Intel / article

Practical Guide Open Source Agent Frameworks And Elevenagents

Source: elevenlabs.io
Author: ElevenLabs editorial sitemap
Date: Published Sep 23, 2026

Why it matters

Gives a three-step pattern (generate, extract final text, re-emit as OpenAI-style SSE chunks) for adding voice to an existing agent framework. chunks early cuts perceived latency without rewriting orchestration.

Key takeaways · AI-distilled

ElevenLabs' voice orchestration sends a generation request to the Custom endpoint when it detects a likely end of the user's turn, and it accepts both Chat Completions and Responses formats.
For LangGraph, stream_mode="values" is simpler but sends fuller state snapshots that add latency; ElevenLabs prefers stream_mode="messages" for voice because incremental chunks cut time-to-first-audio.
LangGraph's messages stream includes intermediate steps such as tool-call updates, so the proxy must filter them out and pass only user-facing text to text-to-speech.
In the guide's minimal LangGraph setup, the receives the full conversation history on every request to keep state; server-side persistence via checkpoints is possible but left out.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
streaming — Sending a model's response token by token as it is generated, so the reader sees text immediately instead of waiting for the whole answer.

Read the source elevenlabs.io