Vigyata.AI
Is this your channel?

Claude Code: Data Analysis Without the Spreadsheet Headaches

3.0K views· 76 likes· 18:44· Mar 30, 2026

🛍️ Products Mentioned (2)

How to use Claude Code for data analysis. Two spreadsheets, one sentence, and an AI agent turned raw data into a full business analysis in about 10 minutes. In this video, I'll show you step by step how to build a reusable agentic business analyst with Claude Code that finds trends, flags problems, and gives you specific recommendations. This is part three of my Claude Code series. In part one, we installed Claude Code and built an app from scratch. In part two, we learned how to build AI agents using agentic workflows. Today, we're applying those skills to a real business problem. We'll use a fake coffee company called Basecamp Brew Co. with 12 months of sales data and 200 customer records, so you can follow along with the exact same files. You'll see how to set up a CLAUDE.md file, build a reusable data analysis workflow, run it on sales and customer data, and iterate on the reports -- all without writing a single formula or pivot table. Timestamps: 0:00 Intro 0:55 Meet Basecamp Brew Co. 1:35 The data we're working with 2:02 Setting up the project in VS Code 3:10 The CLAUDE.md file explained 5:02 Adding the data files 5:40 Plan mode and writing the workflow prompt 6:38 Watching the agent build the workflow 7:53 Running the sales data analysis 9:12 Reviewing the sales report 11:14 Analyzing the customer data 12:43 Reviewing the customer report 15:04 Iterating on the report 16:29 Why reusable workflows matter 17:38 Wrapping up and your challenge Basecamp Brew Co. data files (sales + customer + CLAUDE.md): https://go.teachers.tech/Claude_Code_BA Claude Code Part 1: https://youtu.be/7uyJ9H4hfkU Claude Code Part 2 (Agentic Workflows): https://youtu.be/gHB4JFG9i3k First Prompt: I want to build a reusable data analysis workflow. When I point you at a data file, you should ask me what I’m looking for, check the data for any issues, analyze it thoroughly, find trends and patterns, flag anything unusual, and produce a clear report with specific recommendations. The report should include actual numbers, formatted tables, and a section at the end with the top three things the business should do next. Second Prompt: Analyze the sales data in resources/sales-data.xlsx. This is 12 months of sales for Basecamp Brew Co. I want to understand what’s working, what’s not, and where we should focus next quarter. First Answer to Questions: Give me all three compared but lead with profit since that's what actually matters to a small business. Equal attention across all channels but I have a feeling the mall kiosk might be underperforming so keep an eye on that. Nothing planned for next quarter yet, that's what this analysis is for. And go broad, hit all six lenses. I want the full picture. If something jumps out that deserves a deeper look, go for it. Second Answer to Questions: Lead with spend and frequency together, that makes sense. Yes, definitely do the deep dive on repeat versus one-time customers. I want to know what makes someone come back. For acquisition channels, I want to know which channels produce higher-value customers over time, not just raw numbers. And for the cross-reference, work with what's joinable and flag the limitation. Subscribe: https://www.youtube.com/@TeachersTech?sub_confirmation=1 Website: https://www.teachers.tech Instagram: https://www.instagram.com/teacherstechlab LinkedIn: https://www.linkedin.com/in/teacherstech

About This Video

Two spreadsheets, one sentence, and an AI agent turned it into a full business analysis in about 10 minutes. In this video, I show you exactly how I did that using Claude Code—step by step—so you can build the same reusable “business analyst” workflow with me. We use a fake coffee company (Basecamp Brew Co.) with 12 months of sales data and 200 customer records, and I walk you through setting up a simple project in VS Code, dropping your files into a resources folder, and using a CLAUDE.md file to lock in business context and rules. The big idea is: plan first, build second. I switch into plan mode, write a workflow prompt in plain English, and let the agent create a repeatable process that asks clarifying questions, checks data quality, finds trends and anomalies, and outputs a clean report with real numbers and specific recommendations. Then I run it on sales data, run it again on customer data (with cross-referencing and honest limitations), and show how easy it is to iterate—like adding a channel comparison table—without starting over. Once you build this once, you can reuse it every month with new spreadsheets, no formulas, no pivot tables, and no spreadsheet headaches.

Frequently Asked Questions

🎬 More from Teacher's Tech