Vilya

Market position

Vilya operates in AI-enabled drug discovery, specifically the macrocycle modality positioned between small molecules and biologics. It is one of several protein-design companies spun out of David Baker's lab at the Institute for Protein Design in Seattle. Third-party sources identify Isomorphic Labs, Roivant Sciences and Arrowhead Pharmaceuticals as comparable companies. One profile notes Vilya was cited among top 2024 Seattle-area funding deals alongside other IPD-affiliated firms.

Vilya emphasizes de novo design over screening: rather than searching existing compound libraries, it computationally generates molecules for a specified target, including chemistries outside the canonical amino-acid set. Its stated differentiators are macrocycles that combine biologic-like selectivity with oral, membrane-permeable properties; proprietary all-atom models (Vilya-1, Vilya-2) that handle non-canonical chemistries where general protein-structure models reportedly fail; and the ability to conduct much of its screening in silico, reserving wet-lab work for the most promising candidates.

Technology

Vilya's platform performs generative design, structure prediction and in silico validation of macrocycles, including peptides containing non-canonical amino acids and exotic side chains that expand accessible chemical space beyond the 20 natural amino acids. Vilya-1, released in January 2026, is described as an all-atom neural network foundation model built from first principles for macrocycle structure prediction and design; the company reports it outperforms physics-, rule- and deep-learning-based methods at recovering crystal conformations on a combined public and internal benchmark, more than doubling the next best method, and that it can be fine-tuned to enrich for membrane-permeable compounds, rank its own sampled conformations, and generate conformational ensembles and energy landscapes to prioritize designs likely to adopt binding-competent conformations. Vilya-2, announced in July 2026, is described as an all-atom foundation model for modeling chemically diverse molecular interfaces. Earlier work included cyclic peptide structure prediction and design using AlphaFold (bioRxiv, 2023). The company cites the rigidity of cyclic versus linear peptides as improving metabolic stability and the predictability of structure and binding.

Go-to-market

The company markets primarily through scientific publication and technical reports (Science 2024, bioRxiv 2023, Vilya-1 and Vilya-2 technical reports), a partnering inquiry channel on its website, and recruiting across computation, chemistry, biology and engineering.

Sources describe Vilya as serving the pharmaceutical and healthcare industries; its therapeutic focus areas are oncology, immunology and gastroenterology, including inflammatory bowel disease, and it targets patients with diseases involving intracellular or otherwise undruggable protein targets.

Geography

Headquarters at 188 E. Blaine Street, Suite 124, Seattle, WA 98102, with an additional presence in South San Francisco. One third-party profile lists the headquarters as South San Francisco, CA, conflicting with the company's own site and other sources that state Seattle.

History

Following its August 2022 launch, Vilya appointed Cyrus Harmon as CEO in October 2023. In April 2024 it published work in Science describing the computational design of 15 million candidate macrocycles using non-canonical amino acids, with 18 synthesized and 15 closely matching their design models; around the same time it expanded its Series A, bringing total funding to $71 million. In August 2024 scientific founder David Baker was awarded the Nobel Prize in Chemistry, and in November 2024 C&EN featured Vilya among early-stage firms to watch. The company subsequently released two foundation models: Vilya-1 in January 2026 and Vilya-2 in July 2026.

Risks & controversies

Sources indicate all disclosed programs are preclinical, with no candidate reported past development-candidate selection, so clinical validation of the oral macrocycle approach remains unproven. Third-party aggregator data is internally inconsistent — one profile lists both funding rounds as 'Series A' with an employee band of 1,001-5,000 and a South San Francisco headquarters that conflicts with the company's stated Seattle base, and includes an apparent error referencing an award to a 'CEO Han' — so unverified aggregator figures should be treated cautiously.