Copilot
AI Copilot is a powerful tool available to Boost Plan users and above. It transforms your internal knowledge base and external resources into an interactive, user-friendly support system. Integrated into the Life Ring Button, AI Copilot provides quick, accurate answers in real time, reducing the need for manual support.
Enabling & Configuring Copilot
AI Copilot is available as a “Chat” tab within the Life Ring Button, and you can enable or disable this tab based on your preferences. To enable Copilot feature navigate to Help → Copilot → click on the “Enable copilot” button.
You can set the configuration options and decide what is displayed by choosing between chat, Copilot, or both. There are three main configurations for AI Copilot, allowing you to customize how AI Copilot and live support work together. For this to work, you must have the Chat & Copilot toggle enabled under the Life Ring Button Configuration settings, as mentioned earlier. By clicking enabling the toggles, you can select from the following options:
- External Chat Only – Keep the traditional external chat in your Life Ring Button and contact a live agent directly. To learn how to integrate third-party chat providers and view the list of supported providers, refer to this article.
- Copilot Only – Use AI Copilot as the sole support system.
- Copilot + External Chat – AI Copilot acts as the first layer of support, but users can choose to connect with a live agent through a third-party provider. After receiving the first reply from AI, users can contact live support by clicking a button. The text on this button can be customized, and you can also define language variants for it.
If you enable external chat, you can select your preferred third-party chat provider from the integrated options in your application.
Targeting
You can decide which users have access to the Copilot and/or the support chat. The Targeting section allows you to define rules to select specific users who can access these features, in case you don’t want to enable them for everyone. Options include:
- Enabling AI Copilot for all users.
- Setting a custom rule to define access.
- Using a preexisting segment to specify eligible users.
To learn the difference between a segment and a custom rule, as well as how to create one, check out this article.
Additionally, URL filtering lets you control where Copilot and chat are displayed. You can choose to show them only on specific URLs or hide them on certain pages.
Localization
Under this section, you can customize the messages AI Copilot presents to users. You can:
- Modify the text on the support contact button.
- Define a personalized greeting for AI Copilot.
- Define the message when the search yields no results.
This ensures a tailored experience that aligns with your brand and user expectations.
Sources
AI Copilot connects directly to your internal knowledge base, providing users with up-to-date and accurate answers. The internal Product Fruits knowledge base syncs almost instantly, while external resources update automatically once a day. Additionally, you can manually trigger synchronization for external resources up to three times per day using the three-dot menu.
Managing Sources
A simple dashboard provides an overview of all sources. Here, you can see:
- Source: Clicking on the source will redirect you to a new page where you can manage individual pages or articles within that source.
- Synchronisation type: Manual or daily
- Last synchronisation time: Track the latest data refresh.
- Training mode toggle: Choose whether a source is for AI training only or is included in search results. Training mode means that a given source or page will be used by Copilot to generate answers, but it will not be shown to the user in the list of sources used. The information from the source will be applied in responses, but the source itself will remain hidden.
Adding External Sources
You can integrate external resources to enhance AI Copilot’s knowledge base. External sources update once daily by default, but you can manually sync them up to ten times a day if needed. Currently, external sources that are behind a login are not supported, but this feature is coming soon.
You have full control over:
- Selecting which parts of a source to include or exclude.
- Choosing automatic or manual sync options.
- Marking some sources as “train only” to improve the search results, without linking the source
- Adding rules to include/exclude specific paths.
Feedback & Insights
Understanding how users interact with AI Copilot is key to improving its effectiveness.
By navigating to Analyze > Copilot, you can access detailed insights into user interactions, including:
- Conversation title – Track individual user interactions.
- Date – See when each conversation took place.
- Upvotes & downvotes – Measure user satisfaction with AI responses.
- Helpfulness – Identify whether users marked AI Copilot’s answers as helpful.
To make analyzing feedback easier, you can filter conversations based on several criteria. The Helped / Not Helped filter lets you see how often AI Copilot successfully resolved user issues, with users manually marking responses as helpful. You can also sort conversations by Upvoted / Downvoted to identify which responses users found useful or unsatisfactory. Additionally, you can filter feedback by Language to assess AI Copilot’s performance across different regions or by Time Range to track trends and improvements over specific periods.
When reviewing individual conversations, you’ll have access to two tabs:
- Feedback Tab – Displays any user feedback, such as upvotes or downvotes, along with the name of the user who provided it.
- Conversation Tab – Shows the full chat, allowing you to review both user inputs and AI responses in detail.
These insights help refine AI Copilot’s performance, ensuring it delivers more accurate and helpful responses over time.
Improving AI Copilot Based on Feedback
By reviewing user interactions, you can identify gaps in AI Copilot’s responses and refine its resource pool accordingly. If certain topics frequently receive downvotes or negative feedback, consider updating your internal knowledge base or external sources to improve accuracy.
Additionally, analyzing repeated queries can help you discover common pain points. If AI Copilot struggles with specific types of questions, you may choose to:
- Expand the Knowledge Base: Integrate any missing details to ensure comprehensive and up-to-date content.
- Refine Source Selection: Optimize the choice of sources to enhance the relevance and accuracy of responses to user queries.
- Integrate High-Quality External Sources: Include new external sources that offer reliable and in-depth coverage of the relevant topics.
By regularly reviewing and acting on feedback, you ensure AI Copilot continues to evolve and provide better assistance over time.
Key Benefits
- Enhanced User Experience: Provides instant, AI-generated responses, reducing wait times and increasing user satisfaction.
- Improved Efficiency: Streamlines support by delivering quick, accurate answers, which minimizes the need for live assistance.
- Cost Reduction: Decreases reliance on live support, lowering operational costs.
- Seamless Resource Integration: Combines internal knowledge bases with external sources to offer the most up-to-date and comprehensive information.
- Personalized Assistance: Tailors responses using natural language processing, ensuring that answers are contextually relevant.
- User Feedback for Continuous Improvement: Incorporates a feedback mechanism (thumbs up/down and comments) to refine the support experience over time.
- Flexible Configuration Options: Allows to enable or disable the AI Copilot Chat tab and configure support settings based on unique needs.
- Enhanced Knowledge Base Utilization: Transforms static knowledge resources into an interactive, easily accessible support tool.
Security & Privacy
We use trusted AI models (OpenAI, AWS Bedrock, Google Gemini) running on AWS, Microsoft Azure, or Google Cloud. While your conversations are processed by these providers, we do not use any of your data to train the models. You control which information is indexed—no private or personal details are ever included automatically. Your information is processed securely, and we maintain ISO 27001, SOC 2, and GDPR compliance at all times