Claude for Data Analysis (AI-CLA200)

Data analysts spend too much of their time on things that shouldn't require a developer: building one-off dashboards, formatting reports, wrangling CSV exports, and waiting for BI tickets to get prioritized. This workshop teaches analysts to close that gap themselves — using Claude to go from raw data to a working, shareable output in a single workflow.

No new tools to install. No infrastructure to set up. Just Claude — and the same plain-English skills you already have.

Who's it for?

• Data analysts and business analysts
• Finance and operations teams who own data but not dev resources
• AP managers, reporting leads, and anyone who lives in Excel
• Anyone who's said "I just need someone to build me a dashboard" more than once

1. Claude as a Data Co-Pilot

   How Claude thinks about data differently than a search engine or a chatbot. The mental model shift: from asking questions to delegating analysis. Live demos of natural-language data exploration — uploading a dataset, asking questions, getting summaries, spotting anomalies — no SQL required.

2. Design Without a Designer

   Introduction to Claude Design. Participants describe a dashboard in plain English and watch it render as an interactive prototype in real time. Iteration loop: refining the layout, adjusting components, reviewing with stakeholders. How to use Claude Design as your design review meeting — no Figma, no UX team.

3. Build Against Live Data (60 min)

   The AP Dashboard lab: participants take their Claude Design wireframe and hand it off to Claude Code connected to a live AP MCP. Claude queries the data, identifies exceptions, and builds a self-contained HTML dashboard — skeleton loaders, color-coded exception alerts, vendor master, KPI cards — in one build pass. Participants finish with a working file they created themselves.

4. Prompting Patterns for Analysis

   Structured techniques for getting useful analytical output from Claude: exception detection prompts, narrative summary prompts, comparison and trend prompts, and "what should I be worried about?" prompts. Pattern library distributed as a take-home reference card.

5. Connecting Your Own Data

   Overview of MCPs: what they are, how they give Claude read access to live systems (ERP, CRM, data warehouse, spreadsheets). The two paths: use an existing MCP connector, or have a developer build one. Q&A on participants' specific data environments.

Each student receives a comprehensive set of materials, including course notes and all class examples.