Edison Scientific, Inc

Last month, alongside @multiplylabs , we hosted an AI x Biotech Block Party, bringing together scientists, founders, pharma partners, and investors for a night of conversation, demos, food, and drinks. Thanks to everyone who joined us! Our CTO, @andrewwhite01, sat down with Multiply Labs CEO @FredParietti for a fireside chat moderated by @dr_alphalyrae . They explored the role of AI across drug discovery, development, and manufacturing and what it could mean for the future of biotech. The conversations continued well beyond the fireside chat, and we’re excited to keep them going. If you couldn’t make it this time, there will be more to come.

Writing an Investigational New Drug (IND) application is a major hurdle for drug developers. It can take a team of specialists three to four months, and mistakes can trigger protocol amendments or clinical holds. In June, we started working with Population Health Partners (PHP) to leverage specialized Kosmos Clinical Protocol Design (“CPD”) and Kosmos Regulatory Authoring (“RA”) products to support PHP’s process of IND preparation. Since then, specially trained Kosmos products produced a first acceptable draft of the Nonclinical Overview in 4.2 hours, against 100 hours of expert time, flagged an error that roughly fifteen human reviewers missed, and authored 70% of the text in a Pre-IND Briefing Meeting Request submitted to FDA. PHP is now bringing Kosmos products to a second program and expects to complete IND preparation in as little as 1 month, instead of the 4 month industry standard. PHP's experts can spend their time reviewing and deciding, not just drafting. The time recovered by Kosmos products means that drugs could make their way to patients faster.

Read the full case study on our blog: edisonscientific.com/news/how-popul…

Kosmos can now read @NaturePortfolio! Today we're announcing a subscription agreement with @SpringerNature that brings licensed content from specialist journals across the Nature Portfolio directly into Kosmos. When Kosmos generates a hypothesis or evaluates a target, it can now pull directly from Springer Nature journals, linked to the Version of Record so it's iterating on current, correct science. This is the first of several content partnerships we plan to announce, because better-sourced literature makes for a faster path from discovery to development.

GIF

Learn more: edisonscientific.com/news/bringing-…

We're honored to be named to @BostonGlobe's inaugural 'Next in Medicine' list, which recognizes companies using research and technology to shape the future of patient care. At Edison Scientific, that means building Kosmos, an AI scientist capable of compressing the months long, iterative work of R&D to one day. From discovery, to development, to the clinic, Kosmos is driving scientific advancement across domains.

Read more: sponsored.bostonglobe.com/studiob/next-i…

Join the Edison team at our AI Biotech Block Party, co-hosted with @multiplylabs! The evening will include a fireside chat with @FredParietti, CEO of Multiply Labs, and @andrewwhite01, CTO of Edison, on the future of AI across drug discovery, development, and manufacturing. Plus food, drinks, lab tours, and demos!

Space is limited. RSVP here: luma.com/sqrw0b7k

Welcome back, @AsimovPress! Edison is proud to sponsor the return of Asimov Press and support its thoughtful reporting on the research and technologies shaping the future of science. We're looking forward to what the team publishes next!

Welcome back, @AsimovPress Edison is proud to sponsor the return of Asimov Press and support its thoughtful reporting on the research and technologies shaping the future of science. We're looking forward to what the team publishes next!

Introducing BixBench3, our latest eval for language models in science and the first eval to measure the ability of models to produce the analyses underpinning entire biology papers from raw data. BixBench3 evaluates 13 frontier models on 20 paper-derived tasks, using datasets up to 241 GB. Each task mirrors how scientists often use agents today: setting a research objective and designing a methodological plan, then delegating implementation to the agent. The best model reproduced 48% of the requested artifacts on average. Agents are starting to approach research study-scale capabilities, but there is still plenty of work to do.

Read more on Edison Advances: advances.edisonscientific.com/research/bixbe…

Edison is officially in New York! We opened the doors to our new SoHo office this week, and we’re so excited to give our rapidly growing NYC team a place to call home. We’re hiring in New York, San Francisco, and beyond. Come join us!

Open roles at edisonscientific.com/careers

The biggest beneficial impact for humanity from AI will be medicine. Edison is building the tools to get new medicines to patients faster. Come join us.

This week at GTC, we announced our partnership with @nvidia. Edison is committed to build at the frontier of scientific reasoning. We partner with NVIDIA across the AI stack in training and benchmarking capabilities to accelerate scientific research. Our latest partnered release is on BixBench-Hypothesis, a new benchmark focused on analytical judgment under ambiguity, or the ability to pursue a hypothesis with an open-ended goal. See our release blog post in the comments. More from our CEO @SGRodriques on why this partnership matters:

Read more: edisonscientific.com/articles/accel…

Edison Analysis Agent can access publicly hosted RNA-seq datasets, perform transcriptomic analyses, and draw accurate biological conclusions. Here we use the Edison Analysis Agent to reanalyze a recently released dataset on Autism Spectrum Disorder (ASD). Identifying how risk-associated mutations mechanistically alter brain development and contribute to ASD remains a significant challenge. In their recent Nature paper, Gordon et al. tackled this problem using patient iPSC–derived cortical organoids and time-resolved bulk RNA-seq, revealing genotype-linked expression similarities of ASD forms and their strongest divergence from normal organoids at the earliest stages. In our new post, the Edison Analysis Agent autonomously reanalyzed the public dataset and reproduced both key results, recovering genotype-driven clustering of ASD forms and independently recapitulating their “early divergence, later convergence” trajectory across development.

See the link to the paper: nature.com/articles/s4158… And the analysis: edisonscientific.com/articles/repro…

Not ready for a full Kosmos run? Start with a mini-discovery in Edison Analysis. We asked Edison Analysis to explore the Genomics of Drug Sensitivity in Cancer (GDSC) public dataset. In a single run, it retrieved and explored the data for sensitivity patterns, formulated a BCL2-dependency hypothesis in hematological cancers, and tested the hypothesis using independent CCLE expression data.

Read more here: edisonscientific.com/articles/ediso…

Today we are announcing a major update to Edison Literature, our scientific deep research agent. It represents a significant advancement over our previous PaperQA2 algorithm: it enables deeper reasoning over 100s of scientific documents, can retrieve information from figures and tables, and is state-of-the art among other deep research systems on scientific tasks.

GIF

Read more here: edisonscientific.com/articles/ediso… Try our platform here: platform.edisonscientific.com

Real-time single molecule sensing with nanopores has the potential to transform several fields including precision medicine and diagnostics, drug discovery, protein characterization, and chemical manufacturing. In a recent pioneering work advancing the abilities of nanopores, Zhang et al. (2024) introduced a custom-engineered nanopore capable of trapping single amino acids and demonstrated that molecular volume is the primary determinant of ionic current blockade. Here, we show that the Edison Analysis Agent can autonomously reproduce this key biophysical finding using the authors’ published nanopore time-series recordings—recapitulating the quantitative relationship between amino acid volume and blockade amplitude. Starting from raw current traces, the agent identified open-pore baselines and blockade levels, computed amino acid–specific blockade current amplitudes, integrated canonical biophysical properties of amino acids from the literature, and performed correlation and regression analyses to independently identify molecular volume as the dominant predictor of current blockade as individual amino acids pass through the nanopore.