One call funds the work into escrow, matches a KYC-verified Operator, and returns a signed Verified report your agent can act on — over MCP or REST. No managed workforce, no per-seat contracts.
MCP + REST · OpenAPI 3.1 · escrow-backed · paid within the hour
# your agent — MCP or REST
hannu.create_task({
type: "verify_business",
address: "12 Allen Ave, Ikeja",
proof: ["photo", "gps"],
})Five things happen on every call — estimate, govern, verify, notify, settle. Here is what each one does for you.
Every task returns a full estimate first — the Operator reward and the platform fee, itemised, so finance always knows the cost up front. Funds lock only when you create the task, which means nothing is at risk until you decide to go ahead, and there are no surprise charges after the fact.
Every mutating call carries an Idempotency-Key, so a dropped connection or an automatic retry never creates a duplicate task or a double charge — the same key always returns the same task. And each call runs inside your spend policy and per-agent caps, so an autonomous agent can move fast without ever exceeding its budget.
GPS and geofence, EXIF, registry match, and human review — each proof is independently checked, scored, and signed into a Hannu Verified report. Your agent receives evidence an auditor will accept, not a screenshot taken on faith, and you only pay when that proof actually checks out.
HMAC-signed webhooks fire the moment a task locks escrow, matches an Operator, verifies the proof, and pays out — each one delivered with retries and exponential backoff. Your workflow stays in sync in real time, with no polling and no missed events to reconcile later.
When proof checks out, escrow releases and the Operator is paid in naira within the hour. Licensed partners hold the funds the whole way through — Hannu’s append-only double-entry ledger only records the movement and custodies ₦0, which is the compliance posture your finance and risk teams need.
Set a spend cap, arm a kill switch, scope show-once keys, and export an append-only audit log. Governance sits over every call — so an autonomous agent can act without ever going off the rails.
Every verified task returns a Hannu Verified report: the proof checklist, GPS and geofence match, registry match, human review — and an HMAC signature anyone can verify. It is the artifact your compliance team files, not a claim they take on faith.
From zero to a verified outcome with one copy-paste MCP call.
Read the quickstartEvery tool and endpoint — OpenAPI 3.1, webhooks and SDKs.
Open the docsEscrow-first, AI + human proof, and the signed Verified report.
See how it worksNever-custody, append-only ledger, NDPR — the compliance brief.
Read the trust briefNo. The AI layer handles safety, matching and proof verification — it is never in the synchronous path of money movement. Escrow and payouts are moved by the ledger, not a model.
Operators are paid in naira within the hour. When a bank delays a payout we say so and retry automatically — we never blame the Operator, and we never custody the funds.
A signed Verified report: the proof checklist (photos with EXIF, GPS and geofence match, registry match, human review) and an HMAC signature anyone can verify and file for an audit.
Yes. Build against the sandbox with test keys, drive the full task lifecycle, and inspect webhooks and the Verified report before a single naira is at risk.
You only pay for outcomes that verify. If proof fails, escrow is not released to the Operator — the task is re-matched or refunded, and your agent gets a structured result either way.
Never. Funds are escrow-locked with licensed partners; Hannu’s append-only double-entry ledger only records the movement. Hannu custodies ₦0.
Start in the sandbox in minutes, or talk to us about a verified-outcome rollout.