• Open-source adoption: 6K+ GitHub stars, an international contributor community, and an actively released npm package.
• Known production use: temodar-agent runs OMA in a WordPress security-analysis platform; Mark Galyan runs it fully offline with local quantized models.
• Public downstream projects: PR-Copilot and StuFlow build on OMA's orchestration core.
• Third-party ecosystem: integrations with Engram for shared agent memory and AgentSonar for multi-agent observability.
• Commercial track record: former WPS Office commercialization product manager, with more than RMB 40M in cumulative revenue impact.

Focused advice for teams deciding where agents create real product value and how to ship them responsibly: opportunity assessment, architecture, build-vs-buy decisions, model and tool selection, cost, reliability, and implementation roadmap.

Scoped pilots and custom implementation for embedding agent capabilities into existing products, including multi-agent orchestration, MCP and tool integration, structured outputs, model routing, self-hosted or local-model deployment, evaluation, and observability.

Architecture support, private deployment, production hardening, custom integrations, sponsored adapters, and ecosystem partnerships around Open Multi-Agent.

Best fit: teams with an existing product or workflow, a concrete business problem, and the ability to start with a focused pilot.

Start a conversation →

Open Multi-Agent is a TypeScript-native framework for dynamic multi-agent workflows. A coordinator turns a goal into an inspectable task DAG at runtime, parallelizes independent work, and synthesizes the result.

• Dynamic goal-to-DAG planning instead of a hand-wired graph
• Model- and cloud-neutral, with cloud, OpenAI-compatible, and local models
• Lightweight, in-process, and designed to embed into existing Node.js backends
• Known production use, public downstream projects, and third-party ecosystem integrations

Website · Documentation · GitHub · npm

• Founder, YuanASI — building and delivering AI-native products and workflows.
• Creator and maintainer of Open Multi-Agent — leading product direction, API and architecture decisions, documentation, release strategy, and an international open-source community.
• Former commercialization product manager at WPS Office — product work with more than RMB 40M in cumulative revenue impact.
• Product-led delivery experience — from customer research and product strategy to architecture decisions, scoped implementation, launch, and iteration.

I work at the intersection of product judgment, commercialization, agent-system design, and open-source ecosystem building.

Open Multi-Agent is being developed toward an open, model- and cloud-neutral agent runtime that companies can embed and run in environments they control.

I welcome conversations with investors and strategic partners who share a long-term interest in sustainable open source, enterprise agent infrastructure, and model-neutral AI systems.

Investment or partnership: jack@yuanasi.com

我帮助已有产品或业务流程的企业团队，把 Agent 能力真正嵌入产品并完成落地：

• 企业付费咨询：Agent 产品机会、技术架构、成本与可靠性评估、落地路线图
• PoC 与定制开发：多 Agent 编排、MCP 与工具集成、多模型、自托管、本地模型、评估与可观测性
• Open Multi-Agent 企业合作：私有部署、生产加固、定制集成、生态与厂商合作
• 投资与战略合作：欢迎认同开放、模型中立 Agent 基础设施方向的投资人与合作伙伴交流

商务联系：jack@yuanasi.com · 预约咨询 · X · 小红书