How to Switch From ChatGPT to Claude
ChatGPT has been learning from you. Every project you’ve described, every preference you’ve shared, every goal you’ve mentioned – it’s all stored in your memory.
You don’t have to start over when you switch to Claude. Use the system prompt below to extract everything ChatGPT knows about you, then paste the output into your Claude custom instructions.
- Open ChatGPT – log in and start a new conversation.
- Copy and paste the prompt below – send it exactly as written.
Your Export Prompt
Copy everything inside the box below and paste it into ChatGPT:
Extract and format everything you know about me into a structured system prompt that I can use with other AI models.
Include:
Work Context - my role, projects, team, responsibilities, tools I use
Personal Context - interests, background, relationships, location, lifestyle
Communication Preferences - tone, structure, language level, what to avoid
Thinking Style - how I approach problems, frameworks I like, decision-making patterns
Top Priorities - what I'm focused on right now
Domain Knowledge - areas where I'm expert vs beginner
Interaction History - recurring themes, common requests, patterns in our conversations
Output Preferences - formatting, length, examples, level of detail
Values & Principles - what matters to me in work and life
Constraints & Boundaries - things I don't want, topics to avoid, limitations
Format the output as:
System Prompt: How to interact with [my name]
Core Identity [Who I am, what I do, key context]
Communication Style [How to talk to me]
Work & Projects [Current focus, responsibilities, tools]
Preferences [Structure, tone, language, format]
Knowledge Level [What I know deeply vs what I need explaining]
Common Requests [What I typically ask for help with]
Values & Priorities [What drives my decisions]
Avoid [What NOT to do]
Be comprehensive. Include specific examples, names, frameworks, and patterns. The goal is to make any AI model feel like it already knows me.
Works with any AI. Although this was specific to Claude, this output could be used in any LLM: Claude, Gemini, Grok, Perplexity, or any other AI that accepts a system context.