Build a Databricks AI Agent Without Custom Tables or Vector Search

A developer created a conversational Databricks agent using only the default samples.tpch dataset, bypassing Unity Catalog and Vector Search. The stack includes OpenAI Agents SDK, Databricks Model Serving with Llama 3.3 70B, SQL Connector, MLflow, and Gradio UI. Key pitfalls include using AsyncOpenAI, applying nest_asyncio, and avoiding Spark context in Apps.