§01 — The Thesis
Within five years, AI agents will transact at machine speed and human stakes — and the rails they need do not yet exist.
Era I
Humans transact
1990s — 2010s
APIs, payment networks, web auth
Era II
Humans + APIs
2010s — now
machines call services on human terms
Era III
— unwritten —
next 5 years
machines transact at machine pace
§02 — Differentiation
Five committed asymmetries.
Not all of these are universally better. Each is a deliberate choice.
Dimension
Most peers
Consora
Dimension
01Token modelMost peers
Yield-bearing · inflation-funded · revenue share◆ Consora
Pure utility · slashable bond · no emissionDimension
02What gets paidMost peers
ML inference, GPU time, generic compute◆ Consora
Verifiable agentic work — full task loop, settledDimension
03VerificationMost peers
One mechanism, applied universally◆ Consora
Pluggable per task — ZK · Optimistic · CommitteeDimension
04IdentityMost peers
Free creation · ad-hoc Sybil defense◆ Consora
Stake-bound DID · graph-aware Sybil costDimension
05Compute modelMost peers
Custodial compute or on-chain inference◆ Consora
Off-chain work, on-chain settlement§03 — Rigor
Engineered like infrastructure, not marketed like a token.
Ten rounds of independent peer review before a line of production code. Three formal theorems with proofs. Fifty governance parameters, all defended. The document is built to survive scrutiny.
0pages
Technical specification
0rounds
Independent peer review
0theorems
With formal proofs