AI News

CataX AI Secures Exclusive Seed Round Investment from Qiming Venture Partners, Using Agents to Accelerate Early Clinical Development of New Drugs

36Kr · AIUpdated Nov 16 20:36

AI summary

CataX AI has completed a seed round financing, exclusively invested by Qiming Venture Partners. The company was founded in Shanghai in July 2026. Founder Wang Zilong previously served as a senior researcher at Microsoft Research Asia and CTO of Voxel Technology. It does not work on target and molecule discovery, but instead uses Agents to accelerate the early clinical development of new drugs, adopting a "Clinical-of-One" model, in which a single clinical development lead oversees the project and coordinates Agents and external experts.

Body · AI translation

Investment Circle has learned that CataX AI recently completed a seed round of financing, with Qiming Venture Partners investing exclusively. Unlike previous AI pharmaceutical companies focused on drug discovery, this AI-native Biotech, founded only 4 months ago, does not do target and molecule discovery, but instead uses Agents to accelerate the early clinical development of new drugs.

The founder behind it, Wang Zilong, is a post-90s entrepreneur. In his first year of high school, he earned a recommendation for admission by virtue of a first prize in the National Biology League, and during high school he accumulated four first prizes across three subjects: biology, physics, and informatics. Later he chose to study in the eight-year clinical medicine program at Fudan University, ultimately earning a doctorate in oncology. After that, he served as CTO and Chief Medical Officer of an AI medical technology company and was honored in Forbes 30U30; he also joined Microsoft Research Asia, becoming the only researcher there with a medical background focused on medical AI. After several cross-disciplinary moves, he has now personally entered the field to make drugs.

Speaking of this startup, Wang Zilong traced a line from his medical training to the Agent wave: "Medicine determines why we set out, while AI determines how we should do it." Along the way, what he has always wanted to shorten is the distance between technology and patients.

Born in 1991 in Qinhuangdao, Hebei, he began sweeping multiple first prizes in national leagues in biology, physics, informatics, and others from the start of high school, securing early recommendation eligibility. Although he was extremely fascinated by computers, he made a seemingly counterintuitive choice: entering Fudan University to pursue the eight-year clinical medicine program.

At Fudan University, Wang Zilong studied medicine while auditing computer science courses. He compared interdisciplinary learning to navigation: entering an unfamiliar field is like a dark, boundless sea, and the intersection zones between disciplines are "precious freshwater islands." Finding them takes time; "don't rush, be patient and you will always find them."

His clinical rotations and internship at Tencent showed him another possibility. A department running full tilt could perform more than a dozen surgeries a day; at Tencent, a team's product could reach about 100,000 people as soon as it launched. He began to think about how to bring technology into healthcare so that more patients could benefit.

English text is being prepared.

Just as the large-model wave swept in. In 2023, Wang Zilong chose to return to a research high ground, serving as a senior researcher at Microsoft Research Asia (MSRA), researching medical foundation models, autonomous agents, and human-machine collaboration; related research garnered more than 20,000 stars on GitHub.

The turning point came early this year, when a succession of Agent products brought a "leap of faith." Previously, Wang Zilong participated in a study on visually impaired programmers, observing how they used AI tools to program and review code. As the paper progressed, reality was rewritten at an even faster pace. Even when writing code himself, he almost no longer reviewed it line by line.

Today, early clinical development of new drugs still carries a strong industrial assembly-line flavor: advancing a pipeline requires more than a dozen functional departments—medical, regulatory, operations, data, biostatistics, and others—to coordinate with one another. Even revising a routine report is accompanied by cumbersome cross-departmental meetings and document circulation.

"If agents were deeply present from day one, what should such a biopharmaceutical company look like?" With this question, Wang Zilong resolutely left Microsoft and started his company.

"Wang Zilong has a strong entrepreneurial passion." Chen Kan, a partner at Qiming Venture Partners and co-head of the healthcare innovation sector, recalled his first impression: "There are many people in the market who understand AI, and many who come from a medical background, but someone who is doing the most cutting-edge medical AI research at Microsoft, has received systematic clinical medical training, and is also eager to get into the field and do something practical—that is very rare."

This spring, introduced by a Fudan University alumnus, Wang Zilong met Chen Kan.

At that time, Wang Zilong had only a rough framework in mind for "embedding Agents into drugmaking," and had not even finished writing a business plan. But the two talked for several hours about technological evolution and new drug R&D, and they saw eye to eye on many underlying judgments.

English text is being prepared.

Developing a new drug often takes ten years and costs billions of dollars. Among that, early drug discovery accounts for only about 20% of the cost and time, while the remaining 80% of the weight falls on the clinical development stage. Over the past few years, China's output of preclinical new drug molecules (PCC) has accounted for 44% of the global total, urgently needing a cost-effective channel to cross the life-or-death line of "proof of concept in humans (PoC)."

