Engineering AI agents
you can ship and hand over

01Who we are

Shenzhen YuanASI Technology Co., Ltd. (brand name YuanASI) was founded in 2023 by Jack Chen and is based in Nanshan District, Shenzhen.

We work with companies that already have a concrete business scenario and need AI wired into existing systems, and we deliver primarily remotely.

02Founder

Jack Chen founded YuanASI and brings eight years of product experience across internet platforms and AI companies. He is the author of the open-source multi-agent orchestration framework open-multi-agent (6,800+ GitHub stars). He combines big-company product method and commercialization experience, a deep grasp of AI technology and hands-on engineering: he breaks business problems into solutions with clear acceptance criteria and a return you can measure, and he wrote the orchestration framework the delivery runs on. He leads YuanASI's project delivery.

• Tencent, SenseTime and others: product work at an internet platform and at AI companies, with a working command of requirements for large user bases and of turning AI technology into product.
• WPS: former commercialization product manager on a product with more than 500 million monthly active users; turns product capability into measurable business results.
• Another startup: product lead, accountable for a product's direction and delivery, and experienced in taking a product from planning to launch.
• open-multi-agent: author and maintainer; MIT-licensed and iterating steadily since v1.0.0 in April 2026, it carries his engineering practice in multi-agent orchestration, approval, tracing and evaluation.

03What we do

• Enterprise AI advisory: scenario assessment, technology selection, POC validation, and effort estimation.
• Custom AI agent development: from scenario definition through to production deployment, delivering systems that run and can be maintained.
• Multi-agent system integration: multi-agent architecture, enterprise knowledge bases, and integration with your existing stack.

Each service can be engaged on its own, or run as one path: assess → validate POC (optional) → build & integrate → deploy → hand over. You enter at whatever stage you're actually at.

04The foundation: open-multi-agent

Our delivery is built on open-multi-agent (OMA), our own open-source orchestration framework. It is TypeScript-native, MIT-licensed, and generates and executes task DAGs inside Node.js applications — with plan approval, execution receipts, Trace / Run Viewer, EvalSets, and task-level checkpoints.

Models, tools, credentials, and deployment environment stay entirely under your control. The orchestration logic is open source: you can audit every line of it, and keep evolving it after delivery.

05How we work

• Prove it's worth doing first: a POC plus a versioned EvalSet gives you an acceptance baseline and a go / no-go recommendation before you expand the investment.
• Put the execution boundary in the code: plan approval, tool permissions, timeouts, budgets, and task-level checkpoints are enforced as constraints in code.
• Delivery means handover: custom project source, tests, deployment scripts, runbooks, and team training ship together, and the system runs in your own environment.

06Milestones

07Company details