This repository contains course materials for AI for Excel: Automate the Spreadsheet Work That Eats Your Week, a practical course for Excel users who want to use AI and modern Excel tools to reduce manual spreadsheet work.
The course is designed for finance, accounting, operations, and business professionals who live in Excel and want hands-on ways to clean data, write formulas, summarize reports, and build repeatable workflows with less copying, pasting, and guesswork.
No coding background is required.
Most Excel users are doing far more by hand than they need to. This course shows how to use AI and modern Excel features to work more efficiently on common spreadsheet tasks, including:
- Cleaning messy data
- Writing and troubleshooting formulas
- Summarizing reports
- Using Python in Excel for analysis
- Building repeatable workflows with Power Query
- Deciding when to use Copilot, Python in Excel, Power Query, or a general AI assistant
The focus is not on using AI as a gimmick. The goal is to build practical workflows that save time, reduce errors, and make spreadsheet work easier to maintain.
By the end of the course, you will have built an AI-assisted Excel workflow that you can reuse and adapt for your own work.
This course is intended for:
- Financial analysts
- FP&A professionals
- Accountants
- Operations managers
- Business analysts
- Excel power users
- Professionals who want practical AI use cases inside everyday spreadsheet work
You do not need prior AI experience or a programming background. You should be comfortable working in Excel with basic formulas and spreadsheets.
The course uses a mix of Microsoft Excel features and general-purpose AI tools, including:
- Microsoft Excel
- Power Query
- Copilot in Excel (including custom skills)
- Python in Excel
- Agent Builder in the Microsoft 365 Copilot app (declarative agents)
- ChatGPT, Claude, or another general AI assistant
A Microsoft 365 subscription is recommended for following along with Copilot in Excel and Python in Excel. Power Query is built into modern versions of Excel. Custom Copilot skills and Agent Builder need a paid Microsoft 365 Copilot licence and are rolling out progressively; Session 3 includes a fallback for anyone without them.
By the end of the course, you should be able to:
- Use AI assistance to plan and improve a recurring spreadsheet workflow
- Clean messy data using Power Query
- Use Copilot and Python in Excel to support formulas, summaries, and analysis
- Decide which tool is best suited for a given Excel task
- Recognize where AI-generated output needs review, correction, or supervision
- Build a reusable reporting or data-cleaning workflow in Excel
- Write a recurring review down as a Copilot in Excel skill (a plain-text SKILL.md)
- Configure a declarative agent in Microsoft 365 Copilot that answers questions about a workflow
The course is organized around three main sessions.
Topics include:
- What AI can and cannot do well in Excel
- Where Copilot, Python in Excel, Power Query, and general AI assistants fit
- Setting up a practical AI-assisted Excel workflow
- First data-cleaning exercise using Power Query and AI assistance
Practice work:
- Clean a provided messy dataset
- Document the steps used to make the workflow repeatable
Topics include:
- Using AI to write, explain, and troubleshoot Excel formulas
- Using Python in Excel for analysis
- Asking better questions of AI tools
- Comparing Copilot, Python in Excel, and general AI assistants for analysis tasks
Practice work:
- Reproduce an analysis from class
- Adapt the workflow to a dataset of your own or a provided practice file
Topics include:
- Turning one-off spreadsheet work into a repeatable process with a Power Query folder connection that refreshes itself
- Writing the monthly review down once as a custom Copilot in Excel skill (SKILL.md), and running it with
@skill-name - Wrapping the process in a declarative agent built with Agent Builder in the Microsoft 365 Copilot app: instructions, knowledge documents, starter prompts, testing, and sharing
- Common failure points (schema drift, unmatched lookups, date locales, agents that invent numbers) and how to review AI output
- Capstone project guidance
Practice work:
- Complete a reusable Excel workflow on new data (store sales)
- Fill in a skill template so it describes your own review, install it, and run it (or use it as a prompt if skills are not available on your build)
- Optionally wrap it in an agent with a short run-book as its knowledge
An optional, self-directed project: a working AI-assisted Excel workflow you
build on your own time to lock in the course. There's no submission, no
deadline, and no grading — it exists because building one is the best way to
make the skills stick, and you walk away with something useful. Everything is in
the top-level Capstone/ folder.
How it works:
- Build it on your own time, using your own data or a ready-made scenario.
- No data you can use? The
Capstone/folder has three Sample Scenarios with datasets — they sidestep the anonymizing question entirely. - Know it's solid by running it past the Definition of Done self-check.
- Share it in Slack only if you'd like — welcome and encouraged, never required.
Examples: a monthly reporting refresh, a repeatable data-cleaning pipeline, a summary report from raw data, or a formula troubleshooting and documentation workflow. The goal is to use AI and modern Excel to cut manual work while still reviewing and understanding the results.
