Kyle Springfield

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.

Artifacts