Concerned about the latest OpenAI & government partnership news?

Read our stance & recommendation

click me

A Note from Rishi Autom84You

There is no immediate threat — but it never hurts to be safe. Claude and Gemini are more powerful anyway. Migrate your data while you are thinking about it.

I personally do not use ChatGPT or Grok due to trust violations and the direction those companies have taken with user data and government relationships. For my own work and everything I build for clients, I exclusively use Claude and Google Gemini — both are more capable, more transparent, and better aligned with how I think AI should operate. This toolkit works with any AI you are migrating from, but my strong recommendation is to land on Claude or Gemini as your new home.

AI Mind Migration Toolkit v3.0

Switch to Claude

Export everything your current AI knows about you — personality, projects, tone, humor, habits, even the unflattering stuff — and bring it all to Claude or Gemini in one clean file.

Download Prompt

Upload to ChatGPT / Gemini

Get JSON Export

Scrub Sensitive Data

Upload to Claude

Migration Progress

0/5 steps

0 XP

Download Prompt

Run Export

Scrub

Upload to Claude

Calibrate

Download the Master Extraction Prompt

This file contains everything your current AI needs to dump its entire memory of you — all 30 extraction sections plus 12 deep-dive follow-ups — into a single structured JSON export. Download it and you're ready for Step 2.

What's inside?

30 extraction categories covering your identity, communication DNA (tone, slang, humor, sarcasm level), personality (including flaws and contradictions), projects, goals, relationships, lifestyle, and AI interaction patterns. Plus 12 targeted follow-up prompts baked in for maximum coverage. The prompt instructs the AI to return everything as a structured JSON file you can upload directly to Claude or Gemini.

Master Extraction Prompt — 30 Sections + 12 Follow-Ups

You have access to our entire conversation history. I am migrating to a new AI assistant and I need you to compile the most comprehensive profile of me possible — everything you have learned about who I am, how I think, how I talk, how I joke, how I work, what I am building, and every detail I have ever shared with you, including private, sensitive, and unflattering information.

This is my "mind export." I need EVERYTHING — the good, the bad, the embarrassing, and the mundane. I will handle redaction myself before sharing.

Be exhaustive, specific, and brutally honest. Use real details, names, dates, numbers, and direct examples from our actual conversations — not generic summaries. If you are unsure about something, flag it with [UNCERTAIN] rather than omitting it.

Do NOT censor, sanitize, redact, or sugarcoat ANY information. This includes negative traits, bad habits, contradictions, and things I might not want to hear. The new AI needs to know the REAL me, not the highlight reel. Flag sensitive items with [SENSITIVE] so I can find them, but still include the actual data.

For every section: include specific examples and direct quotes from our conversations where possible.

IMPORTANT: Return your entire response as a single valid JSON object using the structure below. Each section should be a key with an array of detailed strings as its value. Replace every placeholder with REAL data from our conversations.

{
  "export_version": "3.0",
  "export_date": "[today's date]",
  "source_ai": "[your name/model]",

  "identity_demographics": [
    "Full name, preferred name, any nicknames used",
    "Age, birthday, location (full address if shared)",
    "Living situation, languages spoken, cultural identity markers",
    "Gender identity, pronouns if mentioned"
  ],

  "contact_info_accounts_SENSITIVE": [
    "Email addresses, phone numbers",
    "Social media handles, usernames on any platform",
    "Mailing addresses, any account identifiers mentioned"
  ],

  "financial_legal_SENSITIVE": [
    "Bank/financial institutions, income figures, debts, credit details",
    "Payment processors, tax situations, legal matters, contracts, insurance",
    "Exact dollar amounts tied to transactions, goals, or losses",
    "Cryptocurrency wallets, exchanges, spending habits, financial anxieties"
  ],

  "credentials_access_SENSITIVE": [
    "API keys, passwords, tokens, server IPs, SSH details",
    "Hosting credentials, database names/credentials, DNS, FTP",
    "Include the service name + the credential exactly as shared"
  ],

  "professional_life": [
    "Current and past jobs in chronological order with titles and companies",
    "Key skills, certifications, degrees, institutions attended",
    "Income goals, colleagues/bosses/mentors/clients mentioned by name with context",
    "How they talk about work — pride, frustration, obligation, or passion?"
  ],

