Building a Real-Time E-Commerce Support Platform with an AI Agent Team
Ever abandoned a shopping cart because you couldn't quickly ask, "When will my package arrive?" Industry averages put cart abandonment at around 70% , and slow answers are a major cause: contact forms, ticket portals, and emails that get a reply hours later. That problem led me to build Aura Gear , a real-time customer support platform for e-commerce. It's also an experiment in how an engineer…
Online shoppers often abandon their carts due to slow responses, with cart abandonment rates around 70%. Traditional contact forms, ticket portals, and emails typically result in replies hours later. To tackle this issue, Aura Gear was developed - a real-time customer support platform specifically designed for e-commerce. This platform also serves as an experiment in collaborative engineering and AI team building, from initial specifications to ongoing operations.
Aura Gear comprises two main components: a customer concierge widget and a support agent command center. The widget allows for instant order tracking within the online store, providing an easy-to-use interface for shoppers. The agent command center is a dashboard for the support team, featuring multiple panes displaying relevant information during customer interactions.
The system is built using a lightweight glassmorphic UI (under 150KB) with no build step, and a Python 3.10+ backend powered by FastAPI and Uvicorn. The asynchronous WebSocket manager handles real-time communication, while a thread-safe asynchronous chat store manages conversations, messages, and agent assignments. Data is stored and retrieved with session recovery to ensure continuity.
The project was developed using an AI agent team consisting of six specialized agents, each responsible for a different role: product manager, full-stack engineer, QA engineer, DevOps engineer, agent operations specialist, and incident responder. Their respective responsibilities include defining product specifications, creating API contracts, implementing backend functionality, performing unit and integration tests, managing deployment using Docker and CI/CD, developing agent operations, and handling incident response.
Context management plays a critical role in ensuring AI agents perform optimally. To achieve this, the project uses modular documentation, including AGENTS.md, architecture.md, permissions.md, security/ai-data-policy.md, and ops/runbook.md. This well-organized documentation serves as the agent's working memory, providing clear boundaries and enabling dependable results.
Testing and CI/CD processes are streamlined with a single command running all tests, including unit tests, integration tests, and health endpoint checks. The GitHub Actions pipeline consists of three stages: CI, CD Dev, and CD Prod. Each stage includes specific tasks, such as running flake8, unit and integration tests, building the container, deploying and verifying the dev environment, and manually approving the promotion to production.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
