AutoMeet AI - Zero-Touch Lead Scheduling

Our project builds an agentic coordination system that automates lead follow-up and scheduling through an integrated pipeline of modular agents. The system includes two conversational agents (Researcher and SDR) for data enrichment and outreach, plus a central Coordinator agent that ingests inbound replies, uses a watsonx LLM to classify intent, checks calendar availability, and takes action (book meetings, send templated replies, update Airtable records). The Coordinator is resilient and production-ready: it uses Redis for persistent idempotency, the Google Calendar FreeBusy API for real availability checks, and Jinja2-powered reply templates with an LLM fallback for bespoke responses. The backend runs as a FastAPI service containerized for IBM Cloud Code Engine and pulls secrets from secure environment variables; Airtable houses lead state and interaction logs. The design prioritizes accuracy, auditability, and safe operations (signature-verified webhooks, secret rotation). For the hackathon demo we provide end-to-end flows: incoming email → intent classification → booking or follow-up → Airtable update, plus logs/screenshots. This delivers faster, reliable lead conversion and a repeatable foundation for future agentic workflows.