Seamless AI-Powered Search Solution: Merging AEM Sites and Chatbottery for an Enhanced User Experience

AEM Sites with Document-based Authoring and Edge Delivery Services, formerly named AEM Franklin, is a modern, lightweight, and ultra-slim next-generation content management approach that goes “beyond headless”. It intelligently integrates with existing infrastructure such as the Google Workspace and Microsoft 365. However, one central and frequently requested feature - on-site search functionality - is currently difficult and costly to implement, often requiring the operation of an additional cloud instance of ElasticSearch.

Chatbottery, on the other hand, offers an intuitive conversational interface that utilizes context-sensitive keyword search for intent mapping in user conversations. The Chatbottery Editor has recently been extended with a ChatGPT interface to automate keyword tagging. Chatbottery can also be integrated with minimal overhead and runs entirely in the browser.

In this talk, we are going to demonstrate the integration of both approaches by extending Chatbottery's interface and search functionalities to index and automatically tag web pages created with Sites Document-based Authoring. Users will be able to search websites through an appealing conversational chat interface. OpenAI's ChatGPT will serve as the core technology for this solution. Metadata generated from the web preview markdown of Edge Delivery Services will be processed through ChatGPT prompts upon publication, indexing, tagging, and summarizing each page's content and structure. This metadata will then be converted into a Chatbottery JSON file, which will serve as the foundation for the chat search interface.

The resulting solution will be an easily integrable, AI-powered search functionality that reimagines search in a lightweight and innovative manner for both backend and frontend users.

Our presentation, including a live demo, will delve into the technical details and practical applications of this integration, demonstrating its value for website creators and users alike.

How well can it behave/search in natural language or with synonyms search ?
(see answer in talk video)