Documentation covering the full Growth Chat development process, from discovery and design to implementation and deployment.
What is Growth Chat?
An AI-powered chat widget for a company landing page that replaces passive contact forms with intelligent conversations. It qualifies inbound leads, answers questions about the company’s services, and routes hot prospects to sales in real time.
This project is built as part of a course to explore real-world AI engineering — agent design, RAG, tool use, and production deployment.
Project Modules
This site. Architecture decisions, design docs, and development notes.
FastAPI backend powering the chat agent. Streaming POST /chat endpoint via SSE, with domain-driven structure across conversation, qualification, knowledge, handoff, and analytics.
Status: Scaffolded
React + TypeScript + Vite chat widget delivered as a <growth-chat> Web Component. Embeds on any host page with a single script tag.
Status: Scaffolded
Documentation
- Discovery Artifact — Problem space, open questions, and design considerations
- Problem Statement — Formal problem definition and MVP scope, the authoritative input to the PRD
- Product Requirements Document — Feature scope, requirements, technical stack candidates, and open questions for the MVP
- Considerations — Design considerations for chat behaviour, qualification signals, and human handoff
- User Personas — Target and negative visitor personas for the company website chat
- Architecture Decisions — Immutable records of key technical decisions (ADRs)
- Technical Requirements — Technical specifications covering system architecture, components, data models, and API
- Conversation Design — Conversation principles, stage model, persona rules, dialogue flows, and QA test suite for the chat agent