Xlue, a health AI startup spun out of Carnegie Mellon University, has raised $35 million in seed funding led by Conviction, the firm founded by Sarah Guo. The company announced the round on November 18, 2026, alongside the public launch of its technology: a foundation model trained on patient health states and treatment actions rather than on text tokens. No valuation was disclosed.
The round positions Xlue among a small cluster of startups betting that large language model techniques can be repurposed to forecast how a person's health changes over time, rather than to generate text or images.
What Xlue builds and how it has performed so far
Xlue calls its core product a "human health world model." Instead of predicting the next word in a sentence, the system is trained to predict how a patient's clinical state evolves in response to treatment decisions, with the goal of forecasting disease risk and simulating clinical trial outcomes before they are run in the real world.
The company is already running prospective disease risk studies with health systems and working with pharmaceutical companies interested in simulating treatment responses. In an independent evaluation conducted in October by the firm Protege across 2,500 patients, Xlue's model flagged the most future disease cases on 14 of 16 risk prediction tasks, at a cost of roughly $0.04 per 1,000 predictions, according to figures cited in coverage of the launch.
Who is behind the company and the money
Xlue was founded in 2025 by Chenyan Xiong, a Carnegie Mellon professor who spent five years building language models at Microsoft Research, alongside Ophir Frieder and Hao-Ren Yao. Xiong serves as chief executive. The $35 million seed is the company's first disclosed institutional funding, bringing total raised to date to $35 million.
Where Xlue sits against rivals
Xlue enters a field where Oracle has been building clinical trial AI agents into its health data infrastructure, and Knit Health has developed a competing product called Med-BEAM aimed at similar disease-forecasting tasks. Xlue's pitch is narrower and more technical: rather than wrapping an existing large language model around clinical data, it trains a model explicitly on sequences of health states and interventions, a structure closer to the "world models" being developed for robotics and autonomous systems than to a conventional medical chatbot.
That framing matters for how investors are valuing health AI right now. Seed and early-stage health AI rounds have increasingly rewarded startups that can show a measurable prediction advantage over incumbents, rather than general-purpose AI wrappers.
What to watch next
Xlue has not disclosed the scope or outcomes of its prospective health-system studies, nor named its pharmaceutical partners. The company also has not said how it plans to generate revenue — whether through direct health system licensing, pharmaceutical trial simulation contracts, or both. Investors and competitors will be watching for published results from its ongoing studies and whether it can convert pilot deployments into paying enterprise contracts over the next year.
