Banking Infrastructure
The layered machine inside every bank — core, payments, cards, identity, KYC, AML, fraud, risk, lending, treasury, settlement, custody, compliance, data and security — each layer with its standards and providers.
Strip any bank to its skeleton and you find the same stack: identity lets you open accounts; accounts hold balances; payments move them; settlement makes movement final; assets are what everything ultimately references. Around that spine sit 16 operational layers — each mapped below to its standards and vendor ecosystem.
Core Banking
The system of record that maintains customer accounts, balances, products, and the general ledger, processing deposits, withdrawals, and interest calculations. It is the central platform other banking systems read from and write to, historically batch-based and increasingly cloud-native and API-driven.
Standards
ISO 20022 · BIAN
Providers
Temenos · Thought Machine · Mambu · FIS
for agents: With structured access to the core, an AI agent could query balances and product terms, reconcile ledger entries, and orchestrate account servicing tasks without screen-scraping legacy interfaces.
Payments
Handles the initiation, routing, clearing, and confirmation of money movement across rails such as ACH, SEPA, wires, and real-time payment networks. It translates between internal ledgers and external schemes, managing message formats, cut-off times, and returns.
Standards
ISO 20022 · ISO 8583 · SWIFT gpi
Providers
ACI Worldwide · Form3 · Adyen · Stripe
for agents: An AI agent could select the optimal rail for a given transfer, track payment status end-to-end, and automatically investigate failed or returned payments.
Cards
Covers card issuing and processing: creating physical and virtual cards, authorizing transactions in real time, and managing networks, interchange, and disputes. It connects the bank to card schemes such as Visa and Mastercard through issuer processors.
Standards
PCI DSS · EMV · ISO 8583
Providers
Marqeta · Fiserv · Global Payments · i2c
for agents: An AI agent could issue scoped virtual cards, set and adjust spend controls programmatically, and triage authorization declines or disputes on behalf of customers.
Accounts
Manages the lifecycle of customer accounts — opening, structuring, numbering, and closing — along with sub-ledgers, virtual accounts, and account metadata. Banking-as-a-service platforms expose this layer as APIs so third parties can embed accounts into their products.
Standards
ISO 13616 · FDX API · ISO 20022
Providers
Unit · Treasury Prime · Synctera · Modern Treasury
for agents: An AI agent could open and configure accounts, allocate virtual sub-accounts for specific purposes, and keep account records synchronized across connected systems.
Identity
Authenticates and authorizes users and systems, managing credentials, sessions, multi-factor authentication, and consent for data sharing. In open banking it also governs how third-party providers gain delegated access to customer accounts.
Standards
OpenID Connect · OAuth 2.0 · FAPI
Providers
Okta · Ping Identity · Auth0 · ForgeRock
for agents: An AI agent could act on a customer's behalf through delegated, scoped credentials, with every action tied to an auditable consented identity rather than shared passwords.
KYC / KYB
Verifies the identity of individuals (Know Your Customer) and the legitimacy of businesses (Know Your Business) at onboarding and on an ongoing basis. It combines document verification, database checks, beneficial-ownership resolution, and risk scoring before an account is approved.
Standards
FATF Recommendations · eIDAS
Providers
Alloy · Onfido · Jumio · Middesk
for agents: An AI agent could assemble and pre-validate onboarding evidence, resolve beneficial-ownership structures, and route only genuinely ambiguous cases to human reviewers.
AML
Monitors transactions and customer behavior for money-laundering and terrorist-financing patterns, screening against sanctions and politically-exposed-person lists. It generates alerts and supports the filing of regulatory reports such as suspicious activity reports.
Standards
FATF Recommendations
Providers
NICE Actimize · ComplyAdvantage · Napier · Chainalysis
for agents: An AI agent could enrich and pre-investigate AML alerts with contextual data, drastically reducing false-positive workload while documenting its reasoning for auditors.
Fraud
Detects and prevents unauthorized or deceptive activity — account takeover, payment fraud, scams, and synthetic identities — using device signals, behavioral biometrics, and real-time scoring. It sits inline on transactions and logins, deciding to approve, challenge, or block within milliseconds.
