How do you create an AI digital persona?
Shape your own AI digital persona from experiences, memories, values, and expression through confirmation, conversation, and correction.
Updated September 6, 2026Short answer
Start with a meaningful experience, a thought, or an expression habit. Confirm MadeMe’s understanding, ask familiar questions, and correct what feels wrong. Keep adding new experiences and confirming updates so it can reflect more of who you are. Sharing is optional, not a requirement for completing creation.
Try one complete example: from a draft to a checkable answer
Sign in and open Shape. Create a draft and complete the identity and personality steps shown on the page. Under Memory, add an experience you want to keep. If you extract it from a file, review the parsed material and proposed memories before confirming them. Check the requirements on Generate, confirm generation, then open Chat to verify the result. Follow the labels shown in your current version.
The following career story is fictional. It illustrates a review process, not real user data or a promise of identical model output.
- Input: “In 2024 I moved from a large company to a small team because I wanted to influence the product directly, not because I disliked collaboration.”
- Review the draft: if it says “I prefer working alone,” correct it to “I value direct product impact and teamwork” before confirming the memory.
- After generation, ask “Why did you change jobs?” Check both the stated reason and the memory cited as its source.
- If the answer overgeneralizes, use its correction control: “Do not turn this experience into a rule that I always prefer small companies.” Preview and confirm the rule, apply any requested MadeMe update, then ask again.
- An appropriate illustrative answer: “That move was mainly about having a more direct impact on the product; teamwork mattered to me too.” For details not in the confirmed material, acknowledge uncertainty.
Stay in control after confirming
You can keep changing your material, personality details, and rules. Generation is a checkpoint in an ongoing process: confirm updates and use the same question to check their effect. Creating an invitation is optional; private use is a complete use case.
1. Choose material that matters
Begin with material that explains who you are, what you care about, and how you communicate. It might be a written memory, a document, a photo with context, or a voice sample recorded by you.
- Experiences: what happened and why it mattered.
- Memories: who was involved, when it happened, and which details matter.
- Values: how you usually judge tradeoffs and make choices.
- Communication: your usual tone, wording, and expressions you want to avoid.
2. Confirm understanding before it shapes answers
Information extracted from files, images, or audio should begin as a draft. Check facts, dates, relationships, and wording before confirming it as grounding for the persona.
This prevents incorrect material from silently influencing later answers and keeps you aware of what the AI has retained.
3. Calibrate through conversation
- Ask questions that someone who knows you well could answer.
- Check which memories support an answer, not only whether the tone sounds similar.
- Mark what feels like you or unlike you, then explain whether the fact, judgment, or wording is off.
- Preview changes before saving them as reusable rules.
4. Keep shaping it as your life unfolds
Add new experiences and thoughts, and update views or expression preferences that have changed. Review and confirm updates before using them in future answers.
Shape MadeMe at your own pace. You do not need to invite anyone to complete creation or provide your entire life story at once.
Optional: share when you are ready
If you want someone important to get to know you through MadeMe, create a revocable invitation for that person and choose the scope you want to share. You can also keep using and shaping your MadeMe privately without sharing it.
Continue reading
Frequently asked questions
Do I need a large amount of data to begin?
No. Starting with one meaningful experience is easier to verify and reduces the risk of providing too much personal information at once.
Should chat history automatically become memory?
No. A safer pattern is to extract proposed memories first and require the creator to confirm each one.
How long does it take to feel like me?
There is no fixed timeline. Accuracy depends on source quality, question coverage, and ongoing correction—not simply the amount uploaded.