Skip to main content
MP
AI Agents · Logistics · Fleet Automation2026

DISPATCHMIND

“Autonomous AI dispatch co-pilot for logistics and fleet operators.”

Role

Lead Developer

Stack
PythonGoogle ADKGemini 2.5 ProFastAPIGoogle Maps APIOpen-MeteoTelegram Bot APIFirestoreCloud RunMCP
Status

Live

Overview

DispatchMind is an autonomous multi-agent dispatch system that automates order intake, route planning, live fleet monitoring, exception recovery, and customer communications for logistics operators.

Built with Google Agent Development Kit (ADK) and Gemini models, the system ingests invoices and e-way bills via multimodal vision OCR, computes dynamic route optimizations using Google Maps Routes and Open-Meteo weather APIs, and autonomously recovers from transit exceptions without human intervention.

Deployed to Google Cloud Run in asia-south1 (Mumbai) with Model Context Protocol (MCP) tool integration and Telegram driver coordination.

Screenshots

Key Features

→
Multi-Agent Orchestration - Autonomous agent cluster coordinating Intake, Planner, Monitor, Recovery, Comms, POD, and Feedback agents via Google ADK.
→
Multimodal Order Ingestion - Extracts line items, delivery addresses, and weights from invoices and e-way bills using Gemini vision.
→
Dynamic Route Planning - Real-time route optimization combining Google Maps Routes API with live Open-Meteo weather data.
→
Autonomous Exception Recovery - Detects route delays, road hazards, or vehicle breakdowns and autonomously reassigns drivers and updates schedules.
→
Telegram & Customer Comms - Direct driver messaging via Telegram Bot API and automated SMS/email customer notifications.
→
Proof of Delivery (POD) - Computer vision verification of delivery photos with automated recipient confirmation.

Tech Deep Dive

DispatchMind leverages Google ADK to orchestrate seven specialized agents running on FastAPI in Google Cloud Run. The tool layer connects to Google Maps Routes API, Open-Meteo, and Telegram via Model Context Protocol (MCP). Firestore maintains live fleet telemetry and shipment state machines, while automated retry and escalation loops ensure operational self-healing.

Next ProjectTONELENS →