whitepaper

Agent Orchestration Patterns: Chains, Graphs, and Tool Routing

LLM ROUTER TOOL ATOOL B EVALOUTPUT MCP Chains → graphs → agents.

Thesis

Moving from linear chain-of-thought to graph-based agent orchestration unlocks parallel tool execution and conditional branching without sacrificing observability.

What's inside

  • Comparison of chain vs. graph orchestration patterns
  • Built-in tool-calling protocol with MCP support
  • Observability via LangSmith tracing across multi-step flows

Claims sourced from this document

tool calling protocol
Native MCP support plus function-calling for OpenAI/Anthropic/Google.
evidenced
model agnosticism
30+ model provider integrations without custom adapters.
evidenced
license obligations
MIT license, no deployment restrictions.
evidenced
eval harness maturity
LangSmith-based evaluation with built-in scorers and datasets.
evidenced
← Back to library