Personas

A persona defines how your AI behaves — not just what it says, but what it knows, what it can do, and how it approaches your requests.

​

More than a prompt

You could think of a persona as a job description for the AI. It bundles together:

• A system prompt that sets the tone, focus, and constraints — like telling the AI “you are a research assistant that prioritizes accuracy and cites sources.”
• A model and adapters so the AI can be tuned for specific domains without downloading an entirely new model.
• Capabilities and permissions that control what the AI is allowed to do — whether it can use tools, remember past conversations, access your files, or browse the web.

This is what makes a persona more than just changing what the AI is told. It changes what the AI is for that context.

​

Creating and switching

You start with a default General Assistant persona. From there, you can create as many as you need — each with its own configuration. Switching is instant. Pick a different persona from the dropdown and every new response reflects that context. The model doesn’t reload; only the behavior changes. This makes it practical to move between tasks throughout the day without friction.

​

Examples

​

LoRA adapters

Personas can optionally attach a LoRA adapter — a small file that fine-tunes the base model’s behavior for a specific domain. For example, a medical research persona might load the PubMedQA adapter, while a creative writing persona might use the QwQ Creative adapter. Adapters layer on top of your current model without requiring a separate download. They change how the model responds, not just what it’s told in the system prompt.

​

Execution modes

You can activate multiple personas at once and control how they interact: Chained execution is useful when you want different viewpoints on the same question — for example, a research persona provides facts, then a writing persona synthesizes them into a summary.

​

Presets

webAI ships with built-in persona presets — ready-made configurations you can use immediately or clone and customize. The default General Assistant persona is a good starting point for most tasks.

​

Your persona on the network

If you choose to enable it, your persona can serve as a resource on the Intelligence Delivery Network. Approved contacts can send questions to your device, and your persona generates a response locally — only the answer is sent back.

• Your data, files, and context stay on your machine.
• You control who has access.
• You control how much context the AI draws from when answering.

​

Connected Personas

While all user-created and preset personas run entirely on your device, webAI includes one built-in persona that operates differently. The HR Persona appears automatically in your contacts list and processes queries through a remote webAI-hosted service rather than your local AI runtime. When you send a message to the HR Persona, your query is transmitted to api.webai.com and the response is returned over the network. All other personas — including the default General Assistant, the UltraChat and DPO Aligned presets, and any personas you create — run fully on-device using your local model and inference backend. No data leaves your machine when using these personas.

​

Learn more