Biostate AI, a molecular diagnostics startup combining next-generation RNA sequencing (RNAseq) with generative AI, announced today it has raised $12 million in a Series A funding round led by Accel. The round also saw participation from Gaingels, Mana Ventures, InfoEdge Ventures, and returning investors Matter Venture Partners, Vision Plus Capital, and Catapult Ventures. High-profile angels such as Anthropic CEO Dario Amodei, 10x Genomics CTO Mike Schnall-Levin, and Twist Bioscience CEO Emily Leproust also backed the company.
The new funding fuels Biostate’s ambitious goal: to make biology predictable and unlock precision medicine at scale. Much like how OpenAI trained ChatGPT on trillions of words to understand human language, Biostate is training foundation models on billions of RNA expression profiles to learn the “molecular language” of human disease.
A Netflix Model for Molecular Medicine
The startup, founded by MIT and Rice professors-turned-entrepreneurs Ashwin Gopinath and David Zhang, envisions a new paradigm for diagnostics. Rather than offering isolated sequencing services, Biostate uses a Netflix-inspired self-sustaining business model: the company processes thousands of RNA samples at ultra-low cost, feeds that data into a proprietary generative AI system, and improves its models with every experiment. The result is a virtuous cycle—affordable sequencing powers better models, and better models deliver deeper clinical insight.
“Every diagnostic I’ve built was about moving the answer closer to the patient,” said Zhang, CEO of Biostate AI. “Biostate takes the biggest leap yet by making the whole transcriptome affordable.”
The transcriptome—the complete set of RNA molecules in a cell—provides real-time snapshots of human health and disease. Yet historically, full-transcriptome sequencing has been prohibitively expensive and difficult to interpret. Biostate is addressing both problems with a dual approach: radical cost reduction and cutting-edge AI.
Technical Innovations: BIRT, PERD, and Generative AI
At the core of Biostate’s offering are two patented technologies: BIRT (Biostate Integrated RNAseq Technology) and PERD (Probabilistic Expression Reduction Deconvolution). BIRT is a multiplexing protocol that allows simultaneous RNA extraction and sequencing from multiple samples, reducing cost nearly tenfold. PERD, meanwhile, applies novel algorithms to filter out “batch effects”—variability introduced by differences in lab conditions or sample handling—which often obscures the biological signal in multi-site studies.
This highly standardized RNAseq pipeline feeds into Biostate’s proprietary foundation model, Biobase, which functions much like GPT models in natural language processing. Trained on hundreds of thousands of transcriptomic profiles across tissue types, disease states, and species, Biobase captures the “grammar of biology”—the underlying patterns of gene expression that define health and disease.
Just as GPT can be fine-tuned to write essays or summarize legal documents, Biobase can be adapted to detect early cancer recurrence, predict drug response in autoimmune disease, or stratify patients in cardiovascular trials. Biostate’s Prognosis AI, built on top of Biobase, already shows promise in forecasting leukemia relapse and is being piloted for multiple sclerosis with the Accelerated Cure Project.
“Just as ChatGPT transformed language understanding by learning from trillions of words, we’re learning the molecular language of human disease from billions of RNA expressions,” said Gopinath, the company’s CTO. “We’re doing for molecular medicine what large language models did for text—scaling the raw data so the algorithms can finally shine.”
Building the World’s Largest RNAseq Dataset
To date, Biostate has already sequenced over 10,000 samples for 150+ collaborators, including Cornell and other major institutions. Its goal is to scale that number to hundreds of thousands of samples annually. This exponential growth is made possible by its low-cost RNAseq process and streamlined data ingestion pipeline, OmicsWeb, which standardizes, labels, and securely stores transcriptomic data across jurisdictions.
The company’s cloud infrastructure includes several novel GenAI tools, such as:
-
OmicsWeb Copilot – A natural-language interface for analyzing RNAseq data without code.
-
QuantaQuill – An AI assistant that generates publication-ready scientific manuscripts, complete with figures and citations.
-
Embedding Surfer – A visualization tool that uncovers hidden biological relationships within gene expression data.
With offices in Houston, Palo Alto, Bangalore, and Shanghai, Biostate is expanding globally to support a growing network of clinical and academic partners. The startup is already processing both fresh and decades-old tissue samples—helping labs extract insights from previously unusable specimens.
Toward General-Purpose AI for All Diseases
Biostate’s endgame is bold: to create a general-purpose AI capable of understanding and guiding treatment across all human diseases. This unifying approach stands in contrast to today’s fragmented biotech landscape, where each condition often requires its own siloed diagnostic tool and therapeutic path.
“Rather than solve the diagnostics and therapeutics as separate, siloed problems for each disease, we believe that the modern and future AI can be general-purpose to understand and help cure every disease,” said Zhang.
By treating biology as a generative system—where today’s molecular state determines tomorrow’s outcomes—Biostate believes it can predict not just current health status, but future disease trajectories and optimal interventions.
What’s Next?
With more than $20 million raised to date and a rapidly growing client base, Biostate is accelerating clinical collaborations in oncology, cardiovascular disease, and immunology. The company’s next milestones include regulatory validation of its predictive models and commercial scaling of its AI-driven diagnostic tools.
As Gopinath puts it: “We’re not just interpreting biology. We’re building the biological equivalent of the Large Language Model—only this time, it’s trained on the human body.”
If Biostate AI succeeds, the next wave of precision medicine may not just be reactive—it will be predictive, personalized, and powered by generative AI.
