Korea’s Hits Raises $13 Million to Build an AI Co-Scientist for Drug Discovery

South Korean drug-discovery startup Hits raised about $13 million in a Series B round. South Korean drug-discovery startup Hits has raised about $13 million in a Series B round. The company develops AI tools for identifying and designing drug candidates and is building its HyperLab platform around a collection of specialized capabilities.

An AI co-scientist needs a laboratory

Hits says its technology stack includes the K-Fold model and plans for an AI Co-Scientist that can combine models for protein structure prediction, toxicity analysis and molecular design. The direction is broader than using AI for a single prediction because drug research requires several different types of evidence before a candidate can move forward.

Drug discovery requires repeated cycles of hypothesis generation, computational analysis, experimental testing and revision, making workflow coordination a significant challenge.

Hits’ product information describes its AI co-scientist approach and the planned connection to laboratory automation.

Hits is developing AI software for drug discovery, where researchers have to combine biological data, molecular structures, experiments and scientific literature. The company’s HyperLab platform and K-Fold model are part of that effort.

• Hits says its technology stack includes the K-Fold model and plans for an AI Co-Scientist that can combine models for protein structure prediction, toxicity analysis and molecular design.
• Hits plans to launch an AI Co-Scientist that combines specialized models for protein structure prediction, toxicity analysis and molecular design.

Building a discovery workflow around multiple models

The main limitation is scientific validation. A model can prioritize promising candidates, but experiments remain necessary to determine whether a molecule behaves as predicted and whether it can ultimately become a viable drug.

The company also plans to connect the software with automated laboratory equipment. That connection is important because computational predictions have to be tested experimentally. A model can reduce the number of candidates researchers examine, but the laboratory remains the place where those hypotheses are challenged.

The new funding is therefore about building an integrated research workflow as much as improving individual models. The value of such a system will be measured by whether it helps researchers move through discovery steps faster while keeping the scientific evidence behind each decision visible.