Depending on how a persona is configured, inference may run on your Mac (with a downloaded model), on a network-hosted Persona, or via a cloud model you connect with your own API key. See the Privacy Policy for how each path handles your data.
Where to find them
- The Personas tab opens the list and management page, click + New Persona to start one.
Managing your personas
- Search by name, sort A–Z, and switch between grid and list views.
- Click + New Persona to create one.
- Click a persona card to chat with it, or click … on a card to Edit or Delete it.
- Each persona card shows its icon, name, role or purpose, and personality traits.
Chat with a Persona
- Open the Personas tab and click a persona card to start a conversation with that persona.
- Ask questions and get streamed replies shaped by its identity, traits, and system prompt.
- Click the microphone in the prompt window to dictate your message, then click Send: see Voice input in Chat for details.
- Each persona keeps its own thread, return anytime from the Personas tab.
Creating a persona (guided wizard)
The step-by-step builder has two steps by default:
A live success likelihood meter scores how complete the persona is as you fill it in. An advanced Configuration step is available but off by default.
Editing a persona
You can edit a persona in two places:- On the Personas tab, click … on a persona card and select Edit.
- Open Settings → Model → Personas, then click the pencil icon next to the persona you want to change.
Generate a system prompt with AI
You don’t have to write the system prompt yourself. On a persona’s details, fill in the Description and click AI Generate: the model reads your description and writes a system prompt from it. Use it to create a prompt from scratch or to improve one you’ve already started.Common questions
More answers are in the FAQ.