← Back to All Projects
AI / ML

Real-Time TME Assist Platform

Just Dial — Internal AI Product

Built a real-time AI-powered TME-Assistance platform that provides live insights, suggestions, and intelligence to sales tele-marketing executives during customer calls. Multi-agent architecture with dedicated agents for call monitoring, product recommendations, category analysis, and post-call review.

Real-TimeLive Call Insights
Multi-AgentAI Architecture
8,000+TMEs Supported
Agentic AILangGraphWebSocketNode.jsPythonReact.jsRAG

The Challenge

Sales TMEs at Just Dial handle thousands of live customer calls daily. During these calls, they need to make quick decisions — which products to pitch, how to handle objections, how to steer conversations toward a successful deal or appointment booking. Without real-time support, TMEs rely on experience alone, leading to missed opportunities and inconsistent performance.

The goal was to build a platform that listens to live calls and provides real-time guidance, buying intent signals, product suggestions, and actionable insights — all displayed on a dashboard the TME can glance at during the conversation.

Key Features

📡

Real-Time Call Guidance

Live suggestions on next steps, current call status, and how to steer the conversation toward a successful appointment booking or deal close.

🎯

Buying Intent Analysis

Continuous analysis of customer signals to gauge buying intent and alert TMEs when a customer is ready to convert.

📦

Product Pitch Recommendations

AI-driven product suggestions based on the ongoing conversation context, customer profile, and business category.

🛡️

Objection Handling

Real-time insights on how to address customer objections with relevant talking points and competitor comparisons.

🏷️

Category Suggestions

Automatic analysis of business categories the customer belongs to, with suggestions for reclassification if needed.

📊

Post-Call Analysis

After call completion, a dedicated agent summarizes the interaction and provides insights on what went wrong and what could have been done better.

Technical Approach

01

Multi-Agent Architecture

Main orchestrator agent monitors call status and coordinates with specialized agents — Product Expert, Category Expert, and Post-Call Analysis — each handling dedicated tasks autonomously.

02

Real-Time Processing

WebSocket-based live audio streaming with real-time transcription and analysis, delivering insights to the TME dashboard with minimal latency.

03

TME Dashboard

React-based real-time dashboard displaying live insights, suggestions, intent scores, and recommended actions that update as the conversation progresses.

Outcomes

  • Live call intelligenceTMEs receive actionable guidance during every customer interaction
  • Multi-agent AIdedicated agents for product, category, objection handling, and post-call analysis
  • Buying intent trackingreal-time scoring helps TMEs prioritize conversion opportunities
  • Post-call insightsautomated analysis highlights improvement areas for each call
  • Consistent performancedata-driven guidance reduces dependency on individual TME experience

Interested in building real-time AI platforms?

I've built live AI assistance systems for sales teams at enterprise scale.

← PreviousTyre Information Management System
Next ProjectTele-Marketing Call Audit Automation