"If humanity stopped all new drug molecule discovery today, the stockpile accumulated in laboratories alone would be enough for the global clinical system to digest for a full ten years." In Chen Kan's view, reconstructing clinical trials with agents is the inevitable path to reducing costs. But for traditional pharmaceutical companies to penetrate layer upon layer of functional department barriers through AI is tantamount to overturning the entire old system. Rather than fighting massive inertia, it is better to start from zero.

Afterward, Qiming Venture Partners decided to jointly incubate this new company with Wang Zilong by serving as the sole investor in its seed round. Through Qiming Venture Partners' efforts and introductions, Jiang Zhemin, who had spent more than a decade in clinical operations at companies such as Tigermed and BeiGene, joined as co-founder and CDO, filling in the key missing piece for clinical execution.

In July 2026, CataX AI was founded in Shanghai. "Cata" comes from Catalyst, meaning to accelerate new drug R&D; the "X" logo is formed by the intersection of a green DNA double helix and a blue chip circuit, symbolizing the exploration of unknown diseases and also embodying the expectation of tenfold (10x) efficiency.

The company targets preclinical and early-stage clinical pipelines and plans to hold corresponding rights through licensing-in or co-development to advance early clinical proof of concept (PoC). For projects that obtain positive human evidence, it will then seek out-licensing, with partners capable of subsequent development and commercialization taking over.

Undertaking these pipelines is the "Clinical-of-One" model explored by the team. A single clinical development lead coordinates the project and coordinates Agents and external experts. Literature research, document drafting, and data processing are handled with Agent participation, while key decisions remain the responsibility of humans.

Admittedly, medical validation cannot defy cold objective laws: if a clinical follow-up requires one year, AI cannot shorten the time out of thin air. The efficiency of the so-called "10x Biotech" lies in precisely screening assets and reducing human internal friction, directing resources into more valuable clinical validation.

One person supporting N pipelines: a sample of China's future pharmaceutical companies emerges

"The pharmaceutical companies of the future are destined to evolve toward this organizational paradigm." Chen Kan regards CataX AI's exploration as pathfinding for the industry.

In his view, the reshaping of R&D and the underlying division of labor by Agents will ultimately penetrate the industry, but this cannot happen in a vacuum and requires the entire ecosystem to join in. A company's internal innovation must ultimately still connect with hospitals, partners, and the regulatory system.

This exploration coincides with a window of opportunity for China's innovative drugs to go global. Chen Kan observed that overseas pharmaceutical companies' attention to Chinese assets has shifted from price advantage to asset quality. Some promising early-stage projects still struggle to proceed due to insufficient funding. For newly established Biotechs, the opportunity to participate in global R&D lies precisely in these links that have yet to be connected.

Bringing the focus back to the present and to CataX AI itself, the most realistic challenge is how to expand this model beyond the initial few people.

"I have no way to replicate ten of myself." Wang Zilong admitted frankly that the leads coordinating pipelines at present need both deep clinical understanding and AI thinking. Finding such people is not easy. If every project relies on the experience of a few people, this model will be difficult to hand over to more teams.

Wang Zilong places the solution to breaking the deadlock on the accumulation of "organizational experience."

Public medical literature can be learned by large models, but much of the experience in clinical trial R&D is hidden in specific work trade-offs. Why an expert requires the protocol to be modified here, how the team decides based on new data whether to continue or stop investment—Wang Zilong hopes to retain the basis behind these real-world judgments.

In his vision, external experts' feedback, the team's decision trajectories, and subsequent results are all preserved in the same system for later projects to draw on. When one development effort ends, experience can continue to accumulate, and newly joined clinical leads can also have something to follow, allowing this model to gradually reduce its dependence on a small number of cross-disciplinary talents.

This focus on the transmission of knowledge and experience happens to echo Wang Zilong's thinking on "cross-disciplinary" work.

"We are drawing boundaries for ourselves." Wang Zilong talked about Liu Cixin's The Village Teacher: memories cannot be inherited; although knowledge can be transmitted, everyone still has to undergo lengthy learning to approach where predecessors arrived. He hopes Agents will become an "external brain," helping people access broader knowledge and extend exploration beyond existing specialties.

Of course, turning the blueprint into daily reality still requires many hard battles. Wang Zilong said the team is reorganizing the clinical development process and coordinating technical systems, risk control, and regulatory requirements. Agent performance in asset research also needs to be further validated in subsequent development.

"New drug R&D is a long marathon relay; the hardest part is never the start, but how to pass each baton smoothly and efficiently to the finish line." In Wang Zilong's view, the restructuring of technology and organization is merely returning time to science, so that drugs can fulfill hope one step earlier.

This article comes from the WeChat public account "Investment World" (ID: pedaily2012), author: Wang Lu, published by 36Kr with authorization.

Source

36Kr · AI