The Configurations API serves as the central configuration hub and entry point for Content AI. This API enables organizations to define end-to-end workflows that govern content acquisition, processing, indexing, and generative AI behavior, providing comprehensive control over how sites are discovered, content is ingested into Content AI indexes, and how generative search responses are constructed.

Key Features:

• Site Acquisition Configuration: Defines comprehensive workflows for site discovery and content acquisition (of content not already managed by AEM), including crawling parameters, source identification, and content ingestion pipelines that systematically process websites and feed content into the Content AI indexing infrastructure.

• AI Answers Behavior Configuration: Controls all aspects of AI Answers API behavior, including prompt engineering, response generation parameters, retrieval strategies, and AI model configurations that determine how natural language queries are processed and answered.

• Multi-Step Workflow Orchestration: Manages sophisticated 4-step pipelines encompassing discovery (site crawling), parser (content extraction), index (content organization), and generative (AI configuration) steps, enabling end-to-end content processing automation.

• Solution-Wide Configuration Management: Acts as the foundational configuration layer that orchestrates multiple Content AI services, enabling organizations to build comprehensive content intelligence solutions by defining how different services interact and collaborate within the ecosystem.

• Configuration Versioning and Lifecycle Management: Supports configuration versioning, updates through JSON Patch operations, and complete lifecycle management, ensuring organizations can evolve their Content AI configurations while maintaining stability.

The Indexes API provides endpoints that facilitate interaction with the search infrastructure for a given tenant. It enables clients to discover available indexes managed by Content AI and retrieve configuration information about each index. This information supports effective query creation and efficient data retrieval from the search endpoint.

Key Features:

• List Available Indexes: Get a complete list of all Content AI indexes that are currently available for the provided tenant. This feature helps users identify data sources for search operations.

• List Index Configuration: Retrieve configuration of a specified index, including various vector and lexical spaces that are defined along with their description. This information is essential for constructing queries to the search API. Users can use the descriptions provided to decide which space to use in their queries.

The AI Search API enables powerful retrieval of content across one or more indexes for a given tenant. It supports a wide range of modern information retrieval needs through fulltext, semantic, and hybrid (composed) queries. With support for complex filtering, flexible query composition, and advanced search parameters, this API provides users with the tools needed for sophisticated and efficient data discovery.

Key Features:

• Multiple Query Types: Supports semantic, fulltext, and filter queries individually or in combination, allowing retrieval by meaning, keywords, or structured criteria.

• Query Composition: Enables hybrid searches by composing multiple queries—such as combining filters with fulltext and vector searches—for highly precise and relevant results. Vector and fulltext queries can also be combined to leverage the strengths of both approaches.

• Flexible Filtering: Allows users to define filter groups, apply filters alone, or combine filters as pre-filters within either a vector or a fulltext query for targeted results.

• Search Features: Includes cursor-based pagination for efficient navigation through large result sets, and sorting to order results as needed (e.g., by score, timestamp etc).

• Vector Query Enhancements: For vector-based (semantic) queries, supports retrieval of both original documents and chunked (subdocument) representations for greater result detail.

• Multi-Index Search: Enables searching across multiple indexes simultaneously, maximizing coverage and simplifying cross-dataset discovery.

• Configurable Vector and Lexical Spaces: Users can explicitly select which vector (semantic) or lexical (fulltext) space within an index to use for each query, supporting rapid A/B testing and comparative evaluation of different configurations. This is a very powerful feature that gives the ability to evaluate and fine-tune various parts of the Index configuration in parallel.

The AI Answers API provides AI-powered natural language response generation using advanced Retrieval-Augmented Generation (RAG) mechanisms. This API transforms user queries and contextual metadata into comprehensive, brand-customized answers by leveraging hybrid search capabilities to retrieve relevant content and generate contextually appropriate responses through configurable prompt instructions.

Key Features:

• RAG-Powered Response Generation: Utilizes sophisticated Retrieval-Augmented Generation (RAG) techniques that combine the user's natural language query with metadata to locate and retrieve the most relevant excerpts from site pages before generating comprehensive, contextually accurate answers.

• Hybrid Search Integration: Leverages Content AI services' powerful hybrid search capabilities that combine semantic understanding with metadata-based composition, ensuring retrieved content is both semantically relevant and contextually appropriate for the user's specific needs.

• Dual Response Modes: Supports both blocking (synchronous) and streaming (SSE) response patterns, enabling flexible integration approaches— from simple request-response workflows to real-time streaming experiences that provide progressive answer generation.

• Brand Customization Support: Enables configurable prompt instructions that allow organizations to customize the tone, style, and approach of generated responses, ensuring answers align with brand voice, compliance requirements, and specific organizational communication standards.

• Contextual Metadata Processing: Accepts flexible metadata parameters that enhance query understanding and content retrieval precision, supporting geography-specific responses, user role-based filtering, and other contextual customizations.

• Citation and Follow-up Generation: Automatically generates source citations from retrieved documents and suggests relevant follow-up questions, providing users with transparency into answer sources and natural conversation continuation pathways.

• QMA Tracking Integration: Creates unique QMA (Query/Metadata/Answer) identifiers for each interaction, enabling seamless integration with feedback collection systems and comprehensive analytics tracking across the generative search ecosystem.

The Feedback API provides endpoints that enable end-users to provide structured feedback on generative search results. This API captures user satisfaction through boolean ratings (like/dislike) and optional detailed reasoning, creating a critical feedback loop for improving the quality of AI-generated answers over time.

Key Features:

• Binary Feedback Collection: Captures user satisfaction as a simple boolean value (liked: true/false), easily mappable to common UI patterns like thumbs up/down, like/dislike buttons etc.

• Detailed Reasoning Capture: Accepts optional reason parameter allowing users to provide specific explanations for their feedback, enabling deeper understanding of why answers were considered helpful or inadequate.

• QMA Context Association: Links feedback directly to specific QMA (Query/Metadata/Answer) combinations, ensuring that user ratings are associated with the exact context of the original user query, provided metadata, and generated response from the generative search API.

• Evaluation Engine Integration: Serves as the foundational data source for building sophisticated evaluation engines that analyze feedback patterns to identify areas for improvement in answer quality, relevance, and user satisfaction across different query types and contexts.