Generate:Biomedicines was founded on the premise that generative models of proteins, trained on scaled data, would suggest far better therapeutic molecules than those devised by traditional methods, while also significantly reducing timelines. With the positive results from our Phase 1 study of GB-0669, the first AI-generated viral neutralizing antibody targeting SARS-CoV‑2, we have compelling data that support this notion. Advancing GB-0669 from computer to clinic in just 17 months, we’ve demonstrated our platform’s capability to move beyond the limitations of traditional biologics drug discovery, successfully reaching previously undruggable targets with speed and precision.
From Scientific Hypothesis to Real-World Application
At the heart of GB-0669 was a novel hypothesis: Could we target the S2 domain of the spike protein, a critical region involved in viral fusion with human cells? This region had long been considered undruggable by traditional methods. The S2 domain is highly conserved across 99% of SARS-CoV‑2 variants, meaning it remains largely unchanged as the virus evolves. This stability made it an ideal target for the development of more durable therapeutics. However, traditional biologics drug discovery techniques had struggled to unlock the potential of this target, delivering only weak and poorly neutralizing binders unsuitable for clinical development. The virus has repeatedly mutated in response to immune pressure on other common binding sites, outsmarting the solutions that seemed to work only months ago. The S2 domain presented another point of attack—if only we could create a molecule capable of binding to it and neutralizing the virus with the level of potency needed for an effective therapeutic.
Our platform allowed us to rapidly explore vast molecular spaces, optimizing GB-0669 for potency, stability, and efficacy—all key factors in developing a viable therapeutic. The result was an antibody capable of neutralizing SARS-CoV‑2 while remaining effective across variants and related coronaviruses of pandemic threat potential, such as SARS-CoV‑1 and WIV1. In other words, using our generative AI platform we found a way to optimize a low potency molecule that could bind to a new part of the virus—one that, to date, hasn’t mutated like other parts of the spike protein.
The Role of AI in Redefining Drug Discovery
Generative AI is at the core of how GB-0669 was developed. By enabling us to explore molecular spaces efficiently, our platform provided insights that traditional methods were unable to achieve. Instead of relying on often slow, randomized screening processes, machine learning models helped guide our design choices from the start, allowing us to simultaneously co-optimize multiple properties that are critical to therapeutic development.
For GB-0669, this meant identifying the optimal design for an antibody that not only neutralized the virus effectively at a clinically viable dose but was also manufacturable at scale. This ability to balance potency, stability, and developability from the outset sets our platform apart from traditional antibody drug discovery, where each property is often optimized separately. This process, which could have taken years using traditional methods, allowed us to move GB-0669 from computer to clinic in just 17 months. This milestone is a clear demonstration of how our platform allows us to reach previously undruggable targets and respond quickly to emerging health or pandemic threats.
What the Data Tell Us
The Phase 1 study of GB-0669 was designed to evaluate its safety and tolerability, and project the potential for efficacy in healthy volunteers. Conducted across five dosing cohorts, the study demonstrated that GB-0669 was well-tolerated, with no dose-limiting toxicities. Importantly, significant viral neutralization was observed at the three highest doses, supporting the potential for long-lasting protection.
For patients at high risk of severe outcomes, the ability to receive up to six months of protection from a single dose is an incredibly promising development. The safety profile and neutralization data suggest that GB-0669 could be an important therapeutic option for vulnerable populations, including those with compromised immune systems or those at higher risk of COVID-19 infection.
What Comes Next for GB-0669?
The ability of GB-0669 to target both current and future variants make it a valuable candidate for addressing future viral threats and emerging pandemics. However, despite the promising results, we’ve made the decision to pause the development of GB-0669 as a prophylactic due to evolving market conditions and the high costs of conducting the very large-scale trials needed for such an indication.
The decision to pause its development as a prophylactic reflects the realities of the current commercial landscape. Opportunities for partnerships and further development in other areas where GB-0669 can have the most impact are currently being explored.
The Future of AI-Driven Drug Discovery at Generate
The success of GB-0669 demonstrates the potential of generative AI to accelerate drug discovery and reach previously undruggable targets, but GB-0669 is just one example of how AI can be used to solve some of the most complex challenges in biology. The ability of our platform to design, optimize, and develop therapeutics with speed and precision has the potential to reshape how we approach global health challenges.
We understand there is some skepticism about the emerging field of Generative Biology and AI-enabled molecular design. Like all drugs that enter the clinic, AI-designed drugs will not always succeed. However, the positive data from GB-0669 provides strong proof of concept for what AI can achieve. Our platform enabled the development of a drug that represents a new approach to a virus that has impacted the world, in record time, and by targeting a part of the virus that the scientific community had been unable to reach with any other molecule. This is a significant step forward and a reason for optimism about what generative AI can contribute.
At Generate, our mission is to harness the power of data and machine learning to program biology, and GB-0669 is an early indication of what our platform can achieve. We are excited to continue exploring how our platform can be used to create novel solutions that are faster, better, and more effective for patients who need them most.