Skip to main content

Ecommerce materials

This code sample demonstrates how to create e-commerce materials (product description, images, and audio) using IntelliNode and different AI models like OpenAI, Google, Stability, and Cohere. You can use the following code functions:


const IntelliNode = require('intellinode');

1. Generate product description​

The generateProductDescription() function utilizes the RemoteLanguageModel class, which fetches the product description from an AI model like OpenAI or Cohere. Provide the appropriate apiKey, modelBackend, and modelName.

async function generateProductDescription(textInput, apiKey, modelBackend) {
  // available models: OPENAI or COHERE
  const modelName = (modelBackend === IntelliNode.SupportedLangModels.OPENAI) ? 'text-davinci-003' : 'command';
  const langModel = new IntelliNode.RemoteLanguageModel(apiKey, modelBackend);
  const results = await langModel.generateText(new IntelliNode.LanguageModelInput({
    prompt: textInput,
    model: modelName,
    maxTokens: 300
  }));
  return results[0].trim();
}

2. Generate image description​

The getImageDescription() function uses the Chatbot class to generate tuned image description from user message. The users might not enter a text suitable for image generation models and this layer will ensure the image prompt quality.

async function getImageDescription(textInput, openaiKey) {
  const chatbot = new IntelliNode.Chatbot(openaiKey);
  const input = new IntelliNode.ChatGPTInput('generate image description from paragraph to use it as prompt to generate image from DALL·E or stable diffusion image model. return only the image description to use it as direct input');
  input.addUserMessage(textInput);
  const responses = await chatbot.chat(input);
  return responses[0].trim();
}

3. Generate images​

The generateImage() function uses the RemoteImageModel class, which generates images from the description text. The generated images use stable diffusion or DALL·E models.

async function generateImage(imageText, apiKey, modelBackend) {
  // available models: OPENAI or STABILITY
  modelBackend = IntelliNode.SupportedImageModels.STABILITY
  const imgModel = new IntelliNode.RemoteImageModel(apiKey, modelBackend);
  const imageInput = new IntelliNode.ImageModelInput({
    prompt: imageText,
    numberOfImages: 3,
    width: 512,
    height: 512
  });
  return await imgModel.generateImages(imageInput);
}

4. Generate speech​

The generateSpeech() uses the RemoteSpeechModel and AudioHelper classes, which generate and save audio content based on the text input.

async function generateSpeech(textInput, apiKey, modelBackend) {
  // modelBackend = IntelliNode.SupportedSpeechModels.GOOGLE
  const speechModel = new IntelliNode.RemoteSpeechModel(apiKey);
  const input = new IntelliNode.Text2SpeechInput({ text: textInput, language: 'en-gb' });
  const audioContent = await speechModel.generateSpeech(input);
  return IntelliNode.AudioHelper.decode(audioContent);
}