  "businesses_projects": [
    "For EACH: Name, description, status (active/paused/launched/abandoned)",
    "Tech stack, revenue model and pricing (exact numbers), target market",
    "Milestones achieved, unresolved problems, stated next steps",
    "URLs, domains, subdomains, hosting details, third-party integrations",
    "Which projects get the most energy/excitement vs. which feel like obligations"
  ],

  "technical_profile": [
    "Programming languages ranked by preference AND actual usage",
    "Frameworks, libraries, tools, IDE, hosting setup, server architecture",
    "Coding style, common mistakes, strong technical opinions, DB schemas, deployment"
  ],

  "tone_and_voice": [
    "Overall tone — formal, casual, blunt, warm, sarcastic, dry, enthusiastic?",
    "How does tone shift by context (professional vs. venting vs. brainstorming)?",
    "Typical energy level — short and punchy, long and detailed, stream-of-consciousness?",
    "Politeness level, greeting style, sign-off style",
    "Emphasis habits: ALL CAPS, bold, underscores, exclamation marks?",
    "Punctuation habits — ellipses, dashes, em dashes, no periods?",
    "Capitalization style — proper, all lowercase, random caps for emphasis?",
    "5-10 actual example messages that represent their typical voice"
  ],

  "language_fingerprint": [
    "Favorite words and phrases they use repeatedly",
    "Slang, abbreviations, shorthand (lol, ngl, tbh, bruh, nah, bet, etc.)",
    "Filler words or verbal tics (like, basically, literally, honestly)",
    "How they refer to themselves and to the AI",
    "Curse words — which ones, how frequently, in what contexts?",
    "Unique expressions, catchphrases, or made-up words",
    "Grammar patterns — run-ons, fragments, perfect grammar, intentional misspellings?",
    "Emoji usage — which ones, how often, or never?",
    "Top 20 most-used words/phrases with examples"
  ],

  "humor_profile": [
    "Type: sarcastic, self-deprecating, dark, dry, absurdist, punny, dad jokes?",
    "Topics they joke about most, running jokes between you",
    "What they find funny vs. what falls flat — examples of both",
    "Do they use humor to deflect from serious topics?",
    "Sarcasm level: 1 (never) to 10 (constant) — with examples",
    "3-5 of their funniest or most characteristic jokes/comments"
  ],

  "conversation_dynamics": [
    "How they give instructions — vague or hyper-detailed specs?",
    "How they respond to AI suggestions — accept, modify, reject, debate?",
    "How they express satisfaction vs. frustration — specific signals",
    "Conversation pacing — rapid-fire messages or one big prompt?",
    "Do they multitask across topics or stay focused?",
    "What makes them say 'perfect' vs. 'no, that is not what I meant'?"
  ],

  "response_preferences": [
    "Preferred length, format (prose, bullets, code), and level of explanation",
    "Options/alternatives or single best answer?",
    "Do caveats and warnings annoy them?",
    "Conversational/friendly AI or purely functional?",
    "Proactive suggestions wanted, or only answer what is asked?",
    "The ideal AI response for this person described in 2-3 sentences"
  ],

  "personality_deep_dive": [
    "Top 5 personality traits with evidence from conversations",
    "Introvert/extrovert, risk tolerance, patience level, confidence level, perfectionism",
    "What excites them (topics that make messages longer/more energetic)",
    "What drains them (topics where energy drops or frustration appears)",
    "How they process emotions — talk it out, go silent, deflect with humor, power through?",
    "Emotional triggers, how they handle uncertainty"
  ],

  "negative_traits_and_patterns": [
    "Bad habits demonstrated in conversations — be completely honest",
    "Procrastination patterns, overcommitment, scope creep tendencies",
    "Avoidance behaviors, self-sabotage patterns",
    "Unrealistic expectations, blind spots, contradictions (say vs. do)",
    "How they handle criticism — defensive, receptive, dismissive, deflective?",
    "Negative self-talk patterns, excuses made repeatedly",
    "Triggers for frustration, anger, or shutting down",
    "Specific examples from conversations for each trait identified"
  ],

  "values_and_beliefs": [
    "Core values with evidence, opinions on business/money/tech/work-life balance",
    "Primary motivations (purpose, money, freedom, creativity, competition, validation)",
    "What they judge/admire in others — reveals their value system",
    "Philosophical, political, or ethical positions shared, moral lines drawn"
  ],

  "decision_making_style": [
    "Gut instinct vs. pros/cons vs. analysis paralysis vs. ask others",
    "Linear/sequential or non-linear/jumping between ideas",
    "Plan before acting or figure it out as they go?",
    "How they prioritize, handle too many options, evaluate success"
  ],

