Demo Mode
Fraud Operations Dashboard
Demo UI wired to the GraphQL API. Hosted demos run on free-tier services with limited features. For the full pipeline, run the local stack.
Transactions
0
Fraud Rate
0.0%
Pending Review
0
Auto Decisions
0
Latency p99
0ms
Model Accuracy
0.0%
Fraud queue
Live case list pulled from GraphQL with priority and status.
Network view
D3-based graph from suspicious subgraph queries.
Explainability
SHAP features with a plain-English summary.
No pending cases
Select a case to view details
Model Explanation0% risk
Loading explanation...
Increases risk Decreases risk
Transaction Network
Fraudster User Merchant Device
Investigator Copilot (RAG)
Local-first retrieval with optional Groq/Ollama generation
Ask about pipeline flow, model training, or infra setup.
RAG Tips
Ask about data flow, model training, or infra setup.
Examples:
- How do transactions reach Neo4j?
- What features are used for XGBoost?
- Where is drift detection implemented?
- How does the RAG index get built?