Main Conference Day 3Friday, 29 May 2026 - European Time GMT+1

We present Germinal, a model for performing epitope-targeted design of de novo nanobodies and scFvs against a wide range of epitopes, including the ability to design specificity or breadth across multiple targets.

Generative AI is redefining what’s possible in biologics discovery — enabling scientists to explore protein sequence space more efficiently, design higher-performing molecules, and accelerate the path from concept to candidate. At Cradle, we are developing intuitive software tools that allow protein engineers to directly leverage generative AI models within their existing workflows. This presentation will showcase how Cradle’s technology enables scalable and reproducible protein design, combining advanced AI architectures with user-centered design to reduce cycle times and ultimately create better biologics.

We ran an open protein design competition in where 100+ participants from industry and academia submitted binders against Nipah virus. Testing 1000 sequences experimentally in our lab revealed which computational methods translate to experimental success, with hit rates varying 10-fold between approaches. We'll share lessons on benchmarking design pipelines, the value of negative data, and how open competitions can accelerate the field when paired with standardized experimental validation.

Ability Biotherapeutics presents AbiLeap™ – a proprietary AI-enabled platform to systematically generate conditionally active antibodies. Trained on one of the largest therapeutic antibody databases, AbiLeap achieves unprecedented mutational reach (up to 25 amino acid changes) while retaining epitope specificity and therapeutic-ready humanness. Using AbiLeap we have generated leads against 4 targets and are pursuing IND-enabling studies by 2027.

Systemic toxicities from target expression on healthy tissues limit cancer therapy. The PrimeBody platform employs protease-cleavable linkers and affinity-tuned masking to generate tumor-activated biologics with exceptional systemic stability and efficient tumor activation. VOR-101, a masked, Fc-enhanced CD47 blocker, remains inert in circulation but rapidly activates in tumors, achieving >700-fold selectivity. It delivers ~100-fold higher exposure than conventional agents without toxicity, driving durable tumor regressions and a markedly improved therapeutic index.

T cell engagers (TCEs) show promise in solid tumors but are constrained by cytokine release and on-target, off-tumor toxicity. Amberstone’s T-MATE™ platform overcomes these barriers with a pH-gated, target-selective mechanism that adapts to tumor pH heterogeneity while maintaining an optimal therapeutic profile. This conditional strategy enables a new generation of TCEs that combines both safety and efficacy with broad applicability across diverse solid tumor indications.

PRO-XTEN™ masked T cell engagers (TCEs) are conditionally activated immune therapeutics designed to exploit dysregulated protease activity within the tumor microenvironment. This approach enables tumor-selective TCE activation while minimizing systemic toxicity. The XTEN mask serves a dual function: masking TCE activity and extending the half-life of the masked molecule, while the proteolytically activated TCE exhibits a short half-life, creating conditional pharmacokinetics that enhance tumor exposure while limiting systemic active drug circulation. Preclinical studies demonstrated: (1) robust masking with ~10,000-fold reduction in cytotoxicity in vitro; (2) potent anti-tumor efficacy in vivo at doses comparable to unmasked TCE controls; and (3) >100-fold improvement in maximum tolerated dose (MTD) in non-human primates. These findings support a significantly enhanced therapeutic window through tumor-specific activation and conditional PK. Clinical trials evaluating PRO-XTEN™ TCEs targeting HER2, PSMA, and EGFR are currently underway, with early data demonstrating the translational potential of this protease-activated platform.