Autonomous Fraud Investigation Agent
Every
transaction
investigated.

Argus is a LangGraph-powered AI agent that autonomously investigates transactions, invokes specialized tools, and delivers a verdict with full justification — in under 2.5 seconds.

Run a Live Investigation
Named after Argus Panoptes —
the hundred-eyed giant
who never slept.
PRECISION  ·  100%
RECALL  ·  100%
EVAL CASES  ·  10 / 10
AVG LATENCY  ·  < 2.5s
100%
Precision
100%
Recall
10/10
Eval Cases Passed
<2.5s
Avg. Response Time
Transaction Payload POST /agent-investigate
Investigation Report
Awaiting investigation
Investigation Breakdown
Agent Reasoning
Methodology
How Argus
investigates
A fully autonomous pipeline. No human in the loop. From raw transaction data to a justified verdict — in seconds.
01
Transaction Received
The flagged transaction enters the LangGraph agent pipeline as a structured event payload, triggering the investigation graph.
02
Tool Invocation
The agent autonomously calls specialized tools — transaction history, merchant risk scoring, velocity checks, and geolocation plausibility analysis.
03
Reasoning Chain
Claude synthesizes all signals from tool outputs, constructing an evidence-based case for each finding with explicit confidence scoring.
04
Verdict Delivered
BLOCK · ALLOW · ESCALATE — with full reasoning and confidence score, delivered in under 2.5 seconds via the REST API.
Agent Stack
Investigation
pipeline
// Tool registry
check_transaction_history velocity · patterns
get_merchant_risk_score merchant intelligence
check_velocity rate analysis
check_geolocation_plausibility travel feasibility
get_cardholder_profile behavioral baseline
Technology
Built with
precision
LangGraph
Agent orchestration
Claude API
LLM reasoning
FastAPI
REST interface
PostgreSQL
Transaction store
LangSmith
Trace observability
Docker
Containerization
Railway
Production hosting
Pydantic v2
Schema validation
GitHub Actions
CI / CD pipeline