BUSINESS
We are a clinical-stage generative biology company pioneering the AI revolution in biotechnology and drug design and development. Our vision is to program biology to generate optimal therapeutics for the greatest impact on human health. Central to our vision is the Generate Platform, designed to be a therapeutic area and protein modality agnostic system integrating computational innovation with scalable biohardware to address therapeutic challenges beyond the reach of traditional technologies. We have built our Generate Platform to be a tight and fully-integrated loop (design–build–test–learn) to create proprietary, therapeutically relevant data and differentiated molecular solutions for the biological challenges we aim to address. In addressing these challenges, the Generate Platform can engineer solutions against therapeutic targets by either starting from existing reference proteins or suggesting completely novel ones without a reference starting point, also known as de novo design. The Generate Platform’s therapeutic potential has been demonstrated by our successfully progressing three computationally engineered proteins into human clinical testing, the most advanced of which is GB-0895, an investigational long-acting anti-thymic stromal lymphopoietin (“TSLP”) monoclonal antibody, which is enrolling patients in pivotal Phase 3 clinical trials for severe asthma. The first patient was dosed in one of the Phase 3 clinical trials on January 26, 2026. We also expect to advance two additional computationally generated oncology product candidates into Phase 1 clinical trials in 2026.
Biology is an information science. DNA encodes biological function through the way its sequence determines the structure and activity of the molecules it produces, which, in principle, makes biology programmable. In practice, however, the immense complexity of biology has made programming it very difficult. Historically, drug discovery has emphasized two general approaches to manage this complexity. One approach was an intentional, mechanically guided design approach at low-throughput. The other was a high-throughput experimental exploration approach that was generally less able to encode specific intent. We believe that dramatic reductions in the cost of compute and the cost of making and measuring DNA and proteins enable a new paradigm: intentionality at scale. In this paradigm, our generative models learn generalizable design principles from data to generate hypotheses at scale, and scalable experimental systems verify those hypotheses. The Generate Platform was built to implement this paradigm, generating large numbers of specific molecular and biological hypotheses in response to pre-specified therapeutic objectives and rapidly testing them. We believe intentionality at scale is foundational to achieving programmable biology: enabling systematic generation of medicines across therapeutic areas and protein modalities while producing proprietary data that improves our generative models over time.
The Generate Platform integrates generative and predictive models that learn design principles from proprietary data—e.g., diffusion-based models (such as our Chroma model) and graph neural networks, among other architectures—with advanced experimental biohardware systems for scalable verification. Our biohardware systems include scalable DNA assembly, rapid protein production, and high-throughput, multiplexed assay miniaturization enabling us to measure up to billions of molecules per generation cycle, as well as a cryogenic electron microscopy (“Cryo-EM”) core for high-content structural data generation, which has produced more than 500 high-resolution maps in 2025 alone. These capabilities significantly reduce the cost and time per assay data point, tightening the loop between generative models and real-world biological verification.
We refer to the distinct biological and molecular capabilities of the Generate Platform as “modular capabilities” or “modules.” Our modules are designed to be deployed individually or in combination to engineer differentiated therapeutic candidates. We have successfully translated these modular capabilities to create programs and product candidates with therapeutic potential. For example, our lead product candidate, GB-0895, utilizes our binding affinity and developability optimization modules, and is currently enrolling patients in Phase 3 clinical trials for severe asthma and is also being evaluated in a Phase 1b clinical trial for COPD. We used binding affinity and developability optimization modules, as well as additional modules, including functional optimization, to engineer our other product candidates, including GB-4362 and GB-5267. Investigational New Drug applications (“INDs”) were cleared by the FDA for both GB-4362 and GB-5267 in December 2025, and we expect to dose the first patient for both programs in 2026.
