Agentic Analytics: Building Multi-Step AI Workflows

Agentic Analytics: Building Multi-Step AI Workflows

45mIntermediate2026-09-09

Authors

Mo Chen

Mo Chen

Course details

Agentic AI is rapidly moving from experimentation to practice. As teams move beyond individual prompting toward automated, multi-step workflows, professionals need to understand not just how to use AI, but how to design, orchestrate, and govern AI-powered systems.

In this course, Mo Chen walks you through a hands-on, five-day process for building reusable agentic analytics workflows in Claude Cowork. Build three skills (a problem framer, a data analyst, and a stakeholder communicator), then package them into a single plugin that runs end to end. On the final day, point your plugin at a brand-new business problem, dataset, and data dictionary, and watch it work on data it has never seen.

Learning objectives
Design a multi-step analytical AI workflow before building it.
Write, test, and iterate reusable analytical skills in Claude Cowork.
Chain skills into a pipeline that handles real-world edge cases and errors.
Build a full end-to-end analysis pipeline running from business question to findings summary.
Package a complete agentic analytics workflow as a deployable .plugin file.

Concepts

Introduction

  • Stop rebuilding the same analysis from scratch every time
  • Claude Chat vs. Cowork vs. Code prompt vs. skill vs. plugin
  • Get Claude Cowork ready before day 1

Build the Problem Framer Skill

  • Build your first skill and watch it frame the problem

Build the Data Analyst Skill

  • Build a data analyst skill that works on any dataset

Build the Stakeholder Communicator Skill

  • Build the skill that writes for the person receiving it

Create Your Custom Agentic Analytics Plugin

  • Stack three skills into a plugin that runs in your order

Run Your Agentic Analytics Plugin on a Brand-New Problem, Dataset, and Data Dictionary

  • A different industry, a different question, the same plugin
  • Point your plugin at data it has never seen
40,000 Toman