Context

• The sales team manages 100–300 active deals and conducts dozens of customer conversations every week across multiple channels.
• Sales managers manually qualify leads, review conversations, and prepare weekly reports for management.
• The company wanted to automate lead qualification, improve sales quality, and gain better visibility into pipeline performance.

Problem

• Time-Consuming Manual Qualification: Sales reps spent 2–4 hours per day manually qualifying leads and updating CRM records, diverting time from actual selling.
• Unmonitored Quality & Missed Opportunities: Managers had no scalable, consistent method to review multi-channel conversation quality or detect deal risks and missed cross-sell/up-sell opportunities.
• Labor-Intensive Reporting: Compiling weekly executive reports took 3–5 hours of manual data collection across systems, resulting in delayed and non-actionable insights.
• Pipeline Visibility Bottleneck: Managing 100–300 active deals made it difficult for leadership to track overall deal health and pipeline performance in real time.

Solution

Built an end-to-end AI sales ecosystem consisting of three interconnected workflows operating as a unified intelligence system:

• Interactive BANT Qualification Chatbot: Automatically engages inbound prospects, qualifies them based on Budget, Authority, Need, and Timing (BANT), and logs enriched lead records into Zoho CRM.
• Automated Conversation & Deal Health Analytics: Continuously evaluates multi-channel customer interactions using Gemini API, identifying deal risks, sales gaps, and cross-sell opportunities.
• Automated Executive Reporting: Aggregates weekly pipeline metrics and delivers concise executive summaries with strategic recommendations directly via Telegram.

Implementation Process

1. Prompt design

• Designed specialized system prompts to qualify inbound leads using the BANT (Budget, Authority, Need, Timing) framework.
• Built prompts to evaluate conversation history, calculate deal health scores, and identify key risks/recommendations.
• Enforced strict JSON output schemas with structured parsers to feed clean data into downstream databases and reports.

2. API integration