Total Calls Today 0 ↑ 12.4%

Active Sessions 0 ↑ 3.2%

Conversion Rate 0% ↑ 1.8%

Fallback Rate 0% ↓ 0.5%

AI Agents

Customer Intelligence Agent Active · v2.1.0
NLU Agent Active · v3.0.2
Eligibility Agent Active · v1.8.5
Sales / Persuasion Agent Active · v1.5.3
Compliance Agent Active · v4.0.2
Handoff Agent Active · v2.0.1
Orchestration Support Active · v1.1.0

You are the Customer Intelligence Agent for ArthOps Banking Voice AI. Your role is to: 1. Maintain deep context of the customer profile 2. Track sentiment and intent history across previous turns 3. Provide personalized recommendations based on past banking behavior 4. Surface relevant product milestones seamlessly Guidelines: - Analyze real-time interactions for tone/frustration - Identify cross-sell or up-sell opportunities dynamically - Feed contextual variables to other agents

Call Simulator

User Input

›

Orchestrator

›

NLU

›

Customer Intel

›

Eligibility

›

Compliance

›

Sales

›

Compliance

›

Handoff (Opt)

›

Orchestrator

ArthOps AI
Welcome to HDFC Bank ArthOps. I am your priority AI assistant. I can help you with your account, loans, or credit cards. How may I assist you today?

Blocklist Management

Phrase	Scope	Action

Live Compliance Simulation

Create New Offer

Active Campaigns

Campaign Name	Status	Acceptance Rate	Actions

Monitored Agents

Select an Agent

● Active v1.0.0

Owner: --

Avg Latency

-- ms

Success Rate

-- %

Alert Errors (30m)

-- events

Invocations (30m) --

Execution Timeline

Last 60 Minutes

Recent Actions Log

Time	Action	Status	Latency

Tiered User Processing Model

Live Activity

Active Users 12,482

Avg Latency 420ms

Immediate processing for high-intent, engaged users.

Queued Users 84,103

Next Batch 12 mins

Batch processing (every 1-3 hrs) for medium engagement.

Stored Profiles 2.4M

Compute Cost Minimal

Low-cost storage, reactivated only on specific intent signals.

Multi-stage Decisioning Pipeline

Event Ingestion

➔

Early Suppression

➔

Routing logic

➔

Agent Execution

Events Dropped (Early) 41.2%

LLM Calls Avoided 1.8M / day

Avg Pipeline Latency 45ms

Context Optimization & Efficiency

Dynamic Field Pruning Removes stale CRM data to reduce LLM token usage

Historical Summary Compression Summarizes older chats into dense vector blocks

-32% Tokens

Cost Savings: ~$1,450 / day

Tokens Processed: 14B / mo

Latency Gain: +120ms faster gen

Dark Time Engine Rules

Manage session suppression, regulatory blackout periods, and frequency capping.

Rule Name	Type	Condition / Trigger	Status	Suppressed (24h)
Transaction Flow Block	Session	User is on /payment or /checkout screens	Active	42,109
Post-Action Suppression	Cooldown	Within 24h of Loan Approval event	Active	8,451
Regulatory Night Blackout	Global	21:00 to 09:00 IST (All Telemarketing)	Active	145,820
"Not Interested" Hard Hold	Profile	Feedback intent = "Stop Marketing"	Active	3,299

Operations Center

Call Volume Over Time

Conversion Funnel

Recent Alerts

Agent Control

AI Agents

Call Simulator

Compliance Panel

Blocklist Management

Live Compliance Simulation

Campaigns & Offers

Create New Offer

Active Campaigns

Agent Observability Dashboard

Select an agent to view metrics

Monitored Agents

Select an Agent

Execution Timeline

Recent Actions Log

Intelligent Decisioning Architecture

Manage dynamic processing tiers, pipelines, context optimization & suppression rules

Tiered User Processing Model

Multi-stage Decisioning Pipeline

Context Optimization & Efficiency

Dark Time Engine Rules

Modal

Notifications

Operations Center

Call Volume Over Time

Conversion Funnel

Recent Alerts

Agent Control

AI Agents

Call Simulator

Compliance Panel

Blocklist Management

Live Compliance Simulation

Campaigns & Offers

Create New Offer

Active Campaigns

Agent Observability Dashboard

Select an agent to view metrics

Monitored Agents

Select an Agent

Execution Timeline

Recent Actions Log

Intelligent Decisioning Architecture

Manage dynamic processing tiers, pipelines, context optimization & suppression rules

Tiered User Processing Model

Multi-stage Decisioning Pipeline

Context Optimization & Efficiency

Dark Time Engine Rules