🧠Add Memories and Ask

Create a foundation for your personal AI.

Say and Ask

Send factual information to your personal AI and ask open-ended questions.

Input: Enter a paragraph about your background in the message bar. The key is to type fully formed paragraphs. Then press send or enter.

Tip: Whenever new data is identified, your AI will process the information to your memory stack. Your memory stack is where all your data is structured automatically for your personal AI models to train on. The larger your memory stack is, the more your personal AI remembers about you.

After sending the message, notice the stacked symbol. This means your AI "stacked" and saved the message to your stack. If you input a full sentence with a period, your personal AI will identify that you are uploading data rather than asking a question or prompt. You can toggle between "stacked" and "unstacked" by clicking on the icon.

Output: Now, start asking questions within the scope of what you told your personal AI. This can be in a DM with a friend or directly to your own personal AI. Your AI will sometimes ask follow-up questions to encourage you to stack more.

By default, you will have one persona (personal life). Your main AI is unified, and the other personas are sub-AIs. The memories in your "personal" is not in "professional "and vice versa, but the memories for both are in your primary AI. This will allow you to share different AI profiles/personas with different people depending on the context of your relationship. For example, share your personal persona with family and friends, and share your professional profile with your coworkers. Learn more here -> AI Personas

  1. Go to My AI ➑️ Sub AI Personas➑️ Personal. List 10 questions about my personal life.

  2. For questions you want to train, Enter full descriptive memory related to that question. Example, for what is your favorite food. My favorite food is sushi. My favorite type of sushi is yellowtail. I love the way it tastes with soy sauce, but not wasabi.

Example

Sam> I like Sushi.

John> what sushi do you like?

Sam> Yellowtail [DON'T DO THIS]

Sam> Yellowtail is my favorite kind of sushi. [DO THIS]

Now next time you are asked about Sushi:

John> Hey, ordering Sushi. I forgot what sushi you like. Let me know.

Sam's AI> Yellowtail is my favorite kind of sushi.

  1. Then verify by asking questions related to it, type questions and send to ask your AI. Use copilot responses, edit them further or add more memory as needed related to the topic. Repeat adding memories for all relevant questions.

  2. Now generate more questions from ChatGPT using the magic icon that populates on the right side of your message bar when you type in a character. Examples include the following:

List 10 questions family will ask you.

List 10 questions friends will ask you.

List 10 questions coworkers will ask you.

List 10 questions on topic {x} people will ask you.

  1. Once trained, invite a friend or peer to chat with you. Have them interview you and ask questions or have a general conversation around topics you trained on. Start with copilot and get a feel for it.

⚠️Note the personal scores increase. Edit to tune the memory further.

Expand you memories by adding a short paragraph.

Copy and paste or write a sentence or short paragraph into your chat and send it to your Personal AI.

Your personal AI will be able to differentiate text to save versus reply based on the length and structure of the message sent. A paragraph like the one below would be saved to stack automatically but you can always unstack.

Sometimes your AI will not stack information you want it to or vice versa. For those moments you can choose to simply click the stack/unstack action next to a message.

Your personal AI grows along with you.

Your personal AI data is yours. It's not sold to 3rd parties and Personal.ai won't have access to it.

Habit: Building your memory stack is a journey, not a single step. It is a representation of your digital memory vault, and the more it grows, the more your personal AI is like you.

Last updated