Record ChatGPT sessions without exposing conversation history. Guide for prompt engineering tutorials, AI workflow demos, and Custom GPT recordings.
Record yourself crafting, iterating, and refining prompts. Show the full conversation flow — initial prompt, unsatisfying response, refined prompt, better response. This iterative process only works on video.
Record the same prompt in ChatGPT and compare responses. Side-by-side recordings show differences in style, accuracy, and capabilities that text articles can't convey.
Record how you use ChatGPT in your work — writing emails, generating code, analyzing data, brainstorming. Show the real prompts you use, not hypothetical examples.
Record Custom GPTs or ChatGPT plugins in action. Show the configuration, test it with real prompts, demonstrate the output. Plugin creators need these for documentation.
In Screenify Studio, select Window Capture and pick the ChatGPT window.
1920x1080 is fine for ChatGPT — the text column is narrow (~700px) and doesn't benefit from higher resolution.
Open ChatGPT in Chrome. Before recording, collapse the sidebar by clicking the sidebar toggle (top left). Your conversation history titles are visible and may contain sensitive or embarrassing prompts.
Start a NEW conversation for the recording. Don't record in the middle of an existing thread — viewers need context from the beginning. Click 'New Chat' in the sidebar.
If you're using Chrome, hide the bookmarks bar (Cmd+Shift+B) and close other tabs. ChatGPT runs in the browser — same tab privacy rules apply.
Start recording. Type your prompt OR paste a pre-written prompt. If typing live, that's fine for authenticity, but long prompts waste recording time. For complex prompts, paste them and explain the key parts before hitting Enter.
For prompt engineering tutorials, use a split approach: type a simple prompt first, get the response, then explain what's wrong with it. Then type the improved prompt, get the better response. This before/after structure teaches the lesson naturally.
When ChatGPT generates code, click the 'Copy code' button and switch to your editor to test it. Recording this full loop (prompt → code → test → feedback) shows a real workflow, not a cherry-picked result.
For Custom GPT demos, start by showing the GPT configuration screen (if you built it), then switch to the chat interface and demonstrate it. This gives viewers both the builder perspective and the user perspective.
Collapse the sidebar immediately. Your ChatGPT sidebar is a history of everything you've asked an AI — 'How to deal with a difficult coworker,' 'Write a dating profile,' 'Explain this medical test result.' One click on the sidebar toggle hides it all. Do this before recording, every time.
Use silence detection for streaming waits. ChatGPT streams responses at ~30 words/second. A detailed response takes 15-30 seconds to fully render. You're sitting there in silence waiting. Silence detection marks these gaps so you can trim them in the timeline editor — turning a 10-minute recording into a tight 6-minute video.
Auto-zoom on code blocks. ChatGPT renders code in monospace font at ~13px in a shaded block. After video compression, variables and function names blur together. When you click on a code block to explain it, auto-zoom zooms in so viewers can read the actual code.
Paste complex prompts, don't type live. Watching someone type a 200-word prompt character by character is boring. Pre-write your prompt, paste it, then explain the key parts ('notice I specified the output format here, and the persona here'). Saves recording time and keeps viewers engaged.
Leaving the conversation sidebar open. It shows every conversation title. 'Help me apologize to my ex,' 'Am I being underpaid,' 'Write a complaint about my landlord.' Collapse it before recording. Every single time.
Recording the full streaming time without editing. You press Enter and wait 20 seconds for the response to stream in. Viewers watch text appear word by word. Exciting for the first 3 seconds, boring for the next 17. Record it, then trim the wait in the editor. Or speed up the streaming section to 3x.
Not starting a fresh conversation. You record in the middle of a 15-message thread. Viewers have no context — they don't know what prompt #1 was, what the AI's context window contains, or why the response references something they can't see. Always start a new chat.
Showing your OpenAI account settings or billing. If you navigate to Settings during a demo, your email, payment method, and usage stats are visible. Record only the chat interface, not the account pages.
AI auto-zoom on clicks, instant captions in 50+ languages, cursor highlighting, and studio-grade export. Free plan, no watermark.
Try Screenify Free