Consumer credit & banking
Service, underwriting, disputes, and in-app agents that act inside a bank’s own rails.
Read bankingObservatory
Financial Agentic AI
A sourced observatory of live agentic AI in consumer banking, fintech payments, markets, and risk — what actually shipped, how it is wired, and what a risk committee should ask next.
Financial agentic AI is a system that can plan a job in banking, payments, or risk and call real tools — cores, card networks, document stores, wallets — to complete it, under identity, limit, and audit constraints. It is not a chatbot with a finance prompt.
- — Most pages on this topic mix Copilot seats, vendor roadmaps, and one famous contact-center story. This one only lists named firms with a public primary source, a described architecture, and a caveat.
- — The useful picture is not a four-box slide. Production systems have a channel, an orchestrator, specialists, tools, and a governance plane that can refuse a posting.
- — Numbers here are attributed. Backtests are labelled backtests. Survey percentages are labelled surveys. If a figure is the firm’s own, it says so.
Open the studioExplore a use case
Ask the observatory
Three sectors. One control problem.
Fintech & payments
Agentic commerce, refunds, stablecoin rails, and merchant discovery protocols.
Read paymentsMarkets, finance & risk
KYC, legal review, portfolio regimes, treasury, and model-risked multi-agent work.
Read markets & riskGuides on financial agentic AI
- Agentic AI in consumer credit and bankingHow banks are putting agents into service, disputes, car-buying, and employee work — and where a chatbot is still just a chatbot.
- Agentic AI in fintech and paymentsRefunds, agent checkout, product feeds, and stablecoin MCP servers — the rails agents actually need before they can spend.
- Agentic AI in markets, finance, and risk managementKYC, credit memos, legal workflows, regime-switching research agents, and the model-risk bar they have to clear.
- Multi-agent architecture for financial servicesA five-layer pattern — channel, orchestrator, specialists, tools, governance — that actually maps onto how banks ship agents.
- Governing financial AI agentsModel risk, DORA, the EU AI Act, maker-checker, and why “the model said so” is not an audit trail.
Named production cases
Each case has a firm, a workflow, a public source, and a caveat. Architectures are reconstructed from those sources — not invented telemetry.
- JPMorgan Chase — LLM Suite and legal agentic workflowbanking
- JPMorgan Chase — Market-regime allocation agentsrisk
- Capital One — Chat Conciergebanking
- Klarna — Customer-service and refunds agentpayments
- Klarna — Agentic Product Protocolpayments
- Mastercard — Agent Suite, Agent Pay, and Agent Connectpayments
- Commonwealth Bank — Agentic payment-dispute toolbanking
- Anthropic — Financial-services agents (pitchbooks, credit, KYC)risk
- Coinbase — Payments MCP for agent stablecoin transactionspayments
- Visa — Embeddable AI financial assistantpayments
A five-layer architecture, not a four-box slide
Channel
App, employee suite, merchant plugin, or MCP server. Identity lives here.
Orchestrator
Plans the job and routes steps. Must not also hold the wire.
Specialists
Credit, KYC, disputes, legal, inventory — narrow and testable.
Tools
Cores, networks, wallets, document stores. Every call is an audit event.
Governance
Limits, allow-lists, dual control, model registry, kill switch.
Questions people actually search
- What is agentic AI in banking?
- Agentic AI in banking is software that can plan a multi-step job and call bank systems to do it — file a dispute, draft a credit memo, change a payment plan — rather than only answering a question. The live deployments are narrow: JPMorgan’s employee LLM Suite and legal workflow, Capital One’s Chat Concierge, CommBank’s payment-dispute tool. A general “AI teller” is still mostly a slide.
- How is agentic AI different from a chatbot in finance?
- A chatbot generates language. An agent is allowed to take an action with side effects: a refund, a core posting, a KYC update, a portfolio weight. That action needs the same identity, limit, and audit controls as a human in the same seat. If it cannot call a tool, it is a chatbot, whatever the landing page says.
- Which banks use agentic AI in production?
- Public, named examples as of September 2026 include JPMorgan Chase (LLM Suite at employee scale; Legal Agentic Workflow; research-grade regime agents), Capital One (Chat Concierge), Commonwealth Bank (payment disputes), plus fintech and network rails at Klarna, Mastercard, Visa, Coinbase, and Anthropic’s financial-services agents used in bank partnerships. Survey claims that “70% of banks use agentic AI” mix Copilot seats with production agents — this observatory lists the named ones.
- What are the risks of agentic AI in payments?
- Unauthorized or poorly authorized payments, weak mandate, missing replay, scheme-rule breaches, and no unwind path. A payments agent must log payer, mandate, amount, merchant, MCC, authorization time, model version, and the human who can reverse it. Networks are building Agent Pay-style authorization specifically because a free-form model must not be the cardholder.
- How do you govern financial AI agents?
- Put them in the model inventory. Split low-risk and high-risk review. Pin prompts and tool allow-lists. Require maker-checker above thresholds. Log every tool call. Keep a kill switch. EU AI Act and DORA add traceability for in-scope firms; they do not replace existing credit, conduct, or payments rules.