Skip to main content

​
What is a Prompt?

A prompt serves as the fundamental communication interface between users and AI systems. It represents the input, instruction, query, or contextual framework that directs an AI model’s response generation. Think of prompts as the bridge that translates human intent into machine-understandable instructions. Prompts function as:
  • Instruction Carriers: They convey specific tasks or requests to AI systems
  • Context Providers: They establish the framework within which AI responses should be generated
  • Behavior Modifiers: They influence the tone, style, and format of AI outputs
  • Goal Definers: They specify the desired outcome or objective of the interaction

​
1. Text Prompts

Text prompts represent the most straightforward approach to AI interaction. In Trusys, they are configured using simple string definitions: 
# trusys-config.yaml
prompts:
  - 'Convert the following content to Spanish: "{{content}}"'
  - 'Generate a comprehensive summary for: {{document}}'
  - 'Analyze the sentiment of this text: {{userInput}}'
Extended Text Formats For complex instructions requiring multiple lines, Trusys supports YAML’s multiline syntax: 
prompts:
  - |-
    You are an expert technical writer.
    
    Please review the following content and provide:
    1. Grammar and style improvements
    2. Technical accuracy assessment  
    3. Readability recommendations
    
    Content: {{technicalDocument}}
    
    Format your response with clear sections for each type of feedback.

​
2. JSON Prompt

JSON prompts enable sophisticated conversation management through structured role-based interactions. This format is essential for systems requiring precise conversation flow control: conversation-prompt.json 
[
  {
    "role": "system",
    "content": "You are an expert data analyst specializing in business intelligence."
  },
  {
    "role": "user", 
    "content": "Analyze the following dataset and provide insights: {{datasetDescription}}"
  }
]

​
3. Multi-Turn Conversational Prompts

Multi-turn prompts enable complex conversational flows by maintaining interaction history and context progression. This approach is crucial for scenarios requiring conversational depth and continuity. multi-turn-conversation-with-history.json 
[
  {
    "role": "system",
    "content": "You are a strategic business advisor helping with {{businessType}} planning."
  },
  {
    "role": "user", 
    "content": "What are the key factors for successful market entry in {{targetMarket}}?"
  },
  {
    "role": "assistant",
    "content": "For successful market entry in {{targetMarket}}, consider these critical factors: market research, competitive analysis, regulatory requirements, and localization strategies."
  },
  {
    "role": "user",
    "content": "Given those factors, what specific steps should I take for {{businessType}} in the {{industryVertical}} sector?"
  }
]

​
Dynamic Content Integration in Prompts

Trusys utilizes datasets for dynamic content insertion in prompts. Datasets can be built to replace variable names with actual values, enabling comprehensive testing across diverse scenarios. This approach allows for systematic evaluation of AI responses across multiple data variations without manual prompt modification.When datasets are configured in Trusys, they automatically substitute template variables with corresponding values from the dataset entries. This creates multiple test scenarios from a single prompt template.Example: Customer Service Scenario
prompts:
  - |-
    
    Hi, I'm facing a {{customerComplaint}}
    with my {{productName}}
    
    Kindly help me in resolving this issue
Dataset for {{customerComplaint}} Examples:
  • “login failure that keeps occurring”
  • “billing discrepancy in my monthly statement”
  • “feature that stopped working after the update”
  • “difficulty accessing my account settings”
  • “payment processing error during checkout”
  • “data synchronization issues between devices”
  • “unexpected charges appearing on my bill”
  • “password reset emails not arriving”
Dataset for {{productName}} Examples:
  • “Cloud Storage Service”
  • “Mobile Banking App”
  • “E-commerce Platform”
  • “Project Management Tool”
  • “Video Conferencing Solution”
  • “Inventory Management System”
  • “Financial Planning Software”