CataX AI, a four-month-old AI-native biotech, closed its seed round with exclusive backing from Qiming Venture Partners to build AI agents that accelerate early-stage drug clinical trials. The company skips traditional target discovery and focuses on automating document flows, data processing, and cross-functional coordination that typically stall development timelines.
TL;DR
- CataX AI raises seed funding led by Qiming Venture Partners to deploy AI agents in early drug clinical trials, bypassing traditional target discovery.
- The startup targets preclinical and early PoC pipelines, aiming to cut clinical-phase inefficiencies by 10x through agent-led coordination.
- Founder Wang Zilong combines clinical oncology training with AI research at Microsoft Research Asia to rethink pharma R&D workflows.
What happened
CataX AI completed its seed financing with exclusive investment from Qiming Venture Partners, the firm said. The company was founded in July 2026 in Shanghai and does not perform target or molecule discovery, unlike most AI drug startups.
Instead, CataX AI embeds autonomous agents into early clinical development to automate literature review, protocol drafting, and cross-departmental coordination. One clinical lead oversees each pipeline, while agents handle routine tasks and surface key decisions for human review.
Wang Zilong, the founder, holds an 8-year clinical medicine degree from Fudan University, a doctorate in oncology, and prior roles as CTO and CMO at VoxelCloud. He was named to the Forbes 30 Under 30 list and later led medical AI research at Microsoft Research Asia, where his work garnered more than 20,000 GitHub stars.
Qiming Venture Partners sees the biggest bottleneck in pharma now lies in clinical development, which accounts for roughly 80% of cost and time, according to the firm's internal analysis. The VC incubated CataX AI with co-founder Jiang Zhemin, a veteran clinical operations executive from Tigermed and BeiGene.
Why it matters
For developers, CataX AI's model shows how agents can reshape regulated, multi-stakeholder workflows beyond chatbots or coding assistants. The startup's "Clinical-of-One" approach turns each pipeline into a modular unit coordinated by agents, reducing internal friction that inflates timelines and budgets.
For startup founders in AI-native biotech, the round validates a shift from molecule hunting to process automation in clinical development—a segment with clearer regulatory pathways and faster feedback loops than early discovery. It also highlights the premium on founders who straddle medicine and AI, a rare profile that investors are chasing.
For investors, the model lowers capital intensity by de-risking early PoC execution and enabling faster out-licensing of validated assets. Qiming's exclusive seed bet signals confidence that agent-driven clinical coordination will become a core competency for next-gen pharma companies.
Open questions include whether the agent-led model can scale beyond a handful of highly trained clinical leads, and how regulators will evaluate AI-generated trial documentation. The startup plans to encode trial decisions and outcomes into a shared system to reduce dependency on individual expertise over time.
Key facts
- CataX AI closed its seed round with exclusive investment from Qiming Venture Partners, the firm confirmed.
- The company was founded in July 2026 in Shanghai and targets preclinical and early clinical pipelines, not drug discovery.
- Wang Zilong, founder and CEO, holds an 8-year clinical medicine degree from Fudan University and a doctorate in oncology.
- Prior to CataX AI, Wang Zilong served as CTO and CMO at VoxelCloud and led a 155-center, 47,000-subject study that earned one of China's early NMPA Class III medical AI device registrations.
- He later led medical AI research at Microsoft Research Asia, where related projects earned more than 20,000 GitHub stars.
- Qiming Venture Partners cited internal analysis estimating clinical development accounts for roughly 80% of total drug R&D cost and time.
- CataX AI's "Clinical-of-One" model assigns one lead per pipeline, using agents to automate literature review, document drafting, and data processing while keeping final decisions human-led.
- The company plans to in-license or co-develop early pipelines and out-license assets with positive human proof of concept to partners for later-stage development.
Context
China's biotech sector has produced 44% of the world's preclinical new drug molecules in recent years, but faces bottlenecks in moving assets into clinical trials due to high costs and operational complexity, per industry reports.
AI agents are increasingly applied to regulated workflows beyond drug discovery, including medical writing, regulatory submissions, and site monitoring, as large language models mature and gain domain-specific fine-tuning.
The shift from molecule-centric to process-centric AI in pharma mirrors broader trends in enterprise software, where automation targets high-friction coordination tasks rather than core invention.
