Getting started

Set up your core memories and assistant persona

Give your assistant useful starting context and choose how you want it to work with you.

Updated October 12, 2026

You can give your connected AI a useful starting point without writing a long biography. Add a few facts that affect its help and a few clear directions about how you want it to respond.

Open Core. The main cards have different jobs:

  • About you: standing facts about your work, goals, or circumstances.
  • Preferences: how you like things done, such as the format of a report.
  • Persona: how your assistant shows up for you, including tone and manner.
  • Memory policy: your directions about what to save, what to skip, and what to check before answering.

Choose Edit on an existing card, or the card’s writing button if it is empty. Read the saved text before changing it, add the details you want, and save. Keep temporary events in Notes so they do not become standing context.

You can also tell your AI: “Help me set up Penny. I prefer direct answers, and I want decisions saved with their reasons. Show me the proposed core memories before changing them.”

After a change, start a new conversation and ask the AI to read your Penny core memories. This checks that it has loaded the current version. A chat that already read an older version may need an explicit refresh.