Standards
3-D Secure · PSD2 SCA
Providers
Feedzai · BioCatch · Sift · Forter
for agents: An AI agent could correlate fraud signals across channels in real time, explain why a transaction was flagged, and adapt friction dynamically instead of applying blanket blocks.
Risk
Measures and manages credit, market, liquidity, and operational risk, feeding capital adequacy calculations, stress tests, and exposure limits. It aggregates positions and counterparty data so the bank can price risk and satisfy prudential requirements.
Standards
Basel III · IFRS 9 · BCBS 239
Providers
Moody's Analytics · SAS · Oracle Financial Services
for agents: An AI agent could run on-demand scenario analyses, monitor limit utilization continuously, and surface emerging concentrations of risk before they breach thresholds.
Lending
Runs the credit lifecycle: origination, underwriting, decisioning, servicing, and collections for products from consumer loans to commercial credit. It combines application data, credit bureau inputs, and scoring models to price and approve loans.
Standards
MISMO
Providers
nCino · Blend · LoanPro · Zest AI
for agents: An AI agent could pre-assemble underwriting files, explain credit decisions to applicants, and proactively manage servicing events like modifications or delinquency outreach.
Treasury
Manages the bank's own liquidity, funding, cash positions, and interest-rate exposure across currencies and entities. It forecasts cash flows, executes funding decisions, and ensures obligations can be met as they fall due.
Standards
ISO 20022 · SWIFT gpi
Providers
Finastra · Moody's · FIS
for agents: An AI agent could maintain a continuously updated cash-position forecast, flag funding gaps early, and propose optimized intraday liquidity allocations.
Settlement
Finalizes the transfer of funds and securities between institutions, achieving finality through mechanisms such as real-time gross settlement, netting, and delivery-versus-payment. Financial market infrastructures operate this layer and manage the associated counterparty risk.
Standards
ISO 20022 · PFMI
Providers
CLS · DTCC · Euroclear · Clearstream
for agents: An AI agent could track settlement status across venues, predict and resolve fails before deadlines, and reconcile settled positions against internal ledgers automatically.
Custody
Safekeeps client assets — securities, cash, and increasingly digital assets — handling asset servicing tasks such as corporate actions, income collection, and proxy voting. Custodians maintain segregated records of ownership distinct from the bank's own balance sheet.
Standards
ISO 20022 · SOC 2
Providers
BNY · State Street · Fireblocks · BitGo
for agents: An AI agent could monitor corporate actions and entitlement deadlines across custodied portfolios and initiate instructed responses within governed policy limits.
Compliance
Ensures the bank's activities conform to laws, regulations, and internal policies, covering regulatory reporting, communications surveillance, and policy management. It tracks regulatory change and evidences adherence to supervisors during examinations.
Standards
Basel III · GDPR
Providers
Wolters Kluwer · NICE Actimize · Smarsh · Behavox
for agents: An AI agent could map new regulatory texts to affected policies and controls, draft impact assessments, and assemble examination-ready evidence packs on demand.
Data
Aggregates, governs, and distributes the bank's data — transactions, customers, reference data — across warehouses, lakes, and APIs for analytics and open-banking sharing. Data quality, lineage, and access controls in this layer determine what every other layer can trust.
Standards
BCBS 239 · FDX API · ISO 20022
Providers
Snowflake · Databricks · Plaid
for agents: An AI agent could answer complex questions over governed banking data with lineage-aware citations, instead of relying on stale extracts or manual report requests.
Security
Protects banking systems and data through encryption, key management, hardware security modules, network defense, and threat detection. It underpins every other layer, since payment credentials, keys, and customer data are primary attack targets.
Standards
ISO 27001 · NIST Cybersecurity Framework · PCI DSS
Providers
Thales · Entrust · CrowdStrike · Palo Alto Networks
for agents: An AI agent could correlate security telemetry with banking context to triage incidents, verify control coverage, and rotate or revoke credentials the moment compromise is suspected.
Approximate reference data from public sources · seed-0.1 · 2026-08-28 · not real-time, not advice. See data & methodology.