t become an LLM. Don’t become a marketplace. Don’t pick winners among payment rails. A protocol that picks winners is a platform with extra steps.
Q: What’s the realistic launch window?
Spec v0.1 public draft now. Cross-implementation interop demo by late 2026. Meaningful volume by 2027. If any step slips by a year, the project is still on track.
More on our side of the conversation: open questions and the research history.
Help shape the protocol.
Join the waitlist