Tesorai

Tesorai develops a cloud-native, AI-based platform for biological discovery, focused on mass spectrometry proteomics. Its two main products are Tesorai Search, which processes raw mass spectrometry data using large pretrained deep learning models to identify peptides, proteins and small molecules, and Tesorai Chat, a large-language-model interface that lets researchers query datasets, run analyses such as differential analysis and quality assessment, and generate figures without writing code. Users upload raw spectra and FASTA files through a graphical interface and launch processing jobs with minimal configuration; results are returned through interactive dashboards and the chat interface.

The platform supports Thermo Fisher, Bruker and Sciex instrument files in both DDA and DIA acquisition modes, a set of digestion enzymes including trypsin and chymotrypsin, variable modifications such as oxidation and phosphorylation, and TMT probes from 6plex through 18plex. The company positions the platform for teams spanning discovery through clinical stages, and cites capabilities including batch correction, outlier detection and normalization guidance, enterprise-grade security and a policy of not sharing customer data.

Tesorai also maintains a research output stream, including a preprint describing the development and performance of Tesorai Search and a preprint on cross-modal training of single-cell foundation models with proteomics data, plus collaborations with academic groups at UCSF and NYU.

Founding story

The founding team previously worked at Verily, an Alphabet company operating at the intersection of healthcare, biology and data science, where they applied AI to complex problems in medicine and science. There they observed how difficult it was for scientists in academia, biotech and pharma to extract meaningful insight from growing volumes of experimental data, and started Tesorai on the belief that the next generation of life sciences breakthroughs would come from better understanding of data rather than only from better experiments.

Business model

B2B SaaS, according to an investor profile classifying Tesorai in the biotech and life sciences sector with a B2B SaaS model. The website lists a pricing page and offers both self-serve trial access ("Try it Out") and demos arranged through the sales team.

Traction

Academic case studies report use of Tesorai's AI-enabled peptide identification by NYU researchers to identify 28,446 tumor-specific antigens in a pan-cancer immunopeptidomics atlas and by UCSF researchers to map protein-verified RNA editing sites, described as an 18-fold increase over previously reported evidence. Testimonials from UCSF and NYU users cite improved sequence variant detection accuracy and reduction of processing time for thousands of files from days to hours. A UCSF collaborator deposited a bioRxiv preprint on RNA editing in the cephalopod proteome using Tesorai.

Latest developments

The company highlights a recent preprint from its research team reporting that fine-tuning a single-cell foundation model on proteomics data matches or outperforms scaling to models more than 40 times larger, indicating multimodal training can beat parameter scaling alone.