Conversational Analytics with Looker
From semantic layer to trustworthy data agent.
1. Why conversational analytics changes BI
Natural-language interfaces, business questions, and why removing SQL increases the need for governed meaning.
2. LookML as the semantic contract
Measures, dimensions, descriptions, synonyms, joins, grain, and business definitions.
3. Designing Explores for CA
Field quality, model hygiene, scope, discoverability, and question patterns.
4. Building data agents
Agent instructions, business context, multiple Explores, and verified query patterns.
5. Prompt vs semantic modeling
What belongs in LookML, what belongs in instructions, and what should never depend on prompts alone.
6. Evaluation and trust
Golden questions, expected answers, query inspection, failure taxonomy, and accuracy tracking.
7. Conversational Analytics APIs
Agents, conversations, messages, chat APIs, streaming, persistence, and custom application integration.
8. Enterprise production architecture
Auth, allowlists, audit logs, observability, governance, security boundaries, and deployment patterns.