"AI agents autonomously discover drug candidates with minimal human intervention." On January 6, U.S. biotech company Variant Bio announced the launch of 'Inference,' its drug discovery platform utilizing autonomous AI agents. The Inference platform employs Agentic AI (autonomous AI agent) technology to analyze proprietary and public human genomic data collected from global research, along with large-scale biological datasets, to identify drug candidates. Through this platform, Variant Bio has signed a multi-year research collaboration and licensing agreement with German pharmaceutical company Boehringer Ingelheim, securing a deal worth $120 million (approximately ₩170 billion), including upfront payments and milestone payments. How does Agentic AI differ from traditional machine learning? While conventional AI operates in a simple flow of data input → analysis → result output, Agentic AI possesses autonomy: once a goal is set, it independently formulates plans, explores data, evaluates intermediate results, and modifies strategies as needed. Variant Bio's Inference platform automates drug target discovery, analysis of genetic variant-disease associations, and drug response prediction, even designing the next experimental steps without human intervention. This innovation dramatically reduces time and cost in the early stages of drug development. Variant Bio has already signed collaboration agreements totaling $200 million (approximately ₩290 billion) with several pharmaceutical companies. The company aims to provide the Inference platform to research partners, thereby alleviating bottlenecks in the early stages of drug development. Last year, Eli Lilly also launched its AI-based platform 'TuneLab,' offering biotech companies AI models trained on years of its research data. Bristol Myers Squibb and Takeda have also announced AI-based drug discovery collaborations by sharing proprietary data. 2026 is expected to be a year when pharmaceutical companies actively open up AI platforms and data, fostering a collaborative ecosystem.