AI Chat Application Suite
A collection of purpose-built conversational AI applications — including a contextual document chat, a multi-persona AI assistant, and a team knowledge-base chat interface.
Solo Builder
2024
The Problem
Generic chat interfaces (ChatGPT, Claude) are powerful but context-free. Most real-world use cases require chat grounded in specific documents, data sources, or organisational knowledge — with memory, persona, and context management built in.
The Approach
Built multiple chat applications targeting specific use cases: (1) Document Chat — chat with any uploaded PDF/DOCX with citation-grounded responses; (2) Multi-Persona Assistant — configurable AI personas for different professional contexts; (3) Knowledge Base Chat — RAG-powered chat over a structured knowledge base with source attribution.
Architecture / System Design
Each application uses a RAG architecture with document chunking, embedding, and vector retrieval. Conversation memory managed via sliding window + summarisation for long sessions. Persona configuration via system prompt engineering with guardrails. All three deployed as separate live applications.
Outcomes
Three separate live deployed chat applications. Covers the full spectrum from document-grounded to knowledge-base to persona-driven conversational AI.