AI Engines

The AI Engines page is where you decide which AI providers (the “engines”) your organization can use. For each engine you connect a provider API key, validate it, choose which of that provider’s models people are allowed to use, set a default model, turn the engine on or off, and pick one engine as the organization-wide system default.

This page is for configuration only. It does not show usage or analytics — for that, see Usage Analytics.

Before you start

Find this page in the Control Hub under Manage → AI Engines.

To open the page and see your engines, you need the View AI Platform permission.
To change anything — toggling an engine on or off, adding keys, choosing models, or setting the system default — you need the Manage AI Platform permission.

If you have view access only, you can see every engine card and its current status, but the enable switch and the Configure button won’t appear. AI engine settings apply to your whole organization; there is no department-level configuration here.

What you’ll see on the page

Each connected provider appears as a card in a grid. A card shows:

A colored status dot and the provider’s name and icon.
A default engine badge, if this engine is the organization’s system default.
An enable/disable switch (manage permission only).
The connection status, and a masked preview of the API key if one is configured.
The engine’s default model and how many models are currently whitelisted.
A Configure button (manage permission only).

While the page is loading you’ll see a spinner. If no engines exist yet, you’ll see an empty-state message instead of the grid.

Turning an engine on or off

To quickly enable or disable an engine, use the switch on its card. The change applies right away. If something goes wrong saving it, the switch flips back to its previous state.

You can also change this inside the configure dialog, where it takes effect when you save.

Configuring an engine

Click Configure on a card to open the Configure Models dialog for that engine. It has three sections — API Key, Models, and a status toggle — plus Cancel and Save at the bottom.

Connecting and managing the API key

The API Key section adapts to what’s already set up:

If no key is connected, click + Add API key, type or paste the provider key, and click Add. Pressing Enter also submits.
Use the show/hide toggle next to the field if you want to read the key while typing.
Once a key is saved, you’ll see a masked preview along with Validate, Update, and Delete buttons.

A few things to know:

The key cannot be blank, and any surrounding spaces are trimmed when you save.
A newly added or updated key is marked not validated. Click Validate to confirm it actually works — you’ll see a success or error message with the result.
Update replaces the stored key. Delete removes it; you’ll be asked to confirm right there in the dialog before it’s removed.

Choosing which models people can use

The Models section lists the models the provider offers. Check the ones you want to make available (whitelist) in your organization.

Use Check all to whitelist every available model, or Uncheck all to clear the list.
Pick a Default model from the dropdown. This is the model the engine uses by default.
If the model list is still loading you’ll see a spinner, and if the provider returns no models you’ll see a “no models available” message.

Setting the organization’s system default

In the same section, check the system default box to make this engine the default for your entire organization. When you save, this engine (and its default model) become the org-wide default, and any previously default engine is automatically unset.

If this engine is already the system default, the checkbox is disabled — you can’t remove the default by unchecking it here. To change the default, set a different engine as the system default instead.

Saving

Click Save to apply your changes. The button shows a “Saving…” spinner while it works, then the dialog closes. Cancel closes the dialog without saving.

Good to know

At least one model is required. When an API key is connected, you must whitelist at least one model before you can save.
A default model is required once you whitelist models, and it must be one of the whitelisted models. When you first start whitelisting, the first model is selected as the default automatically.
You can’t uncheck the default model. Its checkbox stays locked, and Uncheck all deliberately keeps the default model whitelisted (you’ll see a warning that explains this).
You can’t remove the system default by unchecking it. If you try, the save is blocked and you’ll get an error message.
Validation problems (missing models, missing default, an out-of-list default) appear as a message and stop the save until you fix them.
API keys are stored securely on the server. The page only ever shows a masked preview, never the full key.
Provider icons automatically switch to match light or dark mode.