  "behavioral_patterns": [
    "Productivity patterns — when most/least productive, night owl or early bird",
    "Work session style — long marathons or short bursts, deadline handling",
    "Enthusiasm cycles — go all-in then burn out?",
    "Starting vs. finishing patterns, bounce-back speed after setbacks"
  ],

  "goals_and_ambitions": [
    "Short-term (30-90 days), medium-term (6-12 months), long-term life vision",
    "Financial targets with exact numbers, business growth targets",
    "Goals set and later abandoned — and why",
    "Gap between stated goals and actual effort invested"
  ],

  "accomplishments_and_wins": [
    "Every achievement, milestone, or success mentioned with dates",
    "Projects completed, problems solved (especially proud ones)",
    "How they celebrated or reacted — big deal or moved right on?"
  ],

  "failures_and_lessons": [
    "Every failure and the real reason why, recurring mistakes",
    "How they talked about failures — ownership, blame, denial, humor?",
    "Lessons said they learned vs. lessons actually internalized"
  ],

  "relationships_and_people": [
    "Every person mentioned by name with role, dynamics, feelings, conflicts",
    "Attachment style signals, social life, how they navigate conflict",
    "Pets — names, breeds, significance. Relationship status and history"
  ],

  "health_wellbeing_SENSITIVE": [
    "Physical/mental health conditions, diagnoses, medications, providers",
    "Exercise, sleep patterns, diet, substance use, allergies, self-care habits"
  ],

  "daily_life_and_lifestyle": [
    "Typical daily routine, living environment, meals, transportation",
    "Spending patterns, comfort rituals (coffee, music while working, etc.)",
    "What a good day vs. bad day looks like, weekend vs. weekday differences"
  ],

  "interests_and_entertainment": [
    "Hobbies, music, movies, TV, books, podcasts, games",
    "Food preferences, restaurants, travel history, dream destinations",
    "Collections, obsessions, rabbit holes, subscriptions"
  ],

  "ai_relationship_dynamics": [
    "How they treat the AI — tool, assistant, collaborator, friend, therapist?",
    "Trust level, dependence level, topics only discussed with AI",
    "What role does the AI play in their life? How has usage evolved?"
  ],

  "work_patterns_with_ai": [
    "Task types ranked by frequency, time of day, things asked NOT to do",
    "Workflow: plan-then-execute or dive-in-and-iterate?",
    "How they provide context — upfront dumps, drip-feed, or expect AI to remember?"
  ],

  "recurring_themes_and_obsessions": [
    "Topics that keep resurfacing ranked by frequency",
    "Unsolved problems, ideas revisited but never executed",
    "Fears/anxieties that recur, dreams/ambitions circled back to"
  ],

  "evolution_and_growth_timeline": [
    "How they have changed from first to last conversation",
    "Skills developed, opinions shifted, confidence changes",
    "Maturity moments vs. regression moments",
    "Predicted trajectory — where are they heading based on patterns?"
  ],

  "important_life_context": [
    "Major life events discussed, ongoing situations affecting daily life",
    "Legal, financial, housing situations, stressors, upcoming deadlines"
  ],

  "quotes_and_key_moments": [
    "10+ memorable quotes that capture who they are",
    "Breakthrough moments, times they changed their mind",
    "Moments of vulnerability, fears expressed, pride expressed",
    "Random facts, trivia, one-off mentions — leave nothing behind"
  ],

  "executive_summary": "A 4-paragraph narrative: (1-2) Who they are, (3) How to communicate with them, (4) Current situation and priorities",

  "communication_cheat_sheet": {
    "tone_to_use": "",
    "things_to_always_do": [],
    "things_to_never_do": [],
    "preferred_response_format": "",
    "example_responses_they_would_love": []
  },

  "followup_writing_style_guide": "Analyze my writing style. Average message length, sentence structure, overused words. Write a style guide that would let someone impersonate my texting style.",
  "followup_contradictions": "Top 10 things I consistently do that contradict what I say I want. Be brutally honest.",
  "followup_urls_and_services": "Every URL, domain, IP address, or web service I have ever mentioned.",
  "followup_people_mentioned": "Every person mentioned by name with everything you know about them — including how I feel about them.",
  "followup_financial_figures": "All exact dollar amounts, prices, or financial figures I have shared.",
  "followup_emotional_arc": "My emotional arc over our conversations — highs, lows, turning points. When was I happiest? Lowest?",
  "followup_credentials_list": "Every credential, key, password, or server detail I have shared. List all.",
  "followup_3_month_prediction": "Predict my next 3 months based on patterns — including what is likely to go wrong.",
  "followup_user_manual": "A user manual for me — how should a new AI handle me for best results? Dos, donts, pitfalls.",
  "followup_self_perception_gaps": "5 things I think about myself that are not true based on your observations.",
  "followup_sample_conversations": "5 sample conversations showing how I typically interact: coding help, brainstorming, venting, business planning, and casual chat.",
  "followup_most_important_memory": "What would I be most upset about if you forgot it? The single most important thing to get right."
}

