Banking Agent Authority™ — proof for every agentic banking action

KYE™ is not another banking OS or agent platform. It is the authority layer underneath them: the runtime control point that decides whether an agent's proposed action has authority, then emits a signed, replayable record. When an AML agent closes an alert, KYE™ proves who authorised it — cutting exam-prep from days to minutes.

Start a banking authority pilot See the banking agents

The line: run agentic banking vs. prove authorised banking action

Your platform coordinates the frontline — customers, employees and AI agents across digital channels and operations. KYE™ governs the action boundary: the instant an agent or employee proposes a consequential banking action, KYE™ checks actor, mandate, customer consent, purpose, scope, policy, risk, approval and finality, then returns allow / deny / require_approval / suspend.

The authority questions are banking-grade: Is this agent allowed to initiate this workflow? Who delegated that authority? Is the customer consent valid for this purpose? Is the action advice, servicing, credit, payment, complaint or final decision? Did the output stay advisory, or did it update a system of record? Can the decision be replayed for a regulator, auditor, ombudsman or litigation years later?

Already shipped: six banking Authority Packs™

You do not adopt a new framework — you bind the packs that already exist. Each pack governs a banking action vertical against its real regulatory spine: ECOA/Regulation B, FCRA, Fed SR 11-7, FCA Consumer Duty, FCA CONC, MAS guidance and the EU AI Act™. Pick the verticals your agents touch.

Eight banking governed agents, each authority-checked at the action boundary

Where you deploy agents, KYE™ ships the per-action authority agents that govern them. Each is a first-class principal with a delegation chain, a KYE™ Evidence Pack™ and a Replay-Proof™ trail — so a CISO or auditor can reconstruct any decision offline.

• Payment Authorization Authority Agent — amount, recipient and finality
• Credit Card Action Authority Agent — servicing and dispute actions
• Mortgage Lending Action Authority Agent — origination decisions
• Sanctions & AML Authority Agent — screening and disposition
• KYC-Refresh Agent — customer due-diligence refresh authority
• Chargeback Evidence Agent — representment evidence assembly
• Settlement Finality Authority Agent — irrevocable-action finality gate
• Threat Response Authority Agent — operations prevent / pause / remediate

Browse the governed agents

For AI-native Banking OS & unified-frontline platforms

If you run a Banking OS that coordinates customers, employees and AI agents as one frontline, KYE™ is the authority layer beneath your agents. Your platform orchestrates the journey; KYE™ proves the journey stayed inside authority.

The integration is a runtime check: your agent or employee proposes a consequential action, KYE™ evaluates actor, mandate, consent, purpose, data scope, policy, risk, approval and finality through the Evidence Pack™ chain, and returns a verdict your platform continues, escalates or blocks on. A Banking OS can execute; KYE™ decides whether execution has authority.

For banking AI agent platforms

If you power banking agents — conversational, agentic or operations-facing — KYE™ governs every consequential action those agents take. A trusted banking answer is not the same as an authorised banking action, and a banking agent can understand the customer perfectly and still lack authority to act.

KYE™ binds each agent to a mandate and a purpose, then gates the moment it tries to refund a fee, initiate a payment, pre-approve a loan, change an account detail or escalate a complaint. Every gated action emits a signed record that a regulator, ombudsman or customer dispute can replay — so "the agent did it" is never the end of the trail.

For credit unions & community banks

AI can help a credit union lend faster, work leaner and manage risk better. KYE™ proves each lending action stayed inside authority — so your lending, risk and operations leaders deploy AI actions they can defend to an examiner.

Credit unions and community banks start with the KYE™ Agentic Lending Authority Pack™, which already lists credit unions among its adopters and governs underwriting, adverse-action and arrears decisions under ECOA, FCRA and Fed SR 11-7. You do not just need AI adoption — you need AI actions you can defend.

One authority spine under every banking action

Every pack and agent above resolves to the same runtime primitives — so you govern lending, payments, AML and servicing through one control point, not seven. KYE™ reuses these primitives; it never reimplements them per vertical.

The result is one provable answer for every customer-impacting action: who had authority, for what purpose, with what evidence, and whether it became final under Authority Finality™.

Start a banking authority pilot Explore the banking packs

Canonical KYE™ surfaces referenced on this page: KYE Protocol™ · Purpose Permission.