Each session has two source-of-truth docs: a Facilitator Guide (run the
session from this) in Instructor Notes, and a Participant Guide (learner
companion: demo recap + their exercise) in the session folder. Each session folder
also has an Extra Practice doc: the second dataset as a standalone exercise with
its answer key, for spare time in class or as homework. Shared references
sit in Course Overview.
Instructor Notes/ (instructor docs)
Session 1 - Facilitator Guide.docx RUN THE SESSION FROM THIS: demo (you) →
Session 2 - Facilitator Guide.docx practice (them) → regroup, with click-by-
Session 3 - Facilitator Guide.docx click steps, targets, fillers, Copilot aside.
Sessions 2 and 3 open with a one-table RUN SHEET
(clock / do / say / check) you can teach from alone
Course Logistics & Prep.docx course-level logistics, prep, community
Read-Aloud Scripts.docx per-slide delivery scripts
Slack and Logistics.docx running the cohort + Slack playbook
Course Overview/ (shared references, all sessions)
Prompt Cheat Sheet.docx copy-paste prompts for every tool
Data Dictionary.docx what every dataset contains
Session 1 - The Modern Excel + AI Landscape/
Participant Guide.docx learner companion: demo recap + exercise
Session 1 - Solution (Power Query).docx worked answer (instructor key)
Session 1 - Power Query M Code.m paste-ready query
Session 1 - Extra Practice.docx standalone your-turn: steps + answer key
Session 1 - Power Query M Code (Extra Practice).m paste-ready query for the expense file
Data/ regional-sales-RAW.csv (demo) +
expense-reimbursements-RAW.csv (your turn)
Session 2 - Formulas, Analysis & Summaries/
Participant Guide.docx page 1 is a follow-along card (every formula, in
order); ends with links to George's posts
Session 2 - Solution (Formulas & Python).docx
Session 2 - Extra Practice.docx standalone your-turn: four rounds + answer key
Data/ budget-vs-actual.xlsx (demo) +
sales-quota-vs-actual.xlsx (your turn)
Session 3 - Build & Ship a Reusable Workflow/
Participant Guide.docx demo recap, cheat sheets (Power Query, skills,
Agent Builder), your turn, check-your-work
Session 3 - Solution (Reusable Workflow).docx
Session 3 - Power Query M Code.m paste-ready CombinedExpenses + CombinedSales queries
Session 3 - Extra Practice.docx standalone your-turn: three rounds + answer key
Data/ monthly-expenses/ + vendor-master (demo);
your-turn-store-sales/ + store-master (your turn)
Skills/ Copilot in Excel skills (one SKILL.md per folder)
monthly-spend-review/ the demo skill, complete
store-sales-review/ the your-turn template with blanks
store-sales-review-SOLUTION/ one finished version (instructor key)
README.txt how to install a skill
Agent/ Agent Builder materials
agent-recipe.md name, description, instructions, starter prompts to paste
Monthly Spend Run-Book.docx the knowledge document the agent reads
Capstone/ (the final project — launched in Session 3)
Capstone Brief.docx what it is, when, how to submit + rubric
Definition of Done.docx "you're done when…" checklist + graded rubric + worked example
Sample Scenarios.docx 3 ready-to-use scenarios (for learners without their own data)
Data/ datasets for each scenario (regional sales,
support tickets, marketing spend)
Slides are delivered separately in the GenAI Works Academy "AI for Excel Slides" file. All datasets are synthetic.
To get the most out of the materials:
- Start with the session materials in order.
- Complete the provided exercises before adapting them to your own data.
- Use the prompts and examples as starting points, not final answers.
- Review all AI-generated output carefully.
- Build one workflow that solves a real recurring spreadsheet problem for you.
George Mount is a Microsoft Excel MVP and analytics author. He is the author of Advancing into Analytics and Modern Data Analytics in Excel, both published by O'Reilly, and has taught Excel, analytics, and AI-powered productivity to hundreds of thousands of learners through online courses, corporate training, and live workshops.
He helps finance and operations professionals modernize how they work in Excel using Power Query, Python in Excel, Copilot, and other AI-assisted tools.
LinkedIn: https://www.linkedin.com/in/gjmount/
AI tools can be powerful assistants, but they are not replacements for judgment.
Throughout this course, you are encouraged to:
- Check formulas before relying on them
- Validate summaries against the source data
- Understand each workflow before reusing it
- Treat AI output as a draft, not as final truth
- Keep sensitive or confidential data out of public AI tools unless your organization has approved their use
The goal is not to make Excel a black box. The goal is to make your work faster, clearer, and easier to repeat.
Unless otherwise noted, the materials in this repository are provided for educational use.
Please do not redistribute course materials as your own training program or commercial product without permission.