We are creating an artificial intelligence (AI) platform that anyone can try out and verify its existence. Our dream is to build an AI drug development ecosystem that enhances competitiveness for everyone from pharmacy students to research institutions and pharmaceutical companies.
On the 22nd, CEO Kim Woo-yeon introduced the recently launched drug development AI, 'HyperLab 2.0,' at the headquarters of HITS in Gangnam, Seoul. Kim, a professor in the Department of Chemistry at the Korea Advanced Institute of Science and Technology (KAIST), founded HITS, an AI drug development startup, in May 2020. Kim is currently serving as an advisor to the Korea Pharmaceutical and Bio-Pharma Manufacturers Association on AI drug development.
HITS initially showcased HyperLab last year, and then unveiled version 2.0 at the Drug Discovery Chemistry (DDC) 2025 conference held in San Diego, USA, last week. HyperLab is a platform for designing drugs based on generative AI. It helps identify candidate substances in the early stages of drug development.
Unlike many domestic AI drug development corporations, HITS offers its AI platform as a paid subscription service that can be utilized by institutions, corporations, or individuals worldwide. Kim noted, "There aren’t many AI drug development companies in Korea and abroad that make their platform available to everyone. I believe that to promote our technology for quickly and accurately finding candidate substances, it is essential to show and allow people to experience its existence first."
The MD Anderson Cancer Center of the University of Texas, the world's leading cancer treatment specialty hospital, is a prominent subscriber of HyperLab. It operates hundreds of new drug research programs annually. After encountering HyperLab at last year's DDC, it subscribed directly for its research purposes.
The minimum concentration unit indicating drug efficacy is referred to as 'mole (M).' In the early stages of drug development, it is crucial to determine whether there is efficacy even at very low amounts, such as micromoles, which are one millionth of a mole, or nanomoles, which are one billionth of a mole.
The MD Anderson Cancer Center sought to find substances that respond to cancer cells even at extremely low quantities for the past five years but failed repeatedly. However, utilizing HITS's HyperLab, it designed a new candidate substance that is effective even at a concentration of 60 nanomoles, hundreds of times lower than previously. This means the problem that had remained unsolved for over five years was resolved in just six months.
However, the HyperLab 1.0 version was limited to virtually searching for new drug candidate substances and predicting outcomes on a computer. It simulated how well this substance could bind with body proteins and whether it could function stably in the body. The actual process of creating substances and testing their efficacy was out of reach.
HITS subsequently developed an upgraded version, 'HyperLab 2.0,' significantly expanding the functionalities and scope of the AI platform within a year and a half. It evolved beyond merely virtual predictions to a level where it can synthesize candidate substances into actual compounds and conduct experiments. Clients can now perform the entire drug development process, from discovering and designing candidate substances to verifying efficacy through actual synthesis, all in a one-stop manner using just HyperLab.
The evolution of HyperLab in a short period is largely attributed to the role of Symple Chem, a Swiss molecular synthesis company, with which collaborations began earlier this year. When HITS sends candidate substances designed by AI along with a synthesis quotation, Symple Chem synthesizes the substances.
Kim stated, "HyperLab 2.0 has implemented AI drug development, which previously took place only in digital or virtual laboratories, to also be realizable in real laboratories." The company is also reviewing plans for delivering synthesized substances directly to customers.
HITS significantly increased the pool of compounds that can be utilized as drug candidate substances. Previously, it was about 7 trillion compounds, but now it has grown to approximately 11 trillion, positioning it at the world's largest level. This expansion is thanks to the vast molecular library (databases) held by eMolecules, a U.S. company specialized in compound synthesis that has been collaborating since last year.
The fast and accurate design of HyperLab 2.0 is also credited to two key technologies developed in-house by HITS: 'Hyper Screening X' and 'Hyper Binding Co-folding.'
Hyper Screening X plays a primary role in filtering candidate substances that have a high likelihood of synthesis and expected efficacy among the 11 trillion compounds available. As a result, promising candidates can be identified within 48 hours.
Hyper Binding Co-folding automatically predicts which substances can bind well to a given protein by simply inputting the name of the target protein for the new drug. For instance, by inputting the name of a protein found in cancer cells, the AI calculates in advance which drug could attach to it and be effective.
While previously focused on small-molecule synthetic drugs, HyperLab 2.0 has broadened its application range to include complex modalities such as antibody-drug conjugates (ADCs) and targeted protein degraders (TPDs). Kim emphasized that it is particularly useful for recent developments in ADCs, which many domestic and international pharmaceutical companies have jumped into.
ADCs are therapeutic techniques that attach drugs to antibodies for targeted delivery exclusively to cancer cells. He remarked, "To develop ADC candidate substances, antibodies, payloads (anticancer agents), and linkers to connect the two are needed, and using HyperLab makes it possible to quickly design the payloads."
HITS is focusing on securing customers through HyperLab 2.0. At DDC 2025, the City of Hope, a cancer specialist hospital in the U.S., and a subsidiary of the global pharmaceutical company Roche are considering subscriptions. It is said that the recommendation from existing client, MD Anderson, has also played a significant role. A major distribution company developing functional foods in Korea is also a client. Currently, there are 30 customers utilizing HyperLab.
Kim mentioned, "Rather than being used by specific pharmaceutical companies, the aim of HITS is to provide an AI infrastructure that anyone can use, from pharmacy students to research institutions and pharmaceutical companies," and added, "We will continue to focus on enhancing HyperLab and will keep launching upgrade versions 3.0 and 4.0."