Eightball Sports — agent-native sports intelligence

Eightball Sports is agent-native sports-betting intelligence. The build starts from an unusual premise: the first consumer is not a person browsing a feed but an agent calling an endpoint. Design decisions follow from that — API-first, machine-readable, priced per request.

At the core sits a quant engine. It computes +EV on offered lines, devigs market prices to strip the sportsbook's margin back to a fair probability, and scans across books for arbitrage — the same three primitives a sharp bettor runs by hand, expressed as functions an agent can call. The engine is a Python/FastAPI service: typed request in, structured probability and edge out.

Settlement is the second design pillar. Rather than gate the engine behind a subscription and a login, each call is metered and paid over x402 — pay-per-call, so an agent settles the moment it asks. That keeps the surface stateless and composable: no accounts to provision, no plan tiers to reason about, just a request and a payment.

The consumer feed — the human-readable read on the biggest games and how the numbers say to play them — is phase two, layered on top of the same engine once the pay-per-call core is solid. Engine first, audience second.

None of this is live yet. The engine, the x402 metering, and the feed are all in the build, and the honest status is early. Coming soon at eightballsports.com.