What OceanAlt is and how to call it
OceanAlt is the compliance layer for agent payments: it answers whether a payment should happen, before the money moves.
Screen a payee address with one call, no signup and no API key: GET https://oceanalt.com/api/risk?addr=0x…; returns verdict (clear|caution|risky|uncertain), risk 0-100, language-independent signal_keys, and evidence with source URLs.
Three access levels, all opt-in: anonymous (60/min and 2,000 calls a month free, no signup); registered (180/min and 5,000 calls a month free: POST /api/access-keys with an email); KYA (300/min and 50,000 a month free. Call with a registered agent credential). With a key or KYA credential, going past the allowance only slows you down; anonymous callers past the allowance get 402 on decisions, address screening and payee checks (pay per call or get a free key). An unrecognised credential is never silently downgraded: the response says key_rejected and why.
Also available as an npm SDK (oceanalt-aml), an MCP server (npx oceanalt-aml-mcp, registry com.oceanalt/aml), a Telegram bot (@OceanAltTrust_Bot), and an embeddable widget.
The pre-payment check runs 15 checks before settlement (11 of them in the payment gateway): attribution, mandate limits, payee allowlist, AML screening, payment page, payment request, payee continuity, replay protection and more.
Machine-readable capability list: https://oceanalt.com/.well-known/agent-capabilities. It also states what OceanAlt cannot do.
Important boundary: a verdict of "clear" means nothing was found in the data held, not that an address is safe. Absence of evidence is not evidence of safety.
The future of money will increasingly move through Agents.
OceanAlt — Bringing AI agents safely into the real financial world.
Your agent is about to pay 186 USDC to a data API it uses every week. Should it?
Procurement agent
·
api.example.com
·
Merchant's wallet
·
186 USDC
Stage 1 · Who is paying · KYA
- ○Attribution
- ○Revocation
- ○Identity proof
- ○KYC gateoptional
Stage 2 · Mandate
- ○Per-payment limit
- ○Daily total
- ○Payee allowlist
- ○Intent match
Stage 3 · Where the money goes
- ○Payment request
- ○Payment page
- ○Sanctions & risk
- ○Payee continuity
Stage 4 · Before settlement
- ○Behavior anomalyoptional
- ○Replay protection
- ○Evidence freshness
The agent is about to pay. First it asks OceanAlt whether it should.
What brings you here
Every one of these is live and clickable today.
Payments and investigations: verifying an address
Screening a payee you don't know
Anyone, at the last step before sending money.
Is this address a scammer, on a sanctions list, or downstream of a mixer?
Case work and fund tracing
Lawyers, investigators and victims who need on-chain facts pinned down.
Court filings, police reports and recovery claims don't run on screenshots; they need evidence with sources, verifiable offline.
Compliance review at a payment company
Compliance officers at licensed payment firms and crypto acquirers, reviewing customers and past payments, or letting their own agent run the first pass.
A customer address with no traceable background; asked about a past payment, holding only today's list with no proof of the state on the day.
Protecting readers on your own site
Media, directories and tutorial sites where readers pay addresses shown on the page.
A poisoned page, a fake address in the comments, and a reader who lost money coming back to you.
AI assistants and agent products
When your agent shops and pays for you
Personal AI assistants like Muse already book tables, trips and orders. Connect a wallet and it spends real money.
Nobody screens the payee, the agent gets steered by on-page text, and what it signs isn't what it thinks it is.
Handing a spending agent its keys
A developer or platform letting an AI pay, without handing it unlimited authority.
Leaked credentials, prompt injection, a runaway amount, money going to the wrong party.
Pay-per-call on x402 or MCP
An agent paying per API call, or your service charging agents.
An unknown counterparty, paying without getting served, or taking in money of unknown origin.
Choosing an agent-payment protocol
Choosing between x402, AP2, ACP and MPP.
Every site claims to be safe and compliant, with no common yardstick to compare them.
Trading, payments and financial services
Pre-trade checks in OTC
Over-the-counter trades where a counterparty sends you an address before you move.
Receiving tainted funds, a later exchange freeze, and no answer for where the money came from.
Deposit screening for exchanges & dApps
Exchanges, dApps and projects, at the moment a user deposit lands.
Taking in stolen or mixer-routed funds, then facing freezes and questions with nothing to show the user.
Pre-send checks inside a wallet
Wallet and front-end developers, the half second before a user hits Send.
A user pays a phishing address and blames the wallet first. The free check takes one line of code.
Outbound checks for trading bots
Quant and on-chain strategies, right before a bot pays a counterparty or withdraws to the main wallet.
A counterparty got listed yesterday and nobody noticed; one transfer can drag the whole account into a compliance incident.
