> ## Documentation Index
> Fetch the complete documentation index at: https://docs.entryway.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# AI providers

> Choose which AI provider and model drafts for your workspace, and optionally use your own OpenAI or Google Gemini key.

Entryway drafts with OpenAI or Google Gemini. An owner or admin chooses the provider and model in **Settings → AI providers**. Other roles do not see this section.

## Provider connections

Each provider shows where its key comes from:

* **Using existing platform key**: Entryway's own provider account
* **Private workspace key**: a key your workspace saved
* **Not connected to Entryway**: the provider is not set up for your workspace yet

To use your own key, choose **Set workspace key** (or **Replace workspace key**) and paste it. Saving a key does not run a model.

<Info>
  Workspace keys are encrypted before they are stored, are never shown again or returned by any
  part of the app, and are decrypted only to make a model request. If secure key storage has not
  been set up for your workspace yet, the page says so and no key can be entered.
</Info>

## Test before you rely on it

Choose **Test connection** to check a provider. A test sends one fixed prompt and nothing from your Vault, files or drafts. If the provider rejects the key, has no billing or credit, hits a rate limit or is unavailable, the test says so and no retry is made. If a result is uncertain, check the provider's usage before testing again.

## Default for new requests

Under **Default for new requests**, choose a configured provider and an allowed model, then **Save default**.

* Changes apply to new requests. Existing drafts are unchanged.
* A failed request never falls back to another provider.
* Each request records the provider and model that produced it.

**Configuration history** lists every change to these settings.

## How your data is used

* Entryway sends the provider only what a draft needs, such as your instructions, relevant Business Vault context and the post or contact details involved.
* Requests to OpenAI are sent with response storage turned off.
* Your workspace content is not used to train AI models.
* If you use your own key, your agreement with that provider also applies.

See the [Privacy Policy](https://growth.entryway.ai/privacy) and [Subprocessors](https://growth.entryway.ai/subprocessors).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.