32dots HEIDELBERG AI
Session 6 medium

Custom GPTs and deeper data analysis

LESSONLesson 6 · ~20 min

🎯Goal. Use a purpose-built custom GPT for a recurring task, and push data analysis further — multi-step analyses and downloadable results from your files.

▶ Try this prompt

You are my lab stats assistant. When I upload a CSV, always: report n per group, check normality, choose and justify the right test, run it, report effect size, and give a publication-ready one-sentence result. Confirm you understand, then wait for my file.

You can paste this into a normal chat, or bake it into a custom GPT (via 'Explore GPTs' → Create) so the behaviour persists and is reusable. Then upload a CSV in the next message.

Steps
  1. 1Discover and use existing GPTs. Browse 'Explore GPTs' for ready-made assistants. Every plan, including free, can discover and use GPTs.
  2. 2Create your own GPT. Building and sharing a custom GPT requires a paid plan (Plus and above) — wrap a recurring workflow (like the stats assistant above) into a reusable tool with its own instructions.
  3. 3Push data analysis. Ask for multi-step analyses across an uploaded file and request a downloadable cleaned CSV or saved figure. ChatGPT runs the Python and returns files you can save — always re-check the statistics against what you expected.

You'll see. A reusable assistant that runs your standard analysis the same way every time, and a worked multi-step analysis with a downloadable result file.

💳Cost. Using GPTs and basic data analysis: free tier (limited). Creating and sharing custom GPTs: Plus and above. Heavier/unlimited data analysis and file uploads: Pro.

💡Takeaway. Anyone can use custom GPTs; creating your own needs a paid plan — wrap a repeatable analysis into one GPT so it runs identically every time.

How was this lesson?