Working with Prompts

Prompts in Hyperleap AI are pre-configured instructions that guide AI responses for your specific use cases. They act as templates that can be reused across your organization, ensuring consistent and accurate AI interactions while reducing prompt engineering effort.

Simple 4 step process to start using Prompts

Create and Configure

Design your AI feature in our API Console.

Define the Prompt template, add dynamic variables or import pre-configured Prompt from Prompt library
Set your preferred AI model
Add Sources to provide context
Fine tune the Prompt configuration

Detailed Guide: Create and Configure Prompt

Learn the step-by-step process of creating and configuring your first prompt

Save and Test

After configuration, your Prompt will have a dedicated REST API endpoint. You can test each Prompt in our playground prior to deployment.

After testing, you can publish the Prompt to your Workspace for reuse by other users

Integrate via API

To integrate with the API, perform a REST API call from your application. Send your variables in a JSON format, and the system will handle the AI context and state management. This is compatible with any programming language or framework.

// Example API integration using JavaScript
const response = await fetch('https://api.hyperleapai.com/prompt-runs/execute', {
  method: 'POST',
  headers: {
    'x-hl-api-key': 'your_api_key_here',
    'Content-Type': 'application/json'
  },
  body: JSON.stringify({
    prompt_id: 'your_prompt_id',
    prompt_version_id: 'your_prompt_version_id',
    replacements: {
      variable_name: 'variable_value'
      ...
    }
  })
});
 
const result = await response.json();

For detailed API specifications and more examples, check out our Prompts API Reference. You'll find complete documentation of all available endpoints, parameters, and response formats.

Deploy and Monitor

Ensure your Prompts deliver consistent responses as configured. You can monitor usage, track costs, and adjust settings without code changes. Access audit logs to review interactions and gather feedback for continuous improvement.

How Prompts Work

Understand how Prompts work in Hyperleap AI under the hood

Input Processing

Your Prompt template receives inputs from your configured fields. These inputs are validated and prepared for processing.

Variable Substitution

Dynamic variables in your Prompt are replaced with actual values, allowing for flexible and reusable Prompt templates.

Context Integration

Relevant context from your connected data sources is incorporated, enriching the prompt with necessary background information.

AI Processing

The final Prompt is processed by our AI engine to generate responses, leveraging advanced language models for optimal results.

Common Use Cases

Content Generation

Create templates for consistent content creation across different formats

Data Analysis

Structure prompts for extracting insights from data

Customer Support

Design response templates for common customer queries

Process Automation

Build prompts for automated decision-making workflows

Prompts can be shared with your team through Workspaces, with role-based access control to manage what each team member can do.

Permission Matrix

Action	Owner	Admin	Contributor	Reader	Guest
Run Prompt	✓	✓	✓	✓	✓
View Details	✓	✓	✓	✓	✗
Edit/Add Versions	✓	✓	✓	✗	✗
Delete	✓	✓	✗	✗	✗
Manage Permissions	✓	✗	✗	✗	✗

For more details on workspace roles, see Managing Workspaces.

Troubleshooting

Common issues you might encounter and how to resolve them:

Variable Substitution Errors

Issue: Prompt variables not being replaced with actual values.

Solutions:

Verify variable names match exactly in both Prompt template and input
Check for proper syntax: {{variableName}}
Ensure all required variables are provided in the request
Use the Studio interface to validate variable substitution

Context Length Exceeded

Issue: Total Prompt length including context exceeds model limits. Solutions: - Reduce context length by summarizing or chunking data - Use shorter Prompt templates - Consider using our automatic context truncation feature - Monitor token count in Studio before deployment

Response Format Mismatch

Issue: AI responses not matching expected format structure.

Solutions: