Assistant Settings

Assistant settings is where you configure how the AI Assistant behaves for your organisation. Here, you can set its name, choose its tone of voice, control how it uses your platform content when answering questions, and connect additional sources for it to reference.

Before you can access these settings, enable the AI Assistant from Configure > AI features, then click the Assistant settings button.

The screen is organised into the following sections.

Personalisation

Assistant Name

Enter a name for your AI Assistant in the Assistant Name field to change how it appears to learners across the platform. Click Save to apply.

Custom Instructions

Give your AI Assistant standing instructions to follow in every response – your terminology, your priorities, your ways of working.

To add instructions, enter them in the Custom instructions field (up to 4,000 characters) and click Save. For example: "Call our onboarding programme 'The Launchpad'. Say 'team members', not 'staff'. Keep responses concise."

Note:

You don't need to add policies or documents here – your AI Assistant already has access to your content. Short, clear instructions work best.

Thrive Labs Bot Prompts

Choose which example prompts your AI Assistant suggests when learners start a new conversation, instead of the default set shown to every customer.

There are two separate sets of four prompts, one for each type of screen:

• General screens: Shown wherever your AI Assistant opens outside of a specific piece of content.

• Content screens: Shown when your AI Assistant opens on a piece of content, such as a course or resource.

Click the General screens or Content screens tab, then click the dropdown arrow on the prompt you want to replace and choose a new one from the list. Prompts are grouped by category and picked from a fixed, pre-translated list – you can't write your own prompt text.

Note:

Only prompts compatible with your platform's enabled features are available for selection. For example, a prompt that relies on assignments will be unavailable to pick if assignments are restricted to admins only on your platform.

Changes apply to every learner immediately.

Behaviours

Tone of Voice

Select how you want your assistant to communicate. Choose the style that best fits your organisation's culture.

Response Behaviour

Control how your AI Assistant uses sources when responding to questions. Select one of the following modes.

Note:

If no response behaviour mode is configured, the assistant defaults to Strict sources only. This maintains consistency with previous behaviour and ensures existing customers experience no unexpected change in how the assistant respond.

Knowledge & Sources

Endorsed Content Only

Toggle Endorsed content only on to restrict the AI Assistant to Thrive content that carries the Endorsed flag when generating responses. Content that hasn't been endorsed is excluded, even if it's relevant to the learner's question.

Trusted Web Search domains and Connector Search sources are not affected by this setting and will still be referenced as normal.

Note:

This setting is off by default. If you turn it on when little of your content is endorsed, the AI Assistant will have fewer sources to draw on and may be unable to answer some questions.

Trusted Web Search

Toggle Trusted Web Search on to let the AI Assistant to reference approved public domains in its responses.

To add a domain, enter it in the Add New URL field and click + Add New URL. Added domains appear in the Approved URLs list, where they can be edited or removed at any time.

Connector Search

Connect your assistant to external file sources, such as Dropbox and Google Drive, so it can reference your organisation's documents when generating responses.

Click Connect next to a provider and sign in to select the files or folders you want to share.

Assistant Tools

The Assistant tools section lists the individual actions your AI Assistant can take on your platform – more than 80 in total, covering both read actions (such as searching content or looking up audience details) and write actions (such as creating events, updating group details, or changing space memberships). For each tool, you can control whether it's switched on, whether it needs approval before it runs, and whether it's restricted to administrators.

Each row in the list shows:

• Tool: The action's name and a short description of what it does.

• Capability: Whether it's a read or a write action, and the area it applies to, such as Content, Events, Audiences, Groups, or Spaces.

• Enabled: Switches the tool on or off for your tenant.

• Needs approval: For write actions, requires the signed-in user to review and approve the action before it runs. Not applicable to read actions.

• Admin only: Restricts the tool to administrators. Tools that expose personal data are tagged PERSONAL DATA and are restricted to administrators by default.

Use the search field or the filter icon to find a specific tool, and use the dropdown at the bottom of the list to change how many tools are shown per page.

Note:

Administrative actions are disabled by default. Enabling a tool only makes it available within the AI Assistant experience – every tool still runs using the signed-in user's own permissions, so users can only perform tasks they're already authorised to do.

Click Save tool settings to apply your changes.