- Getting started
- Capabilities
- Data privacy and governance
- Product availability
- Earlier Autopilot capability sets
- User scenarios
- Prompting guide
- Example prompts
- Autopilot usage pool for testing
- Best practices and limitations
- Supported models and limits
- Quality-check requirements
- Generate tests for requirement
- Import manual test cases
- Find obsolete tests
- Generate tests for SAP transactions
- Generate synthetic test data
- Generate test reports
- Generate coded test cases
- Search Test Manager project
- Troubleshooting
- Troubleshooting
- Autopilot for Everyone
- About Autopilot for Everyone
- Licensing
- User types
- Data sources
- Toolset automations
- Localization
- Prerequisites
- Autopilot widget
- The Autopilot for Everyone tenant card
- Prerequisites for installation
- Enabling Anthropic models
- Installing Autopilot for Everyone
- Updating Autopilot for Everyone
- Uninstalling Autopilot for Everyone
- Configuring Autopilot for Everyone
- Disabling the Autopilot welcome screen in Assistant
- Configuring an LLM for Autopilot for Everyone
- Deploying toolset automations
- Prompt-to-response flow
- Launching Autopilot for Everyone
- Autopilot settings for business users
- Using a specialized Autopilot
- Using a starting prompt
- Uploading and analyzing files
- Running automations
- Interacting with Autopilot answers
- Using suggested prompts
- Starting a new chat
- Chat history
- Providing general feedback
- Clipboard AI Enterprise version
- Designing relevant automations for Autopilot for Everyone
- Troubleshooting
Configuring an LLM for Autopilot
Configure your own LLM subscription for Autopilot via the AI Trust Layer.
The AI Trust Layer card in the Admin section of your organization allows you to configure your own subscription for the models Autopilot supports.
Autopilot supports Bring Your Own Subscription (BYOS) only: you can point any model already in the Autopilot model list to your own subscription. You cannot add a custom model or a custom alias for Autopilot.
Autopilot only offers models validated as proficient enough to build and fix UiPath automations, such as workflows, agents, and Flows. A model outside this list can perform well as a chat model but produce broken UiPath code.
To configure your own LLM, you need a connection to the following AI providers, depending on the model you want to incorporate:
- Amazon Bedrock or Amazon Web Services, for Anthropic models
- Google Vertex, for Gemini models.
Important:
To set up a Gemini model, you need to reach out to your designated UiPath technical account manager. If a Gemini model configuration fails in Automation Suite 2025.10, refer to Troubleshooting Autopilot.
When you select Autopilot as the product for which you want to configure your own LLM, you select the Feature: Generation or Chat. The feature determines which Autopilot capabilities use your configured model.
| Feature | Powers | Supported models |
|---|---|---|
| Generation | Autopilot actions: Generate expression and Fix expression in RPA workflows, and API workflow generation | Gemini only |
| Chat | Autopilot agent, including the classic Autopilot chat, and Autopilot as a coding agent | Refer to Configuring LLMs |
Autopilot agent and Autopilot as a coding agent work without a Generation configuration. If your organization can't use Gemini models, configure the Chat feature only.
To use your own model with Autopilot as a coding agent, configure it for the Chat feature. Studio Desktop 2025.10.17 and later supports configuring your own LLM for Autopilot Chat.
Generation
Configure your own LLM for Autopilot actions: Generate expression and Fix expression in RPA workflows, and API workflow generation. You need to override all the default models with your own:
-
Go to your Admin > AI Trust Layer section of your organization.
-
In the LLM configurations tab, select the tenant where you want to configure your LLM.
-
Select Add configuration and provide the following properties:
- Product - select Autopilot
- Feature - select Generation
-
Provide the Connections Folder.
-
For every default LLM Name, configure a new Connector and a Connection for your own model.
Chat
Configure a different LLM for reading the context of your current page or project and answering questions, explaining automations, or suggesting improvements. The model you configure overrides the existing model. Autopilot uses a primary model for complex tasks and an optional secondary model for simpler, cost-optimized tasks.
For Autopilot as a coding agent, configure one of the models that the Autopilot model dropdown offers, so that your configuration replaces it. A configuration for a model that the dropdown doesn't offer has no effect.
Before you configure the model, block its UiPath-managed version in the AI Trust Layer policy in Automation Ops. Refer to Product-level model controls.
-
Go to your Admin > AI Trust Layer section of your organization.
-
In the LLM configurations tab, select the tenant where you want to configure your LLM.
-
Select Add configuration and provide the following properties:
- Product - select Autopilot
- Feature - select Chat
-
Provide the Connections Folder.
-
Configure your new model.
To use your model in Autopilot as a coding agent, select the model your configuration replaces in the Autopilot model dropdown. The dropdown shows no visual difference between the UiPath-managed model and your model. Your configuration overrides the UiPath-managed model.
If you configure both a primary and secondary model, a tooltip displays both model names and confirms the secondary model is used for optimization. If you configure only the primary model, a warning flags the missing secondary configuration, and the primary model handles all tasks — including simpler ones the system would otherwise offload to a lighter secondary model. Leaving the secondary model unconfigured defaults to a UiPath-owned subscription for Automation Cloud organizations, or may limit chat features such as context compacting for Automation Suite organizations.
For details, refer to Setting up an LLM connection.
For the list of models Autopilot supports and their context-window and output-token limits, refer to Supported models and limits.
The Product and Feature you select here (Autopilot > Generation or Autopilot > Chat) determine which LLM configuration applies. Model hub shows the resulting model, provider, region, and status for products like Test Manager, which is non-selectable there — use the steps on this page (or remove the configuration) to change the model. Model hub shows only the outcome and configuration history.
You are responsible for the lifecycle of any custom LLM configuration you set up here. When the underlying model is deprecated, you are notified at least 30 days in advance, in-app and by email, and must migrate to a replacement before the deprecation date; otherwise requests through that configuration fail.
For details, refer to Model hub and Managing models in Model hub.