01 - Overview

What is AgentEnsemble?

AgentEnsemble is an open-source Java 21 framework for orchestrating teams of AI agents that collaborate to accomplish complex tasks. It provides a clean, idiomatic Java API for defining agents with distinct roles, assigning them tasks, and running them together as an ensemble.

Built on LangChain4j for LLM integration, AgentEnsemble supports any LLM provider that LangChain4j supports (OpenAI, Anthropic, Ollama, Azure OpenAI, Amazon Bedrock, Google Vertex AI, and more).

Core Concepts

Concept	Description
Agent	An AI entity with a role, goal, background, and optional tools
Task	A unit of work assigned to an agent, with a description and expected output
Ensemble	A group of agents working together on a sequence of tasks
Tool	A capability an agent can invoke during task execution (e.g., search, calculate)
Workflow	How tasks are executed (sequential, hierarchical, parallel)
Metrics	Per-task and per-run token counts, latency, and cost estimates
Execution Trace	Complete hierarchical record of every LLM call, tool invocation, and prompt

Goals

• Clean, idiomatic Java API using builders, immutable value objects, and standard patterns
• LLM-agnostic via LangChain4j — users bring their preferred model provider
• Easy to extend with custom tools, workflows, and prompt strategies
• Production-ready with proper error handling, structured logging, input validation, and testability
• Minimal dependency footprint — only what is necessary

Non-Goals (Phase 1)

• GUI or web interface
• Built-in tool library (search, web scrape, etc.) — users bring their own tools
• Persistent memory across ensemble runs
• Agent-to-agent delegation
• Streaming output
• Async/parallel task execution

Inspiration

AgentEnsemble is inspired by the multi-agent orchestration pattern where agents with distinct roles collaborate on complex workflows. The core abstractions (Agent, Task, Ensemble, Workflow, Tool) are common patterns in the AI agent ecosystem.

License

MIT License. See LICENSE.