function showPage(p){ document.querySelectorAll('.page').forEach(el=>el.classList.remove('active')); document.querySelectorAll('.nav-item').forEach(el=>el.classList.remove('active')); const pg=document.getElementById('page-'+p); if(pg) pg.classList.add('active'); const navItems=document.querySelectorAll('.nav-item'); navItems.forEach(el=>{ if(el.getAttribute('onclick')?.includes("'"+p+"'")) el.classList.add('active'); }); if(p==='graph-viz') _initGraphViz(); } SENTINEL — Financial Crime Intelligence
FINANCIAL CRIME INTELLIGENCE
NO DATA
RISK: --
IDLE
TRANSACTION INTELLIGENCE ORB — ACO + MULTI-AGENT SWARM
--
AWAITING ANALYSIS
0
Total Cases
0
Critical
0
High
0
Medium
0
Low
TRANSACTIONS CSV *
📊
Upload Transactions
txn_ref, from_account_id, to_account_id, amount, txn_type, txn_timestamp, geo_lat, geo_lon
No file loaded
ACCOUNTS CSV
👤
Upload Accounts
id, account_number, holder_name, holder_type, city, geo_lat, geo_lon
No file loaded
TRADES CSV (optional)
📈
Upload Trades
account_id, security_symbol, trade_type, total_value, trade_timestamp
No file loaded
MARKET EVENTS CSV (optional)
📰
Upload Market Events
security_symbol, event_type, event_timestamp, is_price_sensitive
No file loaded
0 cases
Case IDCrime TypePatternSeverityConfidenceAccountsStatus
Run analysis to see cases
REAL-TIME ACO ANT COLONY + MULTI-AGENT SYSTEM
NODES 0
EDGES 0
SUSPICIOUS 0
FILTER MIN RISK 0% FOCUS NODE none
// SENTINEL CONFIGURATION
AI Analyst — API Keys
Keys stored in your browser only. Never sent anywhere except the chosen AI provider. Groq is free.
Groq FREE DEFAULT
Llama 3.3 70B · Very fast · Free tier
→ Get free key
Anthropic
Claude Haiku · Fast · ~$0.001/call
→ Get key
OpenAI
GPT-4o mini · Reliable · ~$0.001/call
→ Get key
Gemini
Gemini 1.5 Flash · Free tier available
→ Get key
Active provider: none · SAR generation + CSV schema mapping + case Q&A
Jurisdiction & Thresholds
Switch regulatory context. Adaptive mode auto-calibrates from your data after analysis.
Graph Analytics Status
Run analysis to populate graph metrics.
Entity Resolution Status
Run analysis to populate entity clusters.
⚡ v5: Neural Network + Random Forest
Trained from human-labeled cases. Approve/reject cases to build the training corpus.
No labels yet. Approve or reject cases to train the neural classifier.
Architecture: 23→32(ReLU)→16(ReLU)→1(Sigmoid) + RandomForest(25 trees)
Training: Mini-batch gradient descent + momentum + L2 regularisation + backpropagation
Splits: CART Gini impurity | Feature selection: sqrt(n_features) per tree
| SPEED | IDLE 0 ants | ACO:INIT NN:COLD
ITER 0/100  | MB-SMURF 0