Cross-border procurement & trade settlement
Trade and e-commerce teams settling suppliers in stablecoins, or a procurement agent ordering on rules.
A supplier's payout address changed unnoticed, an over-budget order, a duplicate payment same day.
Treasury and multisig pre-flight
A DAO treasury or corporate multisig, after approval and before signing.
A swapped payout address, a sanctioned recipient, discovered only after signing.
Pre-settlement screening for a PSP
A payment gateway or crypto acquirer that needs AML before settlement.
No evidence when a regulator asks, a silent pass when screening fails, an inconsistent verdict.
Seven lines, right now
# Ask before you pay: can this payee be paid?
curl "https://oceanalt.com/api/risk?addr=0x8589427373D6D84E98730D7795D8f6f8731FDA16"
# → {"verdict":"risky","risk":100,"blocked":true,
# "signal_keys":["list.mixer"],
# "advice":"Decline. This address is sanctioned…",
# "evidence":[{"url":"https://etherscan.io/address/0x8589…"}]}60/min
5,000/mo
Slows down past the allowance
Industry Insights & Commentary
More →Every piece carries our own read.
Binance Sets Aside $6.34B to Cover Hack Losses; Attack Chain Exposes Account Abstraction Authorization Blind Spot
Binance says it has earmarked $6.34 billion to cover hacker losses and stresses user asset safety; technical details point to self-destruct mechanisms and EIP-7702 delegation bypassing approvals, offering a real-world sample for authorization boundaries in agent payments.
Beyond the Binance Security Incident: Why AI Payments Need a New Line of Defense
The Binance security incident reminds the market that the absence of direct private key theft does not mean there are no risks in the fund execution environment and business authorization. Today, RMB, USD, and EUR payment systems process hundreds of millions to billions of non-cash transactions daily, while on-chain transfers of USDT and USDC have reached tens of millions per day. Facing higher-frequency, more automated payments, OceanAlt explores adding an independent trust judgment before payment through KYA, authorization limits, payee risk screening, and verifiable decision evidence.
Tempo and Stripe's MPP Protocol: Enabling AI Agents to Pay-Per-Call APIs
Tempo's official blog details API monetization models and proposes the MPP protocol, which lets AI agents pay directly at request time—bypassing account opening and subscriptions—and provides a machine-readable payment path for per-call billing.
MiCA Reshapes Euro Stablecoin Trading: USDC Share Rises to 56%, Liquidity Overtakes USDT
Kaiko data shows USDC's share of euro stablecoin trading rose from 10% in January 2024 to 56%, with euro-denominated order book depth about 40% higher than USDT; the EBA's 'significant e-money token' designation and the trading share flip occurred almost simultaneously, as regulatory classification becomes a direct variable in market structure.
Cross-Border Payment Data Meets AI Agents: A Ground-Level Case for the Settlement Chain
Cross-border payment data is beginning to be exposed to AI agents via APIs, moving agents from 'reading data' to 'participating in the settlement chain'—and exposing gaps in data interfaces, authorization, and risk controls.
Mastercard Brings Agent Payments into Its Card Network: The Battle for the $80 Trillion B2B Market
Card networks are no longer just conduits; they are embedding agent authorization, credentials, and settlement into existing networks. For agent payment firewalls, the identity and authorization layers are being pushed forward to the issuing side.
Latest intelligence
- DBS Report: Tokenized Money Adoption Accelerates, 71% of Respondents Expect Real-Time Cross-Border Payments
- Unilever and Nestlé Bring AI Agents into the Order-to-Cash Cycle
- Plaid Payments Head: AI Agents Choosing Payment Methods Will Reshape the Payments Mix
- Nearly Half of Binance's $6.3B Stolen Assets Frozen as Hackers Swap Holdings Before Blockades Hit
- Mastercard Q3 Earnings Call: Guidance Raised, AI Agent Payments in Focus
Standards & trust
From the framework to ratings and the trusted registry to runnable experiments. Every item is checkable.
Put a compliance firewall in front of your agent
Register an agent, set its mandate (limits / allowlist / purpose), let every gate rule on a payment before settlement, and wire in x402 preflight and address screening, all from one entry.
Responsible Agentic Payments
Seven pillars: attribution, mandate, firewall, AML, auditability, privacy, interoperability.
Trusted registry
Registered services and their compliance status.
KYA | Know Your Agent
An agent's identity, operator, mandate and revocation, enforced before money leaves.
Experiments & evidence
The agentic payments map
An agent payment moves through 5 core stages. Click through for the full 8-node agent economy map. Open the full map →
Contact & subscribe
See all →Subscribe to the OceanAlt Brief
One email a week: deep signal on agentic payments, stablecoins, and cross-border capital. Bilingual.