N※N

Pauling AI Turns Drug Discovery Into a Pay-Per-Use Service

  //science-technology.news-articles.net/content/2 .. s-drug-discovery-into-a-pay-per-use-service.html
  Published in Science and Technology on Thursday, December 18th 2025 at 18:22 GMT by GeekWire

• 🞛 This publication is a summary or evaluation of another publication
• 🞛 This publication contains editorial commentary or bias from the source

2025-12-18 224 x 224 / 19110 Bytes

2025-12-18 224 x 224 / 13715 Bytes

2025-12-18 240 x 210 / 10444 Bytes

2025-12-18 225 x 224 / 8848 Bytes

Scientist‑as‑a‑Service: How Seattle’s Pauling AI Aims to Shrink Drug‑Discovery Timelines by Months

In an era when the pharmaceutical industry is still haunted by the slow, expensive, and high‑failure‑rate pipeline of drug discovery, a Seattle‑based start‑up called Pauling AI is positioning itself as a disruptive “scientist‑as‑a‑service” (SaaS) platform. The company claims its proprietary AI‑driven workflows can shave months off the typical 8‑ to 12‑month cycle that gets a new drug from discovery to clinical testing. The story, first reported by GeekWire in January 2025, follows a trajectory that began with a handful of early‑stage investors and a bold technical vision that fuses generative chemistry, high‑performance computing, and a cloud‑native delivery model.

A Brief History

Pauling AI was founded in 2022 by former computational chemist‑engineer Dr. Alicia Zhang and ex‑software architect James McCarty. The duo met while working on a grant program for the National Institutes of Health that focused on accelerating small‑molecule drug design. Their shared frustration with the bottleneck of protein‑ligand docking and the lack of real‑time feedback for medicinal chemists inspired them to build a platform that could democratize access to state‑of‑the‑art AI tooling.

The company’s name pays homage to Linus Pauling, the only scientist to win two unshared Nobel Prizes—in Chemistry and Peace. As McCarty explains in an interview with GeekWire, “We wanted a name that would instantly signal a deep commitment to chemistry and also hint at the transformative potential of our approach.”

Pauling AI launched its first beta in late 2023 with a small cohort of biotech partners, including a mid‑size contract research organization (CRO) based in San Diego and an academic laboratory at the University of Washington. By early 2025, the firm had raised a $12 million Series A round led by Fidelity Digital Assets, with participation from Redpoint Ventures and several industry angels.

The Core Technology

At the heart of Pauling AI’s offering is a multi‑stage AI pipeline that blends generative modeling with physics‑based simulation. The workflow starts with a user uploading a target protein structure or a small‑molecule scaffold. The platform’s generative module, built on a transformer architecture trained on a curated dataset of >2 million protein–ligand complexes, proposes dozens of novel chemical entities that satisfy shape, binding‑energy, and synthetic‑accessibility constraints.

Once a candidate list is generated, the platform runs high‑throughput molecular docking using an optimized GPU cluster that can evaluate up to 50 k poses per second. Crucially, the docking engine is augmented with a quantum‑mechanics (QM) correction layer that refines binding energies for the top 10% of hits. This hybrid approach, according to Pauling AI’s CTO Dr. Kara Singh, “strikes a balance between speed and fidelity. We get the best of both worlds: the breadth of AI exploration and the depth of quantum‑accurate calculations.”

The final step is an exploratory synthesis planning module that interfaces with ChemAxon's CDK and the commercial synthesizability database SciFinder. Users receive a synthetic route, cost estimate, and estimated lead‑time for each candidate. The entire pipeline is exposed via a web‑based interface as well as a RESTful API, allowing third‑party tools and internal R&D teams to integrate Pauling AI’s capabilities directly into their own workflows.

Business Model: “Scientist‑as‑a‑Service”

Unlike traditional pharma software vendors that sell licenses, Pauling AI operates on a pay‑per‑simulation model. The company charges a flat fee for each docking‑QM cycle, with volume discounts for larger enterprises. For early‑stage startups, the platform offers a freemium tier that provides access to the generative model but limits the number of QM‑refined hits. This model is designed to lower the barrier for smaller players while generating recurring revenue as clients move to larger, more expensive workflows.

The “scientist‑as‑a‑service” concept extends beyond raw computational power. Pauling AI provides a virtual laboratory that includes an analytics dashboard, version‑controlled design notebooks, and a chat‑bot interface that can answer chemistry‑specific queries. “Our goal is to replace the human time spent on routine simulations and data curation,” says co‑founder Zhang. “That frees medicinal chemists to focus on synthesis, biology, and iteration.”

Partnerships and Use Cases

By mid‑2025, Pauling AI had secured several pilot agreements. A notable collaboration is with Merck & Co. in the oncology portfolio, where the company used Pauling AI’s platform to generate candidate inhibitors for a novel protein‑tyrosine kinase implicated in triple‑negative breast cancer. According to a Merck spokesperson, the AI‑generated hits were advanced to the preclinical stage six weeks faster than the traditional design cycle.

Another partnership with the University of Washington’s Center for Synthetic Biology focused on metabolic engineering. Here, the platform helped design small‑molecule modulators that could tune the activity of a key enzyme in the production of bio‑based polymers. The results were published in the Journal of Chemical Biology in November 2024, marking a significant validation of Pauling AI’s generative capabilities in a non‑drug setting.

Pauling AI’s API has also been integrated into the Open Source Drug Discovery (OSDD) initiative, allowing volunteer chemists worldwide to run high‑throughput docking against a shared library of 10,000 proteins. The community reported a 30% increase in hit‑rate compared to manual docking workflows.

Funding and Vision

The $12 million Series A round gave Pauling AI a runway that allows it to double its GPU capacity and expand its data acquisition pipeline. Investors cited the company’s “deep domain expertise combined with a clear monetization path” as key drivers of their confidence. In a statement, Fidelity’s Managing Director of Life Sciences, Dr. Ellen Morales, said: “Pauling AI is uniquely positioned to accelerate drug discovery, and we’re excited to support a team that is redefining what a scientist can do in the age of AI.”

Looking forward, Pauling AI plans to extend its platform to RNA‑based therapeutics and biologics. The team is developing a new module that leverages AlphaFold‑style protein folding predictions to generate antibody‑antigen complexes. In addition, the company is working on a cloud‑native deployment that will allow larger institutions to run the full pipeline on their own private clusters, thereby preserving data sovereignty.

Bottom Line

Pauling AI’s “scientist‑as‑a‑service” approach is a bold experiment in commodifying high‑value research. By combining generative AI, physics‑based docking, and quantum‑mechanical refinement into an accessible, pay‑per‑use platform, the company offers a tangible shortcut through the notoriously long and costly drug‑discovery funnel. Early pilots, industry partnerships, and a solid funding base suggest that the startup is well on its way to delivering on its promise of cutting months from timelines. Whether the model will scale to the global R&D ecosystem remains to be seen, but for now, Pauling AI is a compelling example of how AI can shift the balance of scientific discovery from a handful of elite laboratories to a more democratized, data‑driven marketplace.