Webinar | Deep dive | December 9

Agentic Development in BI – Curating Context for Agents

This is the third session in our webinar series on agentic development in BI. It focuses on the second building block: context.

Sign up here
tabular-editor-portrait-ruben-and-eugene-1080x1080

Agents only know what you put in front of them. Too little context and the results are generic. Too much and performance drops, costs rise, and the agent gets distracted by information it doesn’t need.

This session is part of a webinar series based on our blog series on agentic development for BI. It builds on the introductory session and expands on the article A beginner’s guide to managing context for AI agents. Eugene Meidinger and Ruben Van de Voorde, Innovation Specialists, show how to curate context that makes agents more useful in BI work.

In this session, you’ll learn:

  • What counts as context, and why it affects both quality and cost
  • How memory files, skills, and MCP servers differ, and when each one loads
  • How to apply progressive disclosure to give agents the right context at the right time
  • How to scope context across organization, user, and project
  • Why context rot happens, and how to manage sessions with /clear and /compact
  • Why human-written context usually beats AI-generated context, and how to keep it current

More sessions will follow. Each one covers a building block of agentic development: model, context, prompt, tools, and environment.

Event Details

Date: Wednesday, December 9
Time: 15:00-16:00 (CET, UTC+1)
Host: Eugene Meidinger and Ruben Van de Voorde