On December 1, 2025, Generate:Biomedicines announced it would launch two global Phase III trials, named SOLAIRIA-1 and SOLAIRIA-2, for GB-0895, a long-acting monoclonal antibody for severe asthma whose molecular structure was engineered using the company’s generative AI platform. The trials are expected to enroll roughly 1,600 adults and adolescents across more than 40 countries in North America, Europe, Latin America and Asia Pacific, and the drug’s path from initial design to a pivotal trial took about four years, a timeline the company is holding up as evidence that AI-guided protein design can compete with, and potentially beat, conventional biologics development.
A Disease That Still Escapes Current Treatment
Severe asthma remains difficult to control for a meaningful share of patients even with the biologic drugs already on the market, which typically target single inflammatory pathways and often require injections every two to four weeks. Laurie Lee, Generate’s chief medical officer, has been direct about the unmet need the company is targeting, noting that severe asthma remains hard to manage for many people despite years of new biologic launches. GB-0895 is designed to target thymic stromal lymphopoietin, a cytokine known as TSLP that sits near the top of the airway inflammation cascade, meaning it can trigger a broader downstream immune response than more narrowly targeted antibodies.
How the AI Actually Shaped the Molecule
Generate’s platform pairs machine-learning models with high-throughput lab experimentation to design proteins with specific target properties, rather than starting from a naturally occurring antibody and modifying it incrementally. For GB-0895, the company says that process was used to engineer ultra-high-affinity binding to TSLP alongside an extended half-life and high target specificity, properties that combined to support a dosing schedule of just one 300 milligram subcutaneous injection every six months. That interval, if it holds up in Phase III, would be markedly less frequent than most currently marketed severe asthma biologics, which patients typically inject every two to four weeks for years at a time.
What Earlier Data Showed
The Phase III launch follows a Phase I study in 96 patients with mild-to-moderate asthma, where GB-0895 was reported to be well tolerated across a dose range spanning 10 to 1,200 milligrams. The company reported an elimination half-life of roughly 89 days, consistent with the extended dosing interval it is now testing, and said biomarker reductions tied to airway inflammation were sustained for at least six months after a single dose, the data point that underpins the twice-yearly dosing strategy now moving into the pivotal trials.
What the Phase III Trials Are Actually Testing
Both SOLAIRIA-1 and SOLAIRIA-2 will track reductions in clinically significant asthma exacerbations over 52 weeks as their primary objective, the standard bar regulators use to judge whether a severe asthma biologic meaningfully changes a patient’s disease course rather than just shifting biomarkers on a lab report. Generate CEO Mike Nally has framed the program as proof of concept for the company’s broader approach, saying the advance “demonstrates the potential of programmable biology to design optimal molecular solutions for patients,” a framing that positions GB-0895 less as a one-off success and more as validation of Generate’s underlying design platform for future drug candidates.
The Industry Backdrop: Zero Approvals, Rising Stakes
GB-0895 is entering Phase III at a moment when the entire AI drug design field is under scrutiny for having produced plenty of promising early-stage candidates but, as of mid-2026, not a single fully approved medicine. A peer-reviewed count presented at the American Society of Clinical Oncology’s 2026 meeting tallied 117 AI-enabled therapeutic assets across 63 companies that have entered human trials, and industry estimates suggest somewhere between 15 and 20 of those programs could reach pivotal Phase III testing within the year. That makes 2026 a genuine inflection point: the field has demonstrated it can design molecules quickly, but Phase III is where efficacy and safety have to hold up in large, diverse patient populations over a full year, a test that has derailed plenty of conventionally discovered drugs long before AI entered the picture.
What Would Have to Go Right
For GB-0895 to succeed, it needs to show a meaningful drop in exacerbations that holds steady even as the twice-yearly dosing interval is put through its most rigorous test yet, in a patient population far more diverse in age, ethnicity and disease severity than the 96-patient Phase I cohort. A positive readout would give Generate a commercially compelling product, since a biologic dosed only twice a year addresses one of the most common patient complaints about existing severe asthma treatments, injection frequency and treatment burden. It would also hand the broader AI drug-design industry a rare, concrete data point proving that a molecule engineered primarily by machine-learning models, rather than modified from an existing antibody, can clear the same bar regulators apply to any other biologic. Results are not expected until well after the 52-week treatment period concludes, meaning patients and investors alike are looking at 2027 at the earliest for a real answer.
