AI Agents
Decision Center
Each agent runs an Observe → Reason → Decide → Act loop continuously across customer signals.
5 agents · 9,153 decisions today · 67% automated
ORDA Trace
Observe → Reason → Decide → Act
ObserveSQL: customer_signals
Pulled engagement, support, NPS for 28,400 active customers
ReasonModel: m-churn-v4
Cross-referenced churn model v4.2 with recent service load and sentiment shifts
DecidePolicy: retention-v3
Selected 142 customers above 0.65 risk threshold; matched each to optimal retention play
ActAPI: orchestrator
Queued 102 actions for auto-execution, 40 for human approval (high-value)
Memory
Long+Short
Tools
SQL · API · Python
HITL
High-value only
Recent decisions
Churn Prevention Agent
VIP customer Aarav Mehta — engagement decay + 2 unresolved tickets
Action: Offer priority concierge + 15% loyalty credit
+$1.8K LTV preservedconf 91%
C-70128· 1h ago
Lapsed high-value Marcus Johnson — 47d silence
Action: Personalised win-back bundle via email
+$420 expectedconf 78%
C-50901· 3h ago
Renewal-window Amara Okafor — price-sensitive
Action: Lock-in 12-month plan at 8% discount
+$680 retained ARRconf 84%