I spent 7 years in a chemistry lab, then built an AI platform that controls real instruments. AMA.

Quick background:

I'm a materials science PhD candidate at East China Normal University & Shanghai Academy of AI for Science. For 7 years, my world was the lab bench — synthesizing materials, running spectroscopy, optimizing reaction parameters. Then I taught myself to code, and everything changed.

The moment that started Xyzen:

We tried to use an LLM to run a real chemistry experiment. It read every relevant paper. Wrote a perfect protocol. Designed ideal parameters. Then we pointed it at the actual equipment and… it couldn't turn on the stirrer.

That gap — between what AI can think and what AI can do in the physical world — became an obsession. Every major AI agent platform (Dify, CrewAI, LangGraph) operates purely in the digital realm. They're brilliant at code, data, text. But 15 million researchers worldwide still have to do every physical step themselves.

What we built:

Xyzen is an open-source (Apache 2.0) platform where AI agents can:

• Control real lab instruments through a Hardware Abstraction Layer (HAL) — any brand, any protocol. Say "stir the solution at 500 RPM for 5 minutes" → safe, monitored hardware execution.

• Collaborate as autonomous teams — proactive multi-agent system with DAG scheduling. You give a vague research goal; agents plan, execute, self-correct, and deliver.

• Trade as knowledge assets — every trained agent becomes a versioned, ownable, tradable digital asset in our Agent Marketplace. We call it Agent-as-Asset: your AI isn't a disposable tool — it's the inheritable embodiment of your expertise.

Where we are (March 2026):

• Deployed at 4 top research institutions: Peking University, Chinese Academy of Sciences, Shanghai Jiao Tong University, Xiamen University

• 1,500+ subscribers

• Model-agnostic: works with any LLM

• Launching open-source kernel public beta this half

Why I'm on Product Hunt:

I believe the maker community here understands something most VCs don't — that the best products come from people who deeply feel the problem. I spent years being the frustrated researcher. Now I'm building the solution.

Our vision: One person, one Self-Driving Lab, limitless science.

I'm here to learn, get honest feedback, and connect with people who care about pushing AI beyond the screen.

Ask me anything — about the tech, the science, the messy reality of getting AI to not break expensive lab equipment, or why a chemistry PhD decided to become a founder.

→ xyzen.ai