Our lead product candidate, GB-0895, is an investigational long-acting anti-TSLP monoclonal antibody in development for severe asthma that is intended to be dosed every six months (“Q26W”). Severe asthma represents a substantial unmet medical need, with industry sources suggesting only 15% to 25% of eligible patients receive biologic therapy. There are adherence and persistence challenges with existing shorter-acting biologic agents and GB-0895’s potential Q26W dosing regimen is designed to reduce injection frequency to address these challenges. We have engineered GB-0895 to have ultra-high binding affinity, reaching an estimated twenty-fold improvement over tezepelumab (106 femtomolar binding affinity) and a YTE amino acid modification, a clinically-validated half-life extension technology. A YTE amino acid modification is a specific change made to three amino acids (M252Y/S254T/T256E) in an antibody’s fragment crystallizable ("Fc") region. Preclinical and Phase 1 clinical data have demonstrated favorable safety results, long half-life (mean terminal half-life of approximately 98 days), and suppression of key biomarkers, such as blood eosinophils ("EOS"), fractional exhaled nitric oxide ("FeNO"), IL-5, and IL-13, supporting its potential Q26W dosing regimen. We are currently enrolling patients in two parallel global Phase 3 clinical trials for GB-0895 initiated in December 2025 (SOLAIRIA-1 and SOLAIRIA-2) for severe asthma, with full enrollment expected by the first half of 2028. The first patient was dosed in our SOLAIRIA-1 Phase 3 clinical trial on January 26, 2026. We intend to develop GB-0895 as a biologic-device combination product.
In parallel, we are currently conducting a Phase 1b clinical trial for moderate-to-severe chronic obstructive pulmonary disease (“COPD”) with expected data in 2026. COPD is a widespread and often fatal lung condition. Current biologics target patients with higher eosinophil counts, leaving the majority of patients without an approved biologic option. The Phase 1b COPD trial for GB-0895 is evaluating safety, tolerability, pharmacokinetics (“PK”), pharmacodynamics (“PD”) and immunogenicity. Preliminary data showed biomarker reductions and a PK profile consistent with our earlier Phase 1 trial for GB-0895 for the treatment of mild-to-moderate asthma, supporting an extended dosing interval in COPD. We plan to evaluate multiple approaches to determine the optimal development path for GB-0895 for the treatment of COPD, taking into account expected clinical timelines, regulatory feedback, costs and the clinical data from our Phase 1b trial.
In addition to progressing GB-0895, we are advancing additional programs and product candidates that leverage the Generate Platform’s modular capabilities. These include GB-4362, a systemically administered monoclonal antibody designed to neutralize free monomethyl auristatin E ("MMAE") as an adjunctive therapy to antibody-drug conjugate (“ADC”) molecules with an MMAE payload, as well as GB-5267, an armored, MUC16-directed CAR-T cell therapy candidate developed in collaboration with Roswell Park Comprehensive Cancer Center ("Roswell Park"), for solid tumors, initially targeting platinum-resistant ovarian cancer. Beyond these product candidates, we are advancing additional preclinical programs, including a next-generation ADC that is being developed as an internal program, along with other early stage preclinical programs. In addition, the Generate Platform’s modular capabilities underpin the confidential programs being developed in collaboration with Amgen Inc. ("Amgen") and Novartis Pharma AG ("Novartis").
We continue to advance our Generate Platform by investing in computational and biohardware innovation to scale productivity, unlock new modular capabilities and create differentiated future programs and product candidates. This should allow us to address important unmet patient needs with validated modules that offer scalable impact, at low marginal cost. Additionally, we intend to opportunistically utilize partnerships and collaborations to leverage our capabilities with sophisticated collaboration partners to work collectively to solve significant challenges, while improving the capabilities of the Generate Platform. In addition, we also intend to opportunistically explore various collaboration solutions to maximize value, secure capital, and direct our technology towards diverse therapeutic applications.
We are led by an experienced team of executives with backgrounds in leading pharmaceutical and life sciences companies and academia and deep experience in generative biology and computational sciences, supported by a distinguished board of directors. We were founded in 2018 by Flagship Pioneering, bringing together advancements in generative biology and computational protein science.
The Generate Platform and Our Modular Capabilities
Since inception, the Generate Platform was designed to create differentiated protein therapeutics and unlock the promise of a new method of designing drugs through a concept we call programmable biology. For biology to be programmable, it means that we must be able to design, write and execute biological functions with pre-specified intent, across therapeutic areas and protein modalities.
Our Generate Platform is designed to implement intentionality at scale by coupling AI models that generate large numbers of design hypotheses with scalable biohardware that verifies them. Each time we engineer, build and then test a set of hypotheses, a process which we refer to as a generation cycle, we generate experimental data that is intended to improve the Generate Platform. We package certain of these learnings into reusable modules—validated capabilities that can be applied across targets and modalities towards differentiated therapeutics.
