Why it worked
It provides a high-value, actionable 'power user' tip that solves a common frustration with AI tools, making the content highly shareable and saveable.
Summary
A user demonstrates a workflow for improving AI interactions by prompting Claude to summarize session feedback into a file. This feedback file is then reloaded into future sessions to maintain context and improve output quality.
Structure
- 1User shows themselves on a treadmill with a laptop
- 2On-screen text introduces a specific prompt for Claude
- 3The prompt is displayed in full on screen
- 4Explanation of why this workflow improves AI performance
Product placement
The video is a tutorial on how to use Claude more effectively, specifically by creating a feedback loop.
On-screen text
The prompt I run at the end of every Claude session
"read our whole conversation, extract every correction made, every session preference I stated, anything you'd do differently next time and format it for a feedback file you'll reload in future sessions."
paste the output into a feedback.md file, load it into your Claude project.
most people correct Claude in the moment and move on, that's information gone, your outputs stay generic because the lessons die with the chat.
Original caption
the prompt that turns Claude into my editor. at the end of every session I run this: "read our whole conversation, extract every correction made, every session preference I stated, anything you'd do differently next time and format it for a feedback file you'll reload in future sessions." paste the output into a feedback.md file, load it into your Claude project. most people correct Claude in the moment and move on, that's information gone, your outputs stay generic because the lessons die with the chat. save them because the compounding is wild. #ai #Tech #dev #developer #claudeai