Workshop: From prompts to dependable workflows: building AI Skills for measurement work

Gunnar Griese

10:20 - 11:10

AI agents can already connect to your tools and act on them. But “capable” isn’t the same as “reliable,” and the gap is where most people are stuck now. In this hands-on workshop you’ll close that gap. You’ll connect an AI agent to a live GA4 property over an MCP server, then watch the same task get progressively more dependable as you add a Skill that teaches the agent how to do the job: What to check, what scope to default to, what never to touch.
We use GA4 because it’s familiar, but nothing here is GA-specific. The MCP-plus-Skills pattern is exactly how you’d make an agent reliable against GTM, BigQuery, your CDP, an ad platform, or any internal tool with an API. You’ll leave with a method you can point at anything.

Gunnar Griese

Measurement Specialist8-bit-sheep

With over nine years of experience in digital analytics, data science, and e-commerce, I am a data-driven leader and educator who is passionate about helping businesses leverage data to optimize their performance and growth.

As Measurement Consultant & Analytics Engineer, I offer my expertise to businesses through 8-bit-sheep. I work to design, develop, and deploy data solutions for clients across various industries, using Google Cloud Platform, Google Analytics, and Google Tag Manager. As a Lecturer at IHM Business School, I teach the courses “Digital Analytics – Systems & Technology” and “Tag Management”, where I share my knowledge in technical digital analytics with aspiring professionals.

My goal is to continue to learn, innovate, and collaborate with diverse and talented people who share my vision of creating value and impact through data.

Last but not least, I do share my thoughts on all things digital analytics frequently on my blog: gunnargriese.com