Roman Kasianov

Researchers at MIT’s Jameel Clinic and CSAIL, in collaboration with Recursion and leveraging NVIDIA’s BioHive-2 supercomputer, have released Boltz-2, an open-source biomolecular foundation model designed to simultaneously predict 3D protein structures and their binding affinities.

Boltz-2 extends beyond earlier innovations such as AlphaFold3 and its own predecessor Boltz-1 (released last year) by integrating affinity predictions with dynamic structural modeling—achieving near-physics-based (FEP) accuracy up to 1000 times faster than traditional computational methods.

FutureHouse has released ether0, a 24B-parameter open-weight language model specialized for chemistry tasks. Built on Mistral 24B Instruct and refined through reinforcement learning, the model is designed for molecular generation under natural language constraints. ether0 is now available under an Apache 2.0 license and has been integrated into FutureHouse’s experimental chemistry agent, Phoenix.

Kiin Bio, a 2024-founded startup developing an AI-native research platform for drug discovery, has raised $2.2 million in pre-seed funding led by b2venture. The company’s platform deploys “Virtual Scientists”—generative AI agents designed to plan, analyze, and orchestrate experiments across lab data, modeling tools, and analysis pipelines. Rather than replacing workflows, Kiin integrates with tools already used by R&D teams, aiming to increase output without requiring additional human hires.

Menlo Park–based biomanufacturing company Antheia, founded in 2015 by Dr. Christina Smolke and Dr. Kristy Hawkins, announced a $56 million Series C funding round led by Global Health Investment Corporation (GHIC) and EDBI. The financing will support commercialization of its first biosynthetic product, thebaine—a key input in naloxone (Narcan)—and help advance additional candidates from its 70+ compound pipeline spanning neurology, oncology, gastrointestinal, respiratory, hematology, endocrinology, and anti-infectives. Several programs—including oripavine, scopolamine, and multiple oncology candidates—are in pre-commercial development, with others in R&D and pilot stages.

Today, Nature Medicine published peer-reviewed Phase IIa clinical results for Rentosertib (ISM001-055), a TNIK inhibitor developed by Insilico Medicine using a generative-AI workflow. The paper details the GENESIS-IPF trial—a randomized, double-blind, placebo-controlled study of Rentosertib in idiopathic pulmonary fibrosis (IPF)—and the same data were presented at the American Thoracic Society 2025 meeting.

A study published in Nature Communications reports the development of ISM5939, an orally active small-molecule ENPP1 inhibitor designed using Insilico Medicine’s generative AI platform. 

The compound works by inhibiting ENPP1, an enzyme that suppresses immune activity in tumors, to reactivate the STING (Stimulator of Interferon Genes) pathway. According to the authors, this may help restore local immune responses in cancers that don’t respond to checkpoint inhibitors or chemotherapy.

CurifyLabs, a Finnish healthtech company, has raised €6.7 million to expand its automated solution for manufacturing compounded medications. The round was led by Springvest, a Finnish investment platform, and includes a €1 million deep tech loan from Business Finland. The funds will be used to scale the company’s operations and secure high-volume customers, including in the United States.

Biostate AI has raised $12 million in Series A funding to scale its RNA sequencing platform and train a general-purpose AI model capable of predicting disease and therapy response. The round was led by Accel with participation from Gaingels, Mana Ventures, InfoEdge Ventures, and others. It brings Biostate’s total funding to over $20 million.

Researchers from Harvard Medical School and Roche have developed COMPASS, a foundation AI model designed to predict patient response to immune checkpoint inhibitors (ICIs) across 33 cancer types and six ICI therapies. The model addresses a long-standing challenge in oncology of identifying which patients will benefit from immunotherapy by learning biological concepts directly from tumor transcriptomes and linking them to treatment outcomes.

Persist AI, a startup focused on accelerating pharmaceutical formulation development through AI-guided robotics, has launched its Cloud Lab platform and secured $12 million in Series A funding. The round was led by Spero Ventures, with participation from investors including Eli Lilly, Shimadzu Future Innovation Fund, MBX Capital, SignalFire, and others.