Designing context for agentic analytics

"You're an expert data analyst, a SQL master, and an in-depth expert on our business domain....now show me our sales in a pie chart".

We've all seen those LinkedIn posts. The 4 prompts you HAVE to try to revolutionize your data analysis.

But where does LLM prompt-engineering-hype end, and genuine impact begin? The Count Agent is probably already better than your average "expert data analyst". Give it the context it needs to understand your organization and its output becomes ever, and consistently more powerful.

In this workshop we'll look at how context at workspace, catalog, project, and canvas level influences the agent and can be used to shape it's output

We'll cover:

1. The type of context that is useful to agents — and how to avoid common mistakes.
2. A strategy for iteratively developing context — building up and building down from "important to this query" to "important for every query"
3. How governance and context overlap — bridging the line between semantic models, and organizational knowledge.

Who should tune in?

• Everyone who has tried asking AI to solve problems and then battled lack of relevance and accuracy.
• Data teams looking at rolling out agentic analytics while building good governance through centralised context.