DOSS designer Josh Armantrout pointed Claude Cowork at a full Verve Coffee install corpus, synthesized gaps into a data-model plan, then handed Claude Code a to-do list that finished a prototype pass in about two hours.
This agent first explores the connected data source and records schema context, then answers natural-language questions with structured results and optional visual output.
Google's tutorial builds a conversational analytics agent that discovers BigQuery metadata and generates SQL against the public thelook_ecommerce dataset, using a maintained toolset rather than custom database plumbing.