My AI Connections

By default, Ian On uses the AI provider your organization set up. If you'd rather run on your own account — your own ChatGPT or Claude subscription, or your own API key — you can add a personal connection. For your work, the runtime uses all your personal connections before any organization connection, then the organization AI defaults in the use order your admin set.

Open Settings and you'll find the My AI Connections section at the bottom of the page. Every connection here is yours (user scope) — organization-wide defaults are managed separately by admins at Admin → AI providers.

Adding your own

Under Add connection there are two ways to connect:

• OAuth — sign in with your existing ChatGPT or Claude account. (For Google models, use an API key.)
• API key — paste a key for a provider, then pick a model.

If you have several personal connections for the same provider (for example, two ChatGPT accounts), the filled star in the Default column marks the Default connection. Among accounts connected the same way, the default connection is used first. To change it, click the empty star (Set as default) on another connection.

Want to connect a self-hosted or on-premise model (vLLM, Ollama, LM Studio)? Use the same form, with a Base URL field and a Check button to discover the model. The full walkthrough — including organization AI defaults, use order, and automatic switching — is in Admin → AI providers.

Changing the model on a connected account

In the My connections table, use the Model list to pick from the models your ChatGPT or Claude account (OAuth connection) actually offers. Your pick is saved right away. The list opens only after the status check that runs when you open the screen has finished. If you see Couldn't load models, your saved model stays as it is; click Refresh status to check again. API key and local connections show only the model name in this table.

When you chat, you can pick Auto or a connected model from the model list in the message box. For details, see Home & Ian conversations.

Checking status and usage

In the My connections table, Status shows the connection's state and Usage shows the remaining usage your provider reports. If the provider sends a reset time, you'll see the date and time in your device's time zone along with the time left; if there's no reset information, you'll see Provider reported no reset time. API key connections show Usage check unsupported. Local and self-hosted connections have no usage limit, so if the server provides metrics, you'll see its current server capacity instead. To get the latest information, click Refresh status.