TL;DR - Intermediate data work means connecting real data to AI for repeatable insight, anomaly detection, and dashboards - and always verifying how a number was reached.
Why it matters
Data fluency is one of the most in-demand AI-adjacent skills in every industry. Going past copy-paste to connected, repeatable analysis is a genuine career lever.
Worked example
[Upload a multi-tab sales workbook]
- Analyze across tabs: top trends, any anomalies.
- Flag the region with an unusual dip and a likely cause to investigate.
- Output a 1-page findings brief with 2 recommended actions.
- Show how you computed the key numbers.Steal this - analysis workflow
1. Describe the data (tabs, columns) up front.
2. Ask for trends + anomalies, not just a summary.
3. Demand a recommendation tied to the data.
4. "Show your work" on key numbers, then spot-check one.Common mistakes (and the fix)
- Trusting the conclusion blindly. Fix: you own it - verify the setup and one number.
- Vague asks. Fix: specify the calculation and the output format.
- Sensitive data in public tools. Fix: anonymize or use a private/enterprise option.
Good to know
ChatGPT's data analysis runs real Python on your files (trustworthy math, not guessing), and Gemini in Sheets/BigQuery brings AI to where your data already lives. For recurring dashboards, pair analysis with an automation (Level 4) that refreshes it on a schedule.