Introduction

The AI API provides a powerful interface for accessing language model capabilities. This documentation covers everything you need to know to effectively use the AI in your applications.

Endpoint Details

POST /api/ai_completion

The main endpoint for AI completions and interactions.

Request Format

The API expects a POST request with the following JSON structure:

{
  prompt: string,    // Instructions for the AI
  data?: any        // Optional context data
}

Prompt Structure

The prompt should include:

  • Clear instructions for the AI
  • Expected response format using TypeScript interface
  • Example response

Example Prompt Structure:

Generate 3 creative names for a pet store.

interface Response {
  names: string[];
  descriptions: string[];
}

{
  "names": [
    "Pawsome Pals",
    "Whisker Wonderland",
    "Tail Tales"
  ],
  "descriptions": [
    "A friendly neighborhood pet store",
    "Magical pet supply emporium",
    "Your pet's favorite story"
  ]
}

Response Format

The API returns JSON matching the specified TypeScript interface in your prompt.

⚠️ Important

Always validate the response against your expected interface to ensure type safety.

Examples

Best Practices

1. Clear Instructions

Provide specific, unambiguous instructions in your prompt

2. Type Safety

Always include a TypeScript interface for expected responses

3. Error Handling

Implement proper error handling for API calls

4. Context

Provide relevant context data when needed

Error Handling

async function callAI() {
  try {
    const response = await fetch('/api/ai_completion', {
      method: 'POST',
      headers: {
        'Content-Type': 'application/json',
      },
      body: JSON.stringify({
        prompt: '...',
        data: '...'
      })
    });
    
    if (!response.ok) {
      throw new Error(`HTTP error! status: ${response.status}`);
    }
    
    const data = await response.json();
    return data;
  } catch (error) {
    console.error('Error calling AI:', error);
    throw error;
  }
}