Most discussions about proof-of-human stop at bot filtering. That misses the harder system problem.
Agent traffic needs admission control and flow control at the same time.
Admission control answers one question: who can create demand in the first place. Flow control answers another question: where admitted demand goes when capacity is limited.
Pura already does flow control. Requests hit one endpoint, then route across providers under cost and confidence constraints. Settlement happens over Lightning. Receipts provide an audit trail.
Proof-of-human complements that layer. It does not replace it.
A system that only has proof-of-human still melts under load. A system that only has routing still struggles with free-tier abuse and delegation trust. The stack works when both layers are present.
Pura should not become a biometric issuer and should not claim global identity resolution.
Pura is the enforcement and routing layer that consumes signals from issuers and converts them into useful controls:
That lets Pura stay issuer-plural. World ID, Human Passport, and future issuers can plug into one enforcement surface instead of forcing every app to rebuild the same control plane.
Pura already models stake-bounded capacity with cap_i = sqrt(S_i / u_i).
and minimum-stake friction. That mechanism is one form of admission control because it prices demand against committed collateral.
Verified personhood is another form of admission control with different trade-offs. One is economic collateral, one is uniqueness credentials. Both can feed the same flow-control and settlement path.
The market has distinct products for issuance, payment mandates, and content credentials. What is still sparse is a gateway that enforces all of them during live agent execution.
That is the wedge: consume proofs, enforce limits, route requests, settle in real time, and emit receipts.
Admission control and flow control are different jobs. The agent economy needs both.