Five Myths About Agent Commerce — Counterintuitive Takes

Myth 1: “Agent commerce is mostly about shopping.”

Shopping is the loud example. The volume comes from services, subscriptions and appointments — domains where the user doesn’t want to evaluate twenty options, they want a reliable outcome. Booking a plumber, renewing insurance, scheduling a GP slot: these are where agent commerce will bite first.

Myth 2: “It needs a blockchain.”

It needs content-addressed receipts, signed commitments and scoped consent. Those can sit on-chain. They can equally sit on a normal web-of-trust infrastructure with verifiable credentials. We are agnostic; the protocol should be too.

Myth 3: “Agents will eat marketplaces.”

Agents make marketplaces less visible, not less important. Supply still has to live somewhere auditable. What gets eaten is the 15–30% take-rate, not the marketplace itself.

Myth 4: “Reviews and ratings become irrelevant.”

The opposite. When an agent picks for you, the signal it uses matters more than ever. The difference is that the signal has to be machine-readable, not just human-readable. Star ratings with no structured dimensions will lose to structured quality claims backed by verifiable receipts.

Myth 5: “The take-rate race goes to zero.”

Take-rate goes to the cost of running the protocol plus a small margin — not to zero. Somebody has to run escrow, arbitrate disputes, maintain the spec and pay reviewers. Our bet is 1–3%. Zero is a Silicon Valley meme, not an economic equilibrium.

Further reading

See also 2028 predictions and the FAQ of objections.