AI AgentFeb 2026
HPCL Lead Intelligence Agent
Autonomous B2B lead discovery for industrial fuel sales
Finalist · Productathon, IIT Roorkee
The problem
HPCL's industrial sales teams find new customers for furnace oil, bitumen and LDO by manually reading news and tender sites. Good leads are missed or reach the wrong regional officer too late.
How it works
- 01News & tender sources
- 02FastAPI + APScheduler
- 03Groq LLM scoring
- 04Dashboard · App · WhatsApp
What I built
- Scheduler that monitors Google News, Bing News and tender aggregators every 30 minutes.
- Entity extraction plus Groq (Llama 3.3 70B) analysis to infer which products a company needs.
- Lead scoring based on signal quality and source trust.
- Manager dashboard with KPIs, pipeline, team performance and regional insights (Next.js).
- Field app for sales officers (React Native / Expo) with deep links and status updates.
- Region-based WhatsApp alerts and daily summaries through Twilio.
Highlights
Fully autonomous
From discovery to the right officer's phone with no manual step in between.
Three surfaces
Web dashboard for managers, mobile app for the field, WhatsApp for instant alerts.
Hackathon finalist
Reached the finals of Productathon, a national-level hackathon at IIT Roorkee.
Stack
FastAPIGroq Llama 3.3Next.jsReact NativeTwilio
Next case study
Alcovia