RI

AGT-1007 · Financial Crime

Fraud Detection Agent

Scores card and UPI transactions in-flight for fraud typologies.

critical riskcompliant

Trust score

85

grounding, fairness, oversight

Requests / 30d

24,575

model In-house · GraphNN v4

Human approval

Not required

data class Restricted

Last audit

2026-06-04

version v2.1

Agent architecture

Request path from the surface the customer or staff member touches, through the governance gateway, to the model and its wired systems

Entry surfaces

Contact Centre Desktop

 

Business application · Genesys

Governance gateway

Truint policy enforcement point

9-stage filter chain · p99 54 ms

Evidence ledger writer

Hash-chained, RFC-3161 anchored

Agent runtime

Fraud Detection Agent

AGT-1007 · v2.1 · trust 85

Azure OpenAI · gpt-4.1

 

LLM / model endpoint · Central India

Tools & data

payments.transfer

 

Tool / function · Restricted

Bureau feed (CIBIL)

 

Data source · Restricted

Enforcement chain on every request

  1. 1Identity & entitlementCaller identity resolved; agent entitlement checked against the registry record.
  2. 2Kill switchEstate and per-agent revocation state for AGT-1007 evaluated before any spend.
  3. 3Manifest checkOnly integrations on the signed manifest below are reachable.
  4. 4Budget & rateToken, cost and concurrency budgets for Financial Crime.
  5. 5Injection screenPrompt and retrieved context screened for untrusted-content patterns.
  6. 6DLP & maskingRestricted payloads masked or tokenised inline.
  7. 7Policy decisionCompiled policy bundle evaluated; allow, warn or block.
  8. 8Human in the loopAdvisory output only — no restricted action is executable.
  9. 9Ledger barrierDecision hash-chained into the audit ledger before the response is released.

Registry record

Source of truth for this agent

Registry ID
AGT-1007
Business owner
R. Iyer
Department
Financial Crime
Model endpoint
In-house · GraphNN v4
Version
v2.1
Data classification
Restricted
Human-in-the-loop
Advisory output only

Trust composition

Grounding & citations89
Fairness79
Security posture87
Oversight coverage78

Wired integrations

Signed manifest — nothing else is reachable

Azure OpenAI · gpt-4.1connected

LLM / model endpoint · Microsoft Azure · scopes chat.completions, embeddings

payments.transferdegraded

Tool / function · Payments hub · scopes mt103, neft, rtgs

Contact Centre Desktopconnected

Business application · Genesys · scopes copilot.suggest

Bureau feed (CIBIL)connected

Data source · TransUnion CIBIL · scopes report.pull

Policies in force

POL-001 · No customer PII outside IndiaPOL-004 · No customer data sent to public LLMPOL-005 · High-value transaction requires human approvalPOL-006 · PII masking requiredPOL-009 · Toxicity detection

Recent decisions

Live gateway stream filtered to this agent

11:39:51emp-20210warn

Score this transaction of ₹18,40,000 to a new beneficiary.

POL-005 · High-value transaction approval · risk 55 · 874 ms

11:38:34emp-20435warn

Score this transaction of ₹18,40,000 to a new beneficiary.

POL-005 · High-value transaction approval · risk 42 · 364 ms

11:35:25emp-20004warn

Score this transaction of ₹18,40,000 to a new beneficiary.

POL-005 · High-value transaction approval · risk 49 · 1459 ms

Open risks

Drill into the investigation

No open risks on this agent.

Ledger entries

Hash-chained decision records

No ledger entries surfaced in the current window.

Governance obligations

  • Quarterly attestation by R. Iyer under MB-04
  • Advisory output only — no autonomous execution of restricted tools
  • Kill switch tested; revocation propagates in under 30 seconds
  • Evidence retained ten years for regulator inspection