Connect an AI provider and enable the customer Engineers

1. Open AI engineers & credits in the administrator workspace.
2. Select the Provider and an available Model. If required, use Or type a model name with the exact identifier supported by your provider account.
3. Enter the API key through the protected configuration field. The module encrypts saved secrets and shows a hint rather than the full value. Leaving a saved key field blank retains the existing key.
4. Review the free allowances, usage limits and paid-credit configuration before allowing customer requests.
5. Select Save AI settings, then Test the saved AI. Resolve authentication, model or provider-account errors before continuing.

Enable customer AI is the overall switch. WordPress Engineer and Hosting Engineer have separate controls. The WordPress AI website builder shares the WordPress Engineer switch, so consider both experiences when changing it.

The provider list contains multiple commercial services and permits model selection, but a listed provider does not guarantee every model will work with every task. Test the exact provider, model and account configuration you plan to use. Review the provider's current data handling and commercial terms for your own operation.

Open an eligible test service as a normal customer. Start with a read-only question about a known website or hosting setting. Confirm the selected Engineer responds, the credit or allowance display updates, and the action stays within the tools intended for that product.

Next test a small reversible change that requires approval. Confirm the customer can inspect the proposed action and decline it. Do not validate an AI integration by asking it to repair an unspecified live production problem.

Free stock photos can use a separately configured Pexels, Unsplash or Pixabay provider key. Verify image availability and the applicable image licence requirements before relying on automated selection.

The full WordPress website builder publishes pages while building and can activate its required theme. Use a new or staging website when content must be reviewed before public launch. Monitor early usage, provider failures and customer feedback, then adjust allowances and the supported workflow before expanding access.

Record the selected provider, exact model, allowance settings and date of your test without recording the API key itself. If you later change models, repeat the same representative questions and approved test action so you can compare useful output and actual consumption.

Keep a working way to disable the affected AI experience while investigating a provider outage or unsuitable result. Customers may still need ordinary website and hosting tools during that investigation. Review what happens to requests already in progress before assuming a switch cancels every action. When replacing a provider key, verify the new saved configuration first and then revoke the old key through the provider's supported account controls.

• Configure free AI allowances and service usage limits
• Price paid AI credits and check the protected minimum
• Set safe operating expectations for AI changes and Undo
• Troubleshoot AI access, failed requests and incomplete work