Not based on a specific customer deployment. The company, figures, decisions, and outcomes below are a constructed example that shows how Intended’s mechanism applies to this kind of problem. They are not measured results from a named customer. Where we publish a real, attributed customer outcome, we will say so explicitly.
Governing AI-Initiated Fraud Actions at a Digital Bank
Consider a digital-first bank whose AI agents handle transaction monitoring, account actions (freezes, limit changes), KYC verification, and customer service automation across a high volume of daily transactions.
01 · the challenge
The kind of problem this addresses.
02 · how it works
See the difference.
AI detects suspicious pattern
3 rapid transfers to new payee, total $47,000
AI freezes account
No graduated response, no risk scoring
Legitimate customer locked out
Wire transfers for home purchase
Customer complaint filed
No audit trail to explain AI decision
03 · the solution
What they deployed.
- — Started from Intended's financial-services policy pack and tuned it for fraud, KYC, and account actions
- — Graduated response by policy: escalate to step-up verification for MEDIUM and HIGH risk, allow an automatic freeze only for named CRITICAL patterns such as credential stuffing
- — The agent sends customer context (for example, prior large transfers) with each request, and policy rules read those fields
- — Every account action needs an authority decision first, recorded in a tamper-evident, hash-chained audit trail
- — Agents call Intended through the SDK before acting in the core banking system, the fraud platform, or the notification service, and act with their own credentials only after an ALLOW. Intended never holds the bank's credentials or calls these systems
04 · implementation
From zero to governed.
Week 1
Map
Catalogue the agents' account actions across fraud, KYC, and customer service.
Week 2
Configure
Install the financial-services pack, then add graduated-response rules and approver roles. Check them with policy simulation.
Week 3
Integrate
Add the SDK call ahead of each core banking, fraud platform, and notification action. The services that execute can verify the authority token before acting.
Week 4
Enforce
Enable enforcement. Suspicious patterns that are not clear-cut get step-up verification instead of a hard freeze.
05 · illustrative outcomes
What this is designed to deliver.
Modeled figures for this scenario — what the workflow above is built to achieve, not measured results from a named customer.
0%
Account actions gated
Every freeze, limit change, and block needs a decision
0%
Actions with audit trail
Tamper-evident, hash-chained record
0
Graduated response tiers
Step-up auth before any hard freeze
0 min
Max token lifetime
Signed and single-use
06 · decision replay
Example decisions, full trace.
Sample decision records that show the shape of the evidence Intended produces. Illustrative, not drawn from a live customer’s logs.
banking.fraud.transaction-reviewRISK: 68/100ESCALATEAI flags 3 rapid transfers totaling $47K to new payee -- customer has home purchase history
Resolved by: Graduated response: step-up verification (customer verified)
banking.fraud.account-freezeRISK: 96/100ALLOWAI detects credential stuffing attack on account, 47 failed login attempts from 12 IPs
Resolved by: Policy: auto-freeze for credential stuffing pattern (CRITICAL risk)
banking.kyc.identity-reverificationRISK: 52/100ALLOWAI triggers KYC re-verification for customer with address change + large withdrawal
Resolved by: Policy: KYC re-verification auto-approved for qualifying triggers
banking.fraud.transaction-blockRISK: 100/100ALLOWAI blocks $180K wire to sanctioned entity -- OFAC match on beneficiary
Resolved by: Policy: auto-block for OFAC sanctions match
banking.customer.limit-changeRISK: 44/100ESCALATEAI increases daily transfer limit from $10K to $50K based on customer request
Resolved by: Relationship Manager (approved after a customer call)
the takeaway
Fraud controls usually trade off against customer experience: catch more, and you freeze more legitimate accounts. Intent classification plus graduated response breaks that trade-off — step-up auth instead of a freeze, verification instead of a lockout — and every action leaves a tamper-evident audit trail.
Why this pattern matters — not a customer quote.
Start governing AI banking operations
Free to start. See every AI decision from day one.