Insider Brief
- Biostate AI launched K-Dense Beta, a multi-agent AI research system designed to compress scientific research cycles from years to days while eliminating generative AI hallucinations.
- In testing with Harvard Medical School, K-Dense built a transcriptomic aging clock from 600,000 profiles, revealing stage-specific gene predictors of aging and completing work in weeks that normally takes years.
- Built on Google Cloud’s Gemini 2.5 Pro, K-Dense outperformed frontier models on a leading bioinformatics benchmark and is being validated with academic, biotech, and pharmaceutical partners worldwide.
PRESS RELEASE — Biostate AI, a company accelerating biological research through innovative AI models, today announced the launch of K-Dense Beta, a comprehensive multi-agent AI research system that can compress research cycles from years to days, while eliminating hallucinations that plague generative AI models. In testing, K-Dense made a scientific breakthrough in longevity research, which will be published in a peer-reviewed journal this year.
K-Dense represents a shift from AI tools that handle isolated tasks to a system that conducts complete research cycles. The system coordinates specialized agents that plan experiments, review literature, design analyses, execute code in secure sandboxes, and generate publication-ready reports. The system eliminates hallucinations by operating like a team of independent scientific reviewers, with agents cross-checking references against external databases, adding feedback loops to improve accuracy, and building full traceability and auditability of every decision and action.
“There is a crisis in science right now, where we have too much data and not enough resources to evaluate it,” said Ashwin Gopinath, Co-founder and Chief Technology Officer of Biostate AI. “We have created an AI scientist that can work 24/7, dramatically accelerating discovery while maintaining rigorous scientific standards.”
With integrated access to resources ranging from standard bioinformatics pipelines, tools like Google’s AlphaFold, curated databases, and multiple small and large specialized Large Language Models (LLMs) like MedGemma, K-Dense can also modularly connect to any tool available through Model Context Protocol (MCP), a universal protocol that allows AI systems to access and coordinate external software.
K-Dense’s capabilities were validated in collaboration with Professor David Sinclair, Co-Director of the Paul F. Glenn Center for Biology of Aging Research at Harvard Medical School. Tasked with building a transcriptomic aging clock, K-Dense analyzed the ArchS4 dataset of more than 600,000 transcriptomic profiles, selecting 60,000 high-quality samples and strategically focusing on 5,000 genes from over 50,000 available. The analysis revealed that different sets of RNA transcripts become important predictors at different points in life. Genes useful in one stage were irrelevant in others, showing that aging is not a uniform process but a sequence of biological programs that each require their own predictive model.
“K-Dense enabled us to complete an entire research study in just a few weeks, work that typically requires months or years of expert analysis,” said Professor David Sinclair at Harvard Medical School. “It pointed us to markers and pathways that warrant deeper study and helped us build a unified AI model for predicting biological age. Importantly, it also provided a measure of how reliable those predictions are, which is critical for scientific applications and has not been available in prior AI approaches.”
The findings have been submitted for peer review and are available as a bioRxiv preprint. Biostate AI is now validating K-Dense with dozens of design partners, including academic institutions, biotechnology startups and major pharmaceutical companies.
“Biostate’s implementation with Gemini 2.5 Pro showcases our model’s transformative potential for complex scientific challenges,” said Bikram Singh Bedi, Vice President, Google Cloud Asia Pacific. “Their multi-agent approach demonstrates how intelligent coordination of advanced language models can accelerate genuine scientific discovery.”
Built on Google Cloud’s Gemini 2.5 Pro, K-Dense achieved 29.2 percent accuracy on BixBench,[3] the industry’s most rigorous bioinformatics benchmark, surpassing frontier models including GPT-5 (22.9 percent), GPT-4o (18 percent), and Claude 3.5 Sonnet (18 percent).
Since closing a $12 million dollar Series A led by Accel earlier this year, Biostate AI has rapidly expanded its capabilities. Prior investors include Dario Amodei (Anthropic), Emily Leproust (Twist Bioscience), and Mike Schnall-Levin (10x Genomics). In the months following the Series A, the company completed development of K-Dense and initiated clinical collaborations with MGH in the US as well as partners in China and India. This momentum marks the start of the next phase of growth.
K-Dense Beta is currently available to select design partners, with broader availability planned for later this year. Organizations interested in exploring how K-Dense can accelerate their research can request information at k-dense.ai.
