05 · Configure

Settings

A working AI connection and model are the only essentials for a first result. The other choices can wait until you need them.

First-time setup: a local model

TextAid is configured for a local Ollama connection by default. Ollama and a compatible model need to be installed and running on this computer.

  1. Open the TextAid application menu (☰) or its Windows tray menu, then choose Settings.
  2. Open This device only and leave Active selected. The default loopback endpoint is http://127.0.0.1:11434.
  3. Choose Test connection. If the connection is ready, choose Load local models and select one from Model.
  4. Choose Save. Return to your first result.
This device only settings with local Ollama ready and a model selected
A ready local connection and selected model. The model name shown is an example, not a requirement.
If no model appears: install a compatible model in Ollama, then use Load local models again. If Test connection fails, check that Ollama is running and the endpoint is correct.

Choose where AI runs

Settings has four independently configured connection types within three locations: this device, your administered network, and External. Each connection has its own Active setting and selected model. OpenAI sign-in is inactive and disconnected until you set it up.

TabHow to configure it
This device onlyUse local Ollama, test its loopback endpoint, load installed models, choose one, then Save.
On-premisesActivate it, enter the root URL of an Ollama server on your administered network, test the connection, load its models, choose one, then Save. Do not append /v1 to an Ollama endpoint.
ExternalConfigure an OpenAI-compatible service with its endpoint and API key, test access, load models, choose one, activate it, then Save. This remains available independently of OpenAI sign-in.
Sign in with OpenAISign in through your browser, authorize an eligible ChatGPT plan, load the offered models and select one, activate the connection, then Save. No API-key field is needed.

Both OpenAI access methods are External: text sent through either leaves your device. TextAid cannot verify who owns or handles data at a user-configured network endpoint. Configure these connections deliberately, and keep credentials out of action prompts.

Model profiles

The Profiles tab shows local-default, network-default, external-default, and openai-account-default with their connection categories and configured models. In Actions, choose a saved profile for an action. The workspace also offers a temporary model choice when several profiles are available.

Connect with your OpenAI account

  1. Open Settings → Sign in with OpenAI and choose Continue with ChatGPT.
  2. Complete sign-in and any MFA steps in your system browser. Review and grant ChatGPT plan usage permission for TextAid. Return to Settings and wait for the final connected status: receiving the browser callback alone does not complete sign-in.
  3. Choose Load external models, then select an offered model from Model. Available models depend on your account and workspace permissions.
  4. Use Test connection if you want an explicit check. This runs a synthetic text transformation; sign-in and model discovery do not generate text.
  5. Select Active and choose Save to make the configured connection available for action routing.
Settings Sign in with OpenAI tab with Active selected, verified plan authorization, and a selected model
An authorized OpenAI connection. The model shown is an example; choose from the models offered to your account.

You can cancel a sign-in attempt; browser and MFA completion can take up to fifteen minutes. A failed replacement sign-in preserves a previously valid session. The renewable session is protected for your Windows user on this device and retained across restarts.

Deactivate, sign out, or reconnect

Clear Active and save to stop using this connection while retaining its session. Sign out clears local tokens and attempts remote revocation. If TextAid reports local-only sign-out, remove TextAid access in ChatGPT settings when you also want to revoke it remotely.

Identity sign-in alone is insufficient for generation: ChatGPT plan usage must be authorized. If permission expires, is revoked, or the session cannot be read, reconnect with Continue with ChatGPT. If no models are offered or the chosen model is unavailable, check account permissions and load the catalogue again. Usage limits still apply; TextAid does not retry generation or switch to the API-key connection automatically after a failure.

Languages and shortcuts

Open Language in Settings. Application language is used for the interface and for quick Translate routing. Preferred output language for quick Translate chooses the other destination. These controls are separate from the output language you choose for an individual action.

The defaults are Ctrl+C, then C for Choose and Ctrl+C, then T for Translate. Edit the Choose shortcut or Translate shortcut fields and choose Save. TextAid accepts valid, non-conflicting sequences; another Windows application may consume a reassigned shortcut first.

Language settings showing application language, preferred quick-translation language, and both shortcut fields
The interface-language choice and quick-translation destination are different settings. This example keeps the UI in English.

For interface translation generation, personal corrections, and language packs, continue to Interface language.

Other preferences

In General, choose whether TextAid starts when you sign in to Windows. This setting applies to your Windows user account. You can also reload the saved configuration there.

In Debug, optional diagnostic logging is off by default. The ordinary log records technical metadata; Full log can include input, prompts, output, and raw errors. Review and remove sensitive passages before sharing a full log. TextAid does not send logs automatically, and credentials are redacted.