# Developer Guide for eGain Composer Platform

## Overview

This guide covers eGain Composer, a comprehensive AI-powered knowledge management and retrieval engine that enables intelligent content discovery, agentic workflows, and personalized user experiences across multiple interfaces. It outlines how you, as a developer, can interact with eGain’s services—from ingestion to delivery—across various channels such as web portals, help centers, and conversational apps.

## Architecture

![Composer Architecture](/assets/ai-cx-automation.f731f9cce68bcde98db31f0ba904140a34a147a62cfb0bae739a52674e8ed13b.afe294b5.png)

### Data Sources

The platform ingests content from three primary sources:

- **Conversations**: User interactions, call and chat transcripts
- **Content**: Documents, articles, and media files
- **Data**: Structured datasets and databases


These sources feed into a centralized **Content/Data Lake** that serves as the foundation for all downstream services.

## Core Platform Components

### Content/Data Lake

- **Purpose**: Centralized repository for all ingested content and data
- **Features**:
  - Scalable storage for multi-format content
  - Data preprocessing and normalization
  - Version control and content lifecycle management
- **Integration**: Feeds processed content to retrieval engine


### Retrieval Engine

- **Purpose**: Intelligent content discovery and matching engine
- **Capabilities**:
  - Vector-based semantic search
  - Hybrid search combining keywords and embeddings
  - Real-time indexing and updates
  - Relevance scoring and ranking
- **Output**: Provides relevant content chunks to downstream APIs


### Agentic Apps

- **Purpose**: AI-powered autonomous applications that can take actions
- **Features**:
  - Goal-oriented task execution
  - Multi-step reasoning and planning
  - Integration with external tools and services
  - Context-aware decision making


## API Layer

### Ingestion APIs

**Key Features**:

- **Bulk Content Import**: Import large volumes of content from legacy knowledge management system
- **Content Validation**: Pre-import validation to ensure content quality and compliance
- **Job Management**: Track and manage import operations with detailed status monitoring
- **Scheduling**: Schedule imports for off-peak hours to minimize system impact
- **Real-time Monitoring**: Monitor import progress and access detailed logs


### Knowledge APIs

**Key Features**:

- **Access Controlled View**: Role-based permissions and content filtering
- **Personalized**: User-specific content recommendations and customization
- **Event-driven**: Real-time updates and notifications


### Search APIs

**Key Features**:

- **Content-aware Retrieval**: Understanding of document structure and context
- **Chunks or Articles**: Flexible granularity for search results
- **Relevance Rank**: Advanced scoring algorithms


### Answer APIs

**Key Features**:

- **Consumable**: Ready-to-use formatted responses
- **Best Answer**: Quality-ranked response selection
- **Source(s) Attribution**: Transparent citation and provenance


### Agentic APIs

**Key Features**:

- **Conversational Turns**: Multi-turn dialogue management
- **Global Context**: Persistent memory across sessions
- **Guided Actions**: Structured workflows and decision trees


## Experience Design Layer

### Web Portals

**Target Users**: Agents, Customers, and Employees

**Features**:

- **Web Page**: Responsive web interface
- **Personalized**: Driven by user permissions, scopes, and preferences
- **Configurable Content/Look**: Themeable UI and layout options


### Conversational Apps

**Target Users**: End users seeking interactive AI assistance

**Features**:

- **Widget/Chatpage**: Embeddable chat interfaces
- **Personalized**: Context-aware conversations
- **Configurable Experience**: Customizable chat flows and responses