Skip to Content
View all events

Agentic Analytics with Claude Code

Published by O'Reilly Media, Inc.

10x more powerful analytics with reliable, traceable, and reproducible agentic workflows

What you’ll learn and how you can apply it

  • Build reusable, traceable analytics workflows with Claude Code
  • Connect to live data sources via Connectors or Model Context Protocol (MCP)
  • Generate Jupyter notebooks that document every step of your analysis
  • Build a working forecasting Skill bundle you can deploy immediately for a variety of business problems

Course description

In this hands-on course with Tobias Zwingmann, you’ll move beyond simple prompts and learn how to build reusable, traceable AI-powered analytics workflows with Claude Code. You’ll work through a complete data analytics lifecycle, from problem scoping to deployment, using the Claude desktop app, which provides Claude Code in a clean, install-and-run interface (no terminal required).

The course is structured around a real-world running example: building an agile project forecasting tool using real-world data. You’ll connect Claude Code to GitHub via the Model Context Protocol (MCP), use Skills (reusable code templates) to handle data preparation and Monte Carlo simulation, and have Claude Code generate Jupyter notebooks that document every step of the analysis. The notebooks are modular, traceable, and reproducible so you can rerun them, modify them, or share them with your team.

By the end of the course, you’ll have a working “agile project forecaster” Skill bundle you can point at your own GitHub repository or any other data source. You’ll also have a clear mental model for how Claude Code differs from pure chat-based AI tools like ChatGPT, and when to use each in your daily analytics work.

This live event is for you because...

  • You’re a data analyst, BI professional, or analytics engineer who already uses Claude or ChatGPT and wants to move from one-off prompts to reusable, agentic workflows.
  • You’re a data scientist looking to integrate agentic AI into your daily analysis without leaving familiar tools (Python, Jupyter, GitHub).
  • You’re a scrum master, agile coach, or engineering manager who needs better forecasting tools for project planning.
  • You’re an analytics team lead evaluating Claude Code as a tool for your team.
  • You’re a Python-comfortable professional who wants to build production-ready AI analytics workflows.

Prerequisites

  • An active Claude account (Pro subscription recommended) or API key
  • Claude Code installed (via Claude Desktop app or command terminal)
  • Git installed on your computer
  • A free GitHub account
  • The ability to read and modify Python code in a Jupyter notebook
  • Hands-on data analytics experience
  • No terminal expertise required; the Claude desktop app handles the environment

Recommended follow-up:

Schedule

The time frames are only estimates and may vary according to how the class is progressing.

Welcome and course introduction (20 minutes)

  • Presentation: Course orientation, agenda, and outcomes; analytical problem-framing for Claude Code and how it differs from Claude Chat
  • Demonstration: Forecasting agile projects using case study data
  • Hands-on exercise: Have Claude generate a “hello world” Jupyter notebook with modular cells

Data preparation with Skills (40 minutes)

  • Presentation: What is a Skill?; how Skills differ from plug-ins and prompt templates; common data prep patterns (deduplication, missing values, type coercion, time-based filtering)
  • Hands-on exercises: Connect to GitHub via Connector and pull exercise data from the sample repository; have Claude Code generate a data preparation notebook with modular, traceable cells; save the data preparation logic as a reusable Skill
  • Break
  • Q&A

Data analysis and modeling (75 minutes)

  • Presentation: Throughput versus velocity—why story points alone fail as time estimates; Monte Carlo simulation for project forecasting—concept and intuition
  • Hands-on exercises: Use Claude Code to perform exploratory data analysis on the issues data; apply a prebuilt Monte Carlo Skill to forecast project completion; compare 50th and 85th percentile predictions; modify the forecasting Skill to handle a different data shape (issue counts versus story points)
  • Q&A
  • Break

Insights and deployment (30 minutes)

  • Presentation: Traceable agentic analysis—notebooks as deliverables and Skills as artifacts; bundling and sharing Skills with your team; governance and reproducibility—when to trust agent output and when to verify
  • Hands-on exercise: Generate a final analysis notebook or presentation with narrative commentary ready to share with stakeholders
  • Break

Wrap-up and Q&A (15 minutes)

Your Instructor

  • Tobias Zwingmann

    Tobias Zwingmann is an experienced senior data scientist and managing partner at RAPYD.AI. His mission is to help companies adopt machine learning and AI solutions faster while creating meaningful business impact. Before founding RAPYD.AI, he worked for more than 15 years in both hands-on and strategic roles in a corporate setting to build out data science use cases. He is author of the book AI-Powered Business Intelligence and coauthor of Augmented Analytics, both for O’Reilly.

    linkedinXsearch

Skills covered

  • Machine Learning
  • Matplotlib
  • Scikit-learn