@modelence/ai

AI engine for Modelence applications with built-in telemetry, based on https://ai-sdk.dev

Installation

npm install @modelence/ai

Usage

generateText

A wrapper around AI SDK's generateText with built-in Modelence configuration and telemetry.

import { generateText } from '@modelence/ai';

const response = await generateText({
  provider: 'openai',
  model: 'gpt-4o',
  messages: [
    { role: 'user', content: 'Hello, world!' }
  ],
});

console.log(response.text);

Supported Providers

• OpenAI: provider: 'openai', models: gpt-4o, gpt-4o-mini, gpt-3.5-turbo
• Anthropic: provider: 'anthropic', models: claude-3-5-sonnet-20241022, claude-3-haiku-20240307
• Google: provider: 'google', models: gemini-1.5-pro, gemini-1.5-flash

Configuration

The function automatically uses API keys from Modelence configuration:

• OpenAI: _system.openai.apiKey
• Anthropic: _system.anthropic.apiKey
• Google: _system.gemini.apiKey

You don't need to manually set any of these configs as long as your application is using a Modelence Cloud backend - simply use the AI > Integrations tab in your Modelence Cloud dashboard to configure keys, and it will be automatically used and recognized by this package.