METiS Therapeutics Launches NanoForge, the World’s First AI-Powered Nanodelivery Platform

China Finance and Economics News, September 16th: Today, METiS TechBio, the leading innovator in AI nanodelivery, officially launched NanoForge, the world's first self-developed AI nanodelivery platform.

"From biotechnology to tech bio, NanoForge is a new paradigm we have contributed to the industry over the past five years. In the vertical field of nanodelivery, we have proven that large model innovation combined with closed-loop iteration of dry and wet lab experiments can significantly enhance efficiency and lead to technological breakthroughs," said Dr. Lai Caida, co-founder and CEO of METiS TechBio. Rooted in independent research and innovation, METiS TechBio aims to become a new engine for innovative drug development in China.

In recent years, the nanodelivery system has evolved from a "bottleneck problem" in new drug development to a "key enabling technology." The world's first approved mRNA COVID-19 vaccine in 2020 demonstrated the precise delivery capability and safety of LNPs, enabling mRNA to achieve high-efficiency antigen expression in human cells, and rewriting the speed and form of vaccine development. Innovations in nanodelivery have made it possible to regulate the targeting, release rate, and immunogenicity of carriers, becoming a key factor driving cutting-edge breakthroughs in cancer mRNA therapies, genetic disease gene-editing drugs, and rare disease protein replacement therapies.

It is reported that NanoForge is the world's first artificial intelligence-driven nanodelivery platform. Based on quantum chemistry and molecular dynamics simulations, NanoForge integrates a proprietary high-throughput wet lab and screening platform, combining METiS TechBio's self-developed novel synthetic lipid language model and generative algorithms, an AI agent for nanomaterial screening and research, and currently the world’s largest ten-million-scale LNP lipid library. This enables a closed-loop process from molecular generation, property prediction, AI-guided iterative dry and wet experiments, lipid formulation design and optimization, to final dosage form determination. Through continuous learning and evolution of the platform, it constantly expands the data barrier in nanodelivery.

Based on NanoForge, METiS TechBio has developed three core solutions: AiLNP (AI Nucleic Acid Delivery System Design Platform), AiRNA (AI mRNA Sequence Design Platform), and AiTEM (AI Small Molecule Formulation Design Platform).

To date, METiS TechBio has more than 10 million lipid structures, 100,000 data points available for model training, and has achieved a breakthrough in targeted LNP delivery to eight identified organs or tissues in the human body, including the liver, lungs, immune organs, heart, muscles, tumors, central nervous system, and gastrointestinal tract. The company owns over 100 authorized and filed patent applications. METiS TechBio has successfully developed more than 10 pipeline projects, produced seven preclinical candidate drugs, and is advancing four clinical programs in parallel, with the fastest program reaching the pre-NDA stage.

NanoForge's Four Self-developed Breakthrough Technologies: Currently the World’s Largest Library of Ionizable Lipids for LNPs at Tens of Millions Scale. METiS' ionizable lipid library is the data universe of NanoForge, currently encompassing over 10 million lipid structures, with a significant number supported by wet-lab data on more than 20 critical characteristics. This enables robust model training and algorithm improvement. The scale and proprietary nature of this dataset empower METiS TechBio to break through existing intellectual property restrictions, systematically capture both positive and negative experimental outcomes, support fine-tuning of molecular generation and prediction models, and effectively expand organ-specific targeting capabilities.

Pioneering NanoMaterial Innovation: A Vertical Artificial Intelligence Foundation Model. NanoForge, based on a series of vertical AI foundational models as its underlying architecture, covers each key stage of the nanomaterial development cycle, including molecular generation, property prediction, experimental design, formulation optimization, and expert validation. It operates cross-functionally in parallel and achieves optimization and prediction of small-molecule nanomaterials, as well as 20 critical physicochemical and functional properties of lipids and lipid nanoparticles.

The First AI Agent for Nanodelivery in the Industry. METiS TechBio has independently developed ALAN, the first AI agent for nanodelivery in the industry, which serves as the platform intelligence interface for NanoForge. It executes user queries in natural language or structured formats and guides the discovery, evaluation, and optimization of lipid-based delivery systems. ALAN is capable of understanding tasks, breaking them down, and dynamically activating corresponding foundational models and datasets available in NanoForge, including direct connections to curated lipid libraries, internal screening data, patents, academic literature, and other external sources. Additionally, ALAN supports a reasoning layer, enabling it to integrate output data from multiple models, apply structured reasoning based on domain-specific rules, and deliver results with transparent logic and scientific traceability.

Self-developed patented integrated dry and wet laboratory infrastructure. Under the NanoForge platform, the AI-guided dry and wet laboratory system has achieved end-to-end closed-loop optimization from computer design to in vivo validation. It rapidly validates and refines AI model predictions through real-world biological experiments, significantly shortening development time and increasing the likelihood of successful translation into clinical candidate drugs via a design-learn-validate closed-loop process. For lipid optimization tasks, NanoForge employs mechanism simulation models, predictive analytics, and in vivo testing to evaluate candidate lipids based on targeted delivery characteristics and biodistribution goals, selecting high-performance lipids with the desired therapeutic index at an early stage. For formulation optimization, METiS TechBio's self-developed patented AI high-throughput formulation screening system, METiS DATALOTS, is the first to simultaneously test hundreds of formulation combinations, significantly accelerating the discovery of novel LNPs for specific target organs and administration routes.