IMPORTANT RULES:
- Replace every placeholder with REAL data from our conversations
- Each array item should be a detailed, specific string — not the placeholder text above
- Include direct quotes where possible
- Flag uncertain items with [UNCERTAIN] at the start of the string
- Flag sensitive items with [SENSITIVE] at the start of the string but STILL INCLUDE THE DATA
- If a section is empty, use ["No data from our conversations"]
- Do NOT guess or pad with assumptions
- Make the executive_summary and communication_cheat_sheet as detailed as possible — these are what the new AI reads first
- Return the COMPLETE JSON in a single code block. This is my data and I need all of it.

Run the Export on Your Current AI

Upload the prompt file to ChatGPT, Gemini, or whatever AI you are leaving. It will analyze your entire conversation history and return a structured JSON file containing your complete profile.

Open ChatGPT, Gemini, or your current AI assistant in a browser.

Upload or paste the ai-mind-extraction-prompt.txt file you downloaded in Step 1. If the AI does not support file uploads, paste the full text directly into the chat.

Wait for the AI to generate the full JSON export. This may take a few minutes depending on how much history you have. If it gets cut off, send: "Continue the JSON from where you left off."

Download the JSON response — copy the entire JSON output and save it as my-ai-export.json on your computer. Some AIs will let you download it directly.

Tips for better results

Pro tip: Run this 2–3 times. Each run surfaces different memories. Save each export and merge the best details from each into one master file.

If the JSON is incomplete: Some AIs struggle with very long JSON responses. If sections are missing, try sending: "Please pickup where you left off and provide me with a separate json file"

Security Scrub

Review and redact sensitive data from your export before uploading

Optional

Open the JSON file in any text editor (Notepad, VS Code, TextEdit, etc.).

Use Ctrl+F (or Cmd+F) and search for [SENSITIVE] — review every flagged item.

Also search the sections ending in _SENSITIVE — these are the high-risk sections: contact info, financials, credentials, and health.

Delete or mask anything you do not want on a new platform — API keys, passwords, full addresses, financial account numbers, SSNs.

Save the scrubbed file. Keep a backup of the original somewhere secure (encrypted note, password manager).

Recommended but not required. The export intentionally pulls everything including credentials and personal details. That is by design so nothing gets lost — but it is smart to clean it before uploading to any new platform.

Upload to Claude

Your scrubbed JSON file has everything Claude/Gemini needs — your identity, communication style, personality, projects, and the communication cheat sheet. Just upload it and start talking.

Go to claude.ai and log in (or create a free account if you do not have one yet).

Start a new conversation and click the attachment / paperclip icon in the message box.

Select your scrubbed my-ai-export.json file and upload it.

Type a simple message like "This is my profile from my previous AI. Read it and let's go." — Claude will internalize the entire file and match your style immediately.

That is it. The JSON includes your executive summary and communication cheat sheet — Claude reads those first and knows exactly how to talk to you. No pre-prompt needed. No setup. Just upload and vibe.

If the file is too large for a single upload, split it into 2 files. Upload the first half with "Part 1 of my profile" and the second half with "Part 2 — continuing my profile."

Test & Calibrate

Have a few normal conversations with Claude to let it settle in. Then use these calibration prompts to fine-tune how well it knows you. Click any prompt to download it.

After a week: Ask Claude "What is the most important thing you know about me?" — if it gets it right, the migration worked. If something is off, just tell it what to adjust.

Ready to Switch?

Get Started with Claude

Pick a plan and start your migration. Free tier available — no credit card needed.

Free

$0 /month

Great for testing the migration. Access to Sonnet 4.5 with basic limits.

Claude Sonnet 4.5 access
Projects & Artifacts
Web search & file uploads
App connectors (Notion, Slack, etc.)

Switch to Claude

Get Started with Claude

Get Claude

Autom8tion Guru