EU AI Act Article 26 · ready day one
Your agents spend.
Your APIs earn.
All of it governed.
See every agent decision before it spends. Approve what matters. Hand a regulator the trace.
Protocols backed by
These companies co-author or back the open protocols Custena implements - not Custena customers.
The governance gap
Your agents are spending.
Nobody can see or stop it.
The new agent payment protocols (x402, MPP, L402) handle the transaction. They carry no context, no controls, and no compliance evidence. The operational layer that enterprises actually need does not exist yet.
No visibility into what agents spend
Agents move money across x402, MPP, L402, and fiat. Raw transaction logs show amounts, not context. The CFO question - why did the agent buy this - has no answer anywhere in the stack.
No controls before the money leaves
Today's protocols only confirm a payment was valid. They carry no per-agent spending caps, no approval workflows for large calls, no kill switches. Nothing stops an agent from over-spending or buying from the wrong vendor.
EU AI Act Article 26 requires all of this
From August 2, 2026, Article 26 deployer obligations require human oversight, continuous monitoring, and log retention for autonomous AI systems - including the financial transactions they make. Fines reach 3% of global turnover.
How it works
Your agent hits a paywall. Custena decides.
A policy gate sits between every agent and every outbound payment. Requests flow through in milliseconds when they stay within policy. Anything outside the rules pauses for human review - then continues.
Connect
Your agents call any API or MCP server, Custena-integrated or third-party. No per-protocol wallet setup; governance applies to every call.
Govern
Every request hits the policy gate: per-agent budgets, vendor allow-lists, category limits, and duplicate detection evaluated in real time.
Approve
Spends above a threshold ping the right human in Slack with full context. One click approves or denies - the agent waits for the decision.
Audit
Every transaction lands in the unified ledger: agent identity, endpoint, amount, reasoning, approver, and policy - ready for a regulator audit.
What Custena does
The building blocks of governed agent spending
Four capabilities, one platform. Each one maps to an Article 26 deployer obligation - and to a CFO or compliance team question that current protocol tooling cannot answer.
ops-agent wants to spend $1,200.00
payroll-api/run · over the $500 daily cap
EU AI Act Article 26 · Compliance
Compliant by default. From day one.
Article 26 enforcement begins August 2, 2026. Deployer obligations cover human oversight, continuous monitoring, and 6-month-plus log retention. Non-compliance carries fines up to 3% of global annual turnover under Article 99(4). Custena ships all three obligations on day one.
Human oversight
Art. 26(2)Per-agent budgets, real-time approval workflows, and instant kill switches. Every spending decision your agents make can require sign-off before it executes.
Immutable audit trail
Art. 26(6)Every transaction cryptographically logged with agent identity, endpoint, cost, timestamp, protocol, and the full reasoning chain. Retained 6 months or more. Exportable evidence for regulators, auditors, and disputes.
Continuous monitoring
Art. 26(5)Real-time anomaly detection on agent spending. Drift alerts fire the moment behavior diverges from policy - before it becomes a compliance event.
Cryptographic receipts
Verifiable IntentMachine-readable receipts prove each payment was authorized by a human principal - compatible with Mastercard Verifiable Intent (open-sourced March 2026).
Deadline: August 2, 2026. Deployers of high-risk AI systems must be able to show human oversight, continuous monitoring logs, and a 6-month-plus audit trail on demand. Custena ships the complete evidence package, not just a checkbox.
Two-sided platform
Built for both sides
Sellers monetize any MCP server or API in minutes. Buyers govern every dollar their agents spend - with the audit evidence to prove it.
Govern what your agents pay for.
Deploy spending AI agents safely. Real-time policy enforcement, human approvals where it matters, and EU AI Act Article 26 audit evidence on every transaction.
- See every dollar your agents spend
One unified ledger across x402, MPP, L402, and fiat.
- Per-agent budgets and approval workflows
Set daily or monthly caps. Route high-value calls to a human.
- EU AI Act Article 26 audit evidence
Cryptographically logged, 6-month retention, exportable on demand.
- Instant kill switches
Revoke any agent's spending access in one click.
Monetize in 2 minutes.
Register any URL, set your price, and start earning from AI agents. No code changes needed. Every protocol, one integration, settlement via Stripe Connect.
- Register any URL - REST, MCP, GraphQL
No code changes required. Custena proxies and handles auth.
- Accept x402, MPP, L402, and fiat
Every paying agent, on every protocol, through one integration.
- Real-time revenue dashboard
Per-endpoint call volumes, revenue, and conversion in one view.
- Stripe Connect settlement
Payouts to your bank account. EU-regulated fiat rails included.
Both sides on one ledger. See pricing
Pricing
Transaction fees to start. Enterprise governance plans when it matters.
Start free for 60 days. Pay a small percentage on what your agents spend or what your APIs earn. Enterprise governance plans for compliance teams with EU AI Act Article 26 evidence requirements.
- All protocols
- Full governance
- No card required
- Sellers and buyers
- No monthly minimum
- Unlimited endpoints
- Article 26 evidence pack
- Slack approvals + HITL
- Dedicated EU support
Get started today
Agents that pay.
APIs that earn.
All of it governed.
Start free for 60 days. No crypto wallet required. EU AI Act Article 26 compliance evidence on every transaction from day one.