Reinvent personalization with generative AI on Amazon Bedrock using task decomposition for agentic workflows

This article presents an automated personalization solution that uses Amazon Bedrock and task decomposition to create personalized website content and visuals for business clients while adhering to company guidelines and curated assets. The solution utilizes generative AI models for various tasks like content personalization, image generation, and front-end development.

Specifically, the article covers:

• The use case of a fictional consulting company that generates personalized landing pages for business clients based on their profiles and industry insights
• An agentic workflow that breaks down the personalization process into subtasks handled by specialized LLMs for content personalization, image generation, and front-end development
• How human-curated sources of truth like design guidelines, testimonials, and offerings are incorporated into the personalization process
• The steps involved in generating personalized HTML, CSS, and JavaScript assets for the website, including example outputs
• Potential enhancements like batch processing, clustering client profiles, and using Haiku instead of Sonnet for cost optimization