How to leverage knowledge base(s) for app creation

After you have chosen knowledge base as your orchestration method in the App Details page, click the 'Next' button at the bottom right corner to proceed to the Configure Bot page, where you will connect and manage your knowledge sources.

1. Configure bot

This section allows you to refine your bot settings by integrating the chosen orchestration method and configuring key parameters to match your specific requirements. These customizations ensure the bot functions seamlessly within your workflow, delivering an efficient, personalized, and purpose-driven experience.

Connecting and managing knowledge base(s) in your app

• Click the ‘Add’ button to choose one or multiple knowledge bases you wish to connect to your app.

• You can use the search function to quickly find specific knowledge bases from the list. After selecting the knowledge base(s), click ‘Add’ to connect them.

• The connected knowledge base(s) will immediately appear in the table with their names. To enable the Advanced Reasoning feature, tick the checkbox and choose a schema from the available options for the connected knowledge base to conduct detailed data queries.

• To remove a connected knowledge base, slide the bar to the right and click the trash icon to delete it.

Establishing system instructions

Define specific instructions or guidelines that the app must follow during user interactions.

Methods to add instructions

1. Manually – Type or edit the text directly in the System Instructions box.

2. Generate – Click ‘Generate’, provide a short prompt, and let the LLM create the instruction or provide a draft of the instruction that the LLM will refine.

1. Library – Click ‘Library’ to browse and import one of your pre-built prompt templates from ZBrain’s prompt library.

   1. Click ‘Add from Prompt Library’, choose the desired prompt from the list, and then click ‘Use’ to apply it.

• Check the 'Import Prompt Settings' box to automatically include the prompt text along with its associated settings, such as model type, temperature, top-p, and max token limits. This helps reduce manual setup, maintain consistency across applications, and streamline prompt reuse.

Writing guidelines (Applicable to all methods)

• Be specific and detailed– Clearly define how the bot should behave, including any specific instructions or constraints it must follow.

• Use natural, conversational language – Avoid unnecessary jargon unless the use case requires it.

• Incorporate context awareness – Instruct the bot to refer to previous turns or stored data so replies remain relevant and personalized.