Tsinghua's DevAll Rebuilds ChatDev Into a Zero-Code Multi-Agent Platform
ChatDev 2.0 (DevAll) turns multi-agent AI orchestration into a zero-code visual experience, backed by NeurIPS-accepted research from Tsinghua University's OpenBMB lab.
Takeaways
- ChatDev 2.0 (DevAll) launched: OpenBMB's complete rewrite turns the platform into a zero-code, general-purpose multi-agent orchestration tool. GitHub
- Visual drag-and-drop designer: Build agent pipelines on a canvas with no Python required; workflows are stored as YAML files for version control.
- MacNet DAG architecture: Directed Acyclic Graph model supports 1,000+ agents per workflow without context window overflow, preventing infinite loops.
- Iterative Experience Refinement (IER): Agents accumulate memory across runs and improve collaboration strategies over time -- unique among open-source frameworks.
- Five ready-to-run template domains: Data visualization, 3D generation (Blender), game dev, deep research, and teaching video generation included out of the box.
- Fully free, Apache 2.0: Python SDK on PyPI (
chatdev), Docker Compose support, and any OpenAI-compatible LLM works per-agent.
ChatDev, the multi-agent platform from OpenBMB at Tsinghua University, just made a major pivot. Version 2.0, branded as DevAll, transforms what was once a specialized "virtual software company" simulator into a general-purpose, zero-code platform for orchestrating AI agent teams across any domain. With over 32,000 GitHub stars and a paper accepted to ACL 2024, this is one of the most influential open-source multi-agent projects in existence, and its 2.0 release is a genuine architectural rethink.
From Software Company to Everything Company
ChatDev 1.0 operated as a virtual software company, with agents playing roles like CEO, CTO, and Programmer to automate the entire software development lifecycle, from design to testing and documentation. That was a compelling proof of concept, but it was also a rigid one. You could build software with it, and not much else.
In January 2026, OpenBMB released ChatDev 2.0 (DevAll), a complete rewrite that transforms the platform into a general-purpose, zero-code multi-agent orchestration tool. The name "DevAll" reflects the expanded scope: develop everything, not just software. ChatDev 2.0 is a conceptual generalization of ChatDev 1.0, and users seeking flexible multi-agent orchestration are encouraged to migrate to this version.
The Core Idea: Visual Agent Graphs
The central problem ChatDev 2.0 solves is accessibility. Every other serious multi-agent framework, whether LangGraph, CrewAI, or AutoGen, requires you to write Python. ChatDev 2.0 empowers users to rapidly build and execute customized multi-agent systems through simple configuration, with no coding required. Users can define agents, workflows, and tasks to orchestrate complex scenarios such as data visualization, 3D generation, and deep research.
The visual console, built with Vue 3, provides a drag-and-drop canvas where you design agent networks. Each agent is a node, interactions are edges, and you configure collaboration patterns visually. Under the hood, every workflow is stored as a YAML file, so it stays version-controllable even when you never touch the raw config.