Guide Index
Guides HomeMangoApps WebsiteCommunity & SupportBlog & Resources
MangoApps AI Guide
MangoApps AI Guide
  • 🥭MangoApps AI Guide
  • Overview
    • Overview of MangoApps AI Studio
    • Introduction to AI Assistants
      • HR Support AI Assistant Use Case
      • Company Knowledge AI Assistant Use Case
      • IT Support AI Assistant Use Case
  • User Portal
    • User Getting Started
    • 🆕AI Assistants
      • 🆕Using AI Assistants
      • 🆕Out of the Box AI Assistants
    • AI Agents
      • Out of the Box AI Agents
      • Using AI Agents in Trackers
  • Admin Portal
    • Admin Getting Started
    • AI Admin Role
    • 🆕AI Studio Module
      • 🆕AI Studio Insights
      • AI Assistants
        • Create an Assistant
        • 🆕Writing Instructions for AI Assistants
        • Use Case Examples
          • Setting up an HR Support AI Assistant
          • Setting up an IT Support AI Assistant
      • AI Agents
        • Create an Agent
      • AI Service Providers
      • Templates
      • 🆕AI Settings
    • Understanding Your Knowledge Base
    • Navigation Menu
    • Widget
  • Bring Your Own Assistant
    • 🆕Custom Embed Assistant (BYOA)
    • Integration with AWS Bedrock
  • External Knowledge Bases
    • 🆕Overview of External Knowledge Bases
    • 🆕Knowledge Base: Confluence (Basic Authentication)
    • 🆕Knowledge Base: Confluence (OAuth 2.0 Authentication)
    • 🆕Knowledge Base: SharePoint
    • 🆕Web Page Crawler
  • Mobile
    • Download the App
  • Additional Resources
    • AI Studio by MangoApps
Powered by GitBook
On this page
  • Overview
  • Best Practices for Writing AI Assistant Instructions
  • Best Practices for Writing Prompts
  • Managing & Refining AI Assistants
  • Setting AI Usage Guidelines for Your Organization
  1. Admin Portal
  2. AI Studio Module
  3. AI Assistants

Writing Instructions for AI Assistants

Overview

AI Assistants are flexible tools designed to retrieve and generate information based on configured instructions and knowledge bases. Because there is no single "correct" way to write prompts or instructions, this guide provides best practices for both configuring AI Assistants and writing prompts.

Additionally, because AI Assistants work best when continuously refined, this guide also includes strategies for improving Assistants over time and ensuring they remain aligned with business needs.


Best Practices for Writing AI Assistant Instructions

Admins setting up AI Assistants should provide clear, structured, and constraint-driven instructions to ensure predictable behavior. Here are key guidelines:

Define the Assistant’s Role Clearly

  • ✅ "You are the RFP Assistant. You assist employees by retrieving past RFP responses and providing company-approved answers."

Specify the Knowledge Base Scope

  • ✅ "Only respond based on the uploaded knowledge base. Do not provide external recommendations."

Outline Key Tasks & Capabilities

  • ✅ "You help employees retrieve past case studies and summarize key messaging points from previous marketing materials."

Guide Response Formatting

  • ✅ "Organize responses in bullet points for clarity and cite relevant documents where applicable."

Anticipate Edge Cases & Avoid Overgeneralization

  • ✅ "If a user asks a question outside the scope of available documentation, inform them that the requested information is unavailable and suggest they contact [relevant team]."

Prevent Speculative or Inaccurate Responses

  • ✅ "Do not generate new policies or modify existing responses. Only retrieve and present information as written in the uploaded knowledge base."

By following these principles, admins can ensure Assistants remain focused, reliable, and aligned with company-approved content.


Best Practices for Writing Prompts

Employees interacting with AI Assistants should phrase their prompts in a way that ensures clear, relevant, and useful responses. Here’s how to optimize prompt-writing:

Be Specific About the Information Needed

  • ✅ "What is the correct process for handling a customer refund?"

  • ❌ "Tell me about refunds." (Too vague)

Use Keywords That Relate to Uploaded Content

  • ✅ "Find past blog posts about sustainability initiatives."

  • ❌ "What are some general sustainability trends?" (Might not retrieve company-specific data)

Use Step-by-Step Requests for Complex Topics

  • ✅ "List the steps for processing a customer refund according to our policy."

  • ❌ "How do I handle refunds?" (Too broad—Assistant may miss details)

Avoid Multi-Question Prompts in a Single Query

  • ✅ "What is our standard process for refunding a customer?"

  • ❌ "How do we process a refund, when does it get approved, and who should the customer contact if they have a problem?"

Clarify What Type of Answer Is Needed

  • ✅ "Summarize our policy on data retention in 3-4 sentences."

  • ❌ "Tell me about data retention."

Ask Follow-Up Questions to Refine Responses

  • ✅ "That’s helpful—can you clarify the timeframe for refunds?"

By guiding users to ask focused, structured questions, organizations can improve the quality of AI-generated responses and enhance productivity.


Managing & Refining AI Assistants

AI Assistants are not static tools—they work best when continuously improved based on real-world interactions. Regular updates to instructions, prompts, and knowledge bases help ensure accuracy and effectiveness.

Regularly Improve the Assistant Based on Feedback

Start with a Baseline, Then Iterate

  • Begin with clear instructions and structured prompts, then adjust based on user feedback

Monitor AI Performance

  • Track whether employees are getting accurate, useful answers

  • Identify and adjust instructions that lead to unclear or incomplete responses

Expand and Update the Knowledge Base

  • If users frequently ask about a topic that isn’t well-documented, add relevant content

  • Keep policies, procedures, and messaging aligned with the latest company updates

By continuously refining the Assistant, organizations can ensure AI remains a trusted and effective tool for employees.


Setting AI Usage Guidelines for Your Organization

Since AI Assistants are highly flexible, organizations should consider defining internal AI usage policies to ensure employees understand when, how, and to what extent they should rely on AI-generated responses.

Define Internal AI Usage Expectations

  • Should employees use AI strictly for information retrieval, or can they rely on AI-generated content with human oversight?

  • Are there specific topics or scenarios where AI responses should always be reviewed before being used externally?

Ensure Employees Are Trained on Effective AI Use

  • Provide guidance on how to write clear, structured prompts for the best results

  • Educate employees on AI limitations and when to escalate questions to a manager

Reinforce Responsible AI Use

  • Make sure employees understand that AI outputs are only as reliable as the knowledge base provided

  • Encourage employees to verify AI-generated responses for accuracy and compliance with company policies


By setting clear AI policies and ensuring employees are properly informed, companies can maximize AI’s value while maintaining accuracy, security, and trust in its responses.

PreviousCreate an AssistantNextUse Case Examples

Last updated 3 months ago

🆕
🆕