Case Study
chatdeb
Full-stack chatbot for analyzing business data.
aidatafullstack
Context
Business users need conversational analytics that still respects modeled data contracts.
Approach
Built a full-stack assistant with FastAPI + OpenAI/LangChain on top of DuckDB/dbt models and a React/Next interface.
Architecture
UI query layer -> API orchestration -> retrieval + LLM response synthesis -> dbt-modeled warehouse tables.
Results
Enabled faster exploratory analysis while maintaining traceability to governed data models.
Lessons
Conversation UX improves only when response grounding and metric definitions are explicit.
Artifacts
If I had 2 more weeks...
Ship richer source citations and query execution observability.