Catch the embarrassing mistake before your viewers do.

In this hands-on project you build a small Python tool that reads any text - a video script, a blog post, a caption - pulls out every claim, and flags the ones that look wrong, risky or impossible to check, with a reason and what to search. It uses Google Gemini’s free tier, or your own local AI with Ollama.

What you’ll build

• check_facts.py: point it at a text file (or paste text), get a clear report
• A saved report.txt for every check
• A settings file to switch between Gemini and Ollama

What you’ll learn

• Getting a free Gemini key and keeping it safe
• Asking an AI for structured answers (JSON) and handling it when it gets it wrong
• What AI checkers are good at - and, honestly, what they miss
• Fixing the usual problems (quota reached, no internet, file errors)

What you get

• A 13-lesson step-by-step course with real screenshots
• The finished tool, sample scripts with planted mistakes, an example report and a README

You need: Windows or Mac, an internet connection (or Ollama), a free Google account, and the free lessons What is the terminal?, What is Python? and What is an API? (links inside).

Time: about 25 minutes.