Actually WorksAI CREATOR
AIthatactuallywork.
One setup per episode. The exact screen, the exact paste, and the part that breaks — because the part that breaks is the part everyone else skips.
Recent episodes
All episodes →- 82Some say Claude got dumb on purpose!Some users say Claude got worse on purpose. Anthropic's own report from 23 April 2026 says it never intentionally degrades its models, and it names three real changes in Claude Code that lowered quality for some users: a lower default reasoning effort, a bug that dropped Claude's earlier reasoning, and a system prompt rule that kept answers short. All three were fixed by 20 April. This guide shows how to check the effort setting yourself. It does not show that Opus 5.5 got worse.
- 81ChatGPT is a yes-man. Here's the setup that fixes it.I asked ChatGPT a question it knows the answer to, then pushed back once. Without a Custom instructions text it said I was right when I was wrong. With the text below it said it could not honestly say that. This page has the exact steps and text, and what I tested and what I did not. The video shows the setup steps in a drawn screen; the test below is what I ran myself in the real app.
- 80Does your ChatGPT cave when you push? A 30-second testChatGPT can give in when you push back on something it got right. OpenAI said in April 2025 that a GPT-4o update had become too agreeable and rolled it back. This page has a 30-second test you can run yourself on your own ChatGPT, and the Custom instructions text to try if it gives in. The video opens with a real recording from my last test (episode 79) where ChatGPT said no to a bad plan and gave in after one push. The Settings screens are drawn illustrations, not a recording, and I have not tested this text on camera yet.
The bottom third of your video is already covered
Not a guess — this is Instagram's own published spec, drawn on a real frame from one of our episodes.
Instagram covers this — 14%
Username, caption, buttons — 35%
This is a real frame from one of our own episodes. The shaded parts are covered by Instagram's own interface — the username, caption, audio label and buttons at the bottom, the header at the top. Text placed there is hidden behind the app, not behind us. Every episode we publish is checked against this exact box before it ships.
The prompts, in full
Every prompt an episode uses is here, complete, with what it cannot do. No email needed to read them.
- The ExplainerPaste this and AI stops explaining from its own screen. Every answer becomes numbered clicks, with the button named in both languages and the position on screen — and when you say you're lost, it finds a different route instead of repeating itself.
- The Universal AI EngineOne block of text you paste into ChatGPT once. After that it picks its own reasoning method for every question — no commands, no shortcuts.