An advanced, AI-powered customer support ecosystem that leverages an autonomous AI agent to resolve customer queries by interacting with legacy systems via n8n workflows.
This project is a decoupled full-stack application consisting of:
- Frontend: Next.js (TypeScript) with Tailwind CSS, authenticated via Clerk.
- Backend: Java Spring Boot microservice managing ticket lifecycles.
- AI Layer: LangChain4j integration with Google Gemini (or OpenAI) to create an autonomous Agentic AI.
- Automation Layer: n8n workflows acting as the bridge between the AI agent and legacy ERP/Database systems.
- Autonomous Agentic AI: The agent doesn't just chat; it reasons. It decides whether to check order status, initiate a refund, or escalate to a human.
- Legacy System Integration: Uses n8n as an abstraction layer to interact with older "legacy" APIs and databases safely.
- Real-time Ticket Tracking: Customers can watch the AI "think" and resolve their tickets in real-time.
- Human-in-the-Loop: When the AI detects complex issues, it automatically escalates the ticket to an Admin Dashboard for human intervention.
- Secure Authentication: OIDC-compliant auth using Clerk for both customers and administrators.
- Frontend: Next.js 15+, React 19, Tailwind CSS, Clerk Auth.
- Backend: Spring Boot 3.x, Spring Data JPA, Spring Security.
- AI/LLM: LangChain4j, Google Gemini / OpenAI.
- Automation: n8n (Workflow Automation).
- Database: PostgreSQL.
- DevOps: Docker, Docker Compose.
- Node.js & npm
- JDK 17+
- Docker & Docker Compose
- Clerk API Keys
- n8n instance (local or cloud)
- Configure your
src/main/resources/application.propertieswith your Gemini API key and database credentials. - Run the Spring Boot application:
./mvnw spring-boot:run
- Navigate to the
clientdirectory. - Create a
.env.localfile (seeclient/.env.local.example). - Install dependencies and start:
npm install npm run dev
Import the JSON workflows located in the /n8n-workflows directory into your n8n instance.
- Backend: Hosted on Render (connected to PostgreSQL).
- Frontend: Hosted on Vercel.
- Live Demo: https://agentic-support-system-inky.vercel.app/
The admin dashboard is restricted to users defined in the NEXT_PUBLIC_ADMIN_EMAIL environment variable. Admins can view all escalated tickets and provide final resolutions.
Built with ❤️ to demonstrate the power of Agentic AI in Enterprise Support.