Authenticate
The consumer is identified from their bearer token in microseconds. No separate auth service — identity is part of the revenue path.
The proxy and ledger are the same process. Every request is measured, authorized, charged, and recorded before the response returns.
Most platforms reconstruct revenue after the fact. Finno creates it in the same atomic flow as delivery.
From authentication to ledger posting — one system, one transaction, one source of truth.
The consumer is identified from their bearer token in microseconds. No separate auth service — identity is part of the revenue path.
Before upstream runs, Finno checks quota, wallet, or credit and reserves the estimated charge. Unfunded requests stop here with 402 — not a surprise invoice later.
The request is proxied upstream with path rewriting, header injection, and compiled transforms. Value is delivered while durability is earned.
Actual usage is measured from the response. The charge reflects real consumption and is committed to the Raft log — durable before acknowledgment.
A double-entry posting updates all six accounts. The algebraic invariant is verified. Financial reports update immediately — not after batch jobs.
The consumer receives their API response. Revenue is already accounted for. Finno adds less than 250 µs on the hot path.
Six accounts satisfy one invariant after every transaction. This is not an end-of-day check. Your books balance by construction.
Three funding sources — quota, prepaid wallet, and postpaid credit — coexist on one API key.
When quota is exhausted, Finno falls back automatically — or returns 402 if none is available.
While requests flow through Reserve → Finalize → Ledger, background loops reconcile usage rollups, run 23 revenue anomaly detectors, sync abuse-investigation signals, reconcile billing disputes, dispatch alerts, and keep financial reports indexed — none of it on the billing critical path.
Finno doesn't just charge requests — 23 background checks continuously scan your ledger, payments, settlements, usage rollups, and gateway telemetry. Every finding ships with evidence, money at risk, and a recommended next step — not a vague alert.
The gateway emits facts; detection runs in the dashboard API against Postgres and gateway /stats. Request overhead stays under 250 µs.
From ledger balance to settlement failures, usage-rollup drift, overdue invoices, error floods, and survival mode — transparent rules first; statistical detectors calibrate in shadow before they alert.
Each item includes severity, money at risk, auditable evidence, an owner role (engineering / finance / SRE), and a concrete next step. Episodes stay open until acknowledged or resolved.
Every request, payment, ledger posting, and /stats signal lands in the Postgres mirror and gateway telemetry.
A periodic loop in the dashboard API scans existing data with SQL and gateway fan-out — never touching the proxy hot path.
Admins see open items ranked by money at risk, acknowledge or resolve them, and get notified via email, Telegram, or other channels.
Anomaly detection finds integrity breaks and revenue leakage on your platform. The abuse investigation board rolls signals into reviewable episodes. Notification channels push critical findings to Slack, email, Telegram, WhatsApp, or your webhook — in the language you choose.
23 background checks on ledger, payments, settlements, gateway telemetry, and usage rollups. Each finding: severity, money at risk, evidence, owner, next step.
Nearline rules and anomaly signals roll into scored cases with an audit trail. Review, override, and escalate — enforcement stays off the request path until you enable it.
Configure webhooks, Slack, SMTP email, Telegram, and WhatsApp. Revenue anomaly and abuse alerts dispatch with cooldowns and a delivery log.
Three cold-path capabilities: machine credentials for ERP and partner systems, maker/checker corrective adjustments, and claim-file billing disputes — without touching the metering hot path.
HMAC-signed credentials for the dashboard API, default-deny scopes, and virtual services for work that happens outside Finno — no sharing of an operator’s browser session.
Goodwill credits, SLA refunds, rate corrections, and opening balances. One operator proposes; a different one approves — separately for consumers and providers.
Consumers or providers open a case, upload a claim CSV, and Finno places every micro of difference in a named bucket. Resolution settles through the same maker/checker adjustment rails.
Deploy revenue infrastructure where proxy and ledger are one. Your CFO will thank you.