Skip to main content
← Back to projects

Agentic CPMS

Architecture Phase

An open-source Charge Point Management System where AI agents handle the operational complexity that currently requires large operations teams: self-healing charge points, predictive maintenance, dynamic pricing, and automated incident response.

The EV charging software market is projected to reach $8-21B by 2033. Current CPMS platforms are monolithic, reactive, and expensive. An agentic approach could automate 80%+ of operational tasks.

Architecture covers OCPP 1.6/2.0.1 gateway (Go), core business logic (TypeScript), Next.js frontend, and 8 specialized AI agents. 24 epics and 114 user stories mapped out across 5 development phases.

Tech Stack

GoTypeScriptNext.jsPostgreSQLTimescaleDBNATSDockerKubernetes

Key Features

  • OCPP 1.6 & 2.0.1 WebSocket gateway
  • Self-healing charge point agent
  • Predictive maintenance agent
  • Dynamic pricing optimization agent
  • Smart load balancing agent
  • OCPI 2.2.1 roaming integration
  • Customer support AI chatbot
  • Automated firmware orchestration

Challenges

OCPP protocol complexity (dozens of edge cases), real hardware testing requirements, and the need for OCA certification. The agentic angle is the key differentiator — no CPMS on the market offers AI-native self-healing and predictive operations.