Wed, Aug 5, 2026
4:00 PM UTC (45 minutes)
Virtual (Zoom)
Free to join
Go deeper with a course
Wed, Aug 5, 2026
4:00 PM UTC (45 minutes)
Virtual (Zoom)
Free to join
Go deeper with a course
What you'll learn
Choose the right first question
Start with a recurring analytics question tied to a real decision, not an agent that tries to answer everything
Give the agent the right business context
Define the metrics, sources, grain, exclusions, and examples the agent needs before it writes SQL
Route and constrain the work
Decide when the agent should query data, retrieve context, clarify the question, or refuse to answer
Prove it works with real questions
Test real questions, inspect failures, collect feedback, and set the proof needed before expanding the pilot
Why this topic matters
Most agentic analytics demos jump from a question to SQL and stop there. Production systems fail for different reasons: missing business context, poor source selection, weak guardrails, and no reliable way to evaluate answers. This lesson gives data professionals a practical sequence for turning one recurring business question into a pilot that real users can test.
You'll learn from
Paras Doshi
Head of Data (Amazon, Opendoor)
Previously at
