Drug Discovery and Delivery Startups funded by Y Combinator (YC) 2025

We make oral drugs to treat and prevent viral infections - like COVID-19.

Invitris radically simplifies how we create novel biological drugs. Our patent-pending technology platform turns DNA into synthetic proteins and was recently featured in the Science magazine. It makes creating new drugs >10,000x more efficient at <100x lower costs of material. Our killer application is enabling the newest generation of synthetic bacteriophages to combat antibiotic-resistant infections, which are threatening to cause more deaths than all cancers combined.

Rosebud Biosciences accelerates drug development by screening drugs against organoids (micro-organs) that have the same gene mutations as the patients. We partner with therapeutics companies to screen their drugs, and we perform our own drug discovery for rare diseases that have no existing treatments. Our organoids are also fetal-like and enable discovery of novel drug targets for pediatric diseases. This technology was validated at Stanford, published in a prestigious journal, and has already led to the discovery of a drug target in a pediatric heart disease that could not have been found using traditional disease models.

Shasqi is leading the development of therapeutics leveraging click chemistry to get active cancer treatments to tumors with its proprietary CAPAC™ platform. Our technology is based on localizing a click chemistry reagent in the tumor area, which then activates a second agent, a systemically infused protodrug. When the two 'click', a powerful localized therapy is activated at the tumor, while keeping its systemic exposure levels below toxic thresholds.

Menten AI designs protein drugs and enzymes using quantum computing and machine learning. Their technology leapfrogs current computational methods making it possible to design previously intractable proteins. The team created the first protein design algorithm for current and near-term quantum computers and created the world’s first protein designed on a quantum computer. They are now working on 3 new proteins, including a peptide to prevent the spread of viruses like Covid-19. They have 2 paid pilot projects with multinational chemical companies and 4 LOI’s with pharma and biotech companies.

At CellChorus, we apply artificial intelligence to evaluate thousands of microscopy videos in parallel to evaluate how immune cells move, interact and perform over time. This allows our customers to discover, develop, manufacture and deliver cell therapies, antibodies, vaccines and other novel therapies faster, at lower expense, and with higher rates of success. Our customers include top-25 biopharma companies and seed-stage biotechs that are developing cell therapies, antibody therapeutics and vaccines (including one of the few companies with an approved CAR T cell therapy). Hear from the team at https://www.youtube.com/watch?v=IE9x_tm0XnI

Arpeggio Bio is a pioneering pharmaceutical company that develops drugs targeting transcription factors using AI and high-throughput RNA-sequencing. With $20M in venture funding, we've targeted "undruggable" proteins like NRF2, TEAD, and GPX4 where our lead program is rapidly progressing towards a DC for the treatment of IO-resistant melanoma. With partnerships with J&J and FORMA, we've validated our platform in rare disease and inflammation with a significant Phase I success.

We're making every therapeutically valuable emotion available on demand in pharmaceutical form. Our Osmanthus platform is an ensemble of AI models that synthesize the world’s largest datasets of psychoactive pharmacology and phenomenology, quantitatively pinpointing the combination of neurotransmitter receptor activity that underlies various emotional as well as cognitive and perceptual states. Our lead program MSD-001 is designed to be rapid-acting and produce heightened emotional and cognitive flexibility without hallucinations. MSD-001 is currently in human trials, and will be the base ingredient for multiple drug combination "emotions in a bottle" such as empathy, awe, clarity, or beauty.

Diffuse is building generative AI for protein design. Our mission is to build AI systems that engineer new and useful proteins with unprecedented control and accuracy. Our team has been behind breakthroughs in AI protein design for the past 7 years, including the first experimental validation of AI-generated proteins and diffusion models for protein structure and sequence.

Finding new cures requires seeing what and where the abnormal molecules are in a diseased tissue, but current tools see less than 1% of those molecules. Biocartesian combines microscopy and new chemistries to see 50X more, offering unprecedented insights into disease biology and new therapies.

Algen is using CRISPR to uncover disease-driving RNA messages to find treatments for cancer, inflammation and diseases with high unmet needs. We are harnessing the power of CRISPR and machine learning to find first-in-class drugs that modulate RNA messages at single-cell resolution. Algen's founders are bioengineer from Jennifer Doudna Lab and biotech commercialization expert, teamed up with experienced pharmaceutical executives.

We have built a platform that enables drug discovery and development for previously undruggable genome regulators ("the regulome"). We accomplish this by moving drug discovery out of artificial systems where regulome proteins fail to function properly, and back into live human cells. Our technology combines innovations in automated cell processing, next-gen proteomics, and advanced data analytics to measure a drug's effect on the entire regulome in one experiment.

Perlara 1.0 (Feb 2014 - Feb 2019) launched early R&D joint ventures called PerlQuests with highly motivated rare disease families, companies or impact investors. We screened gene-edited patient avatars for repurposable drugs in order to build a predictive R&D engine called ArkBase. We planned to monetize programs based on ArkBase predictions by selectively spinning out asset-centric clinical development NewCos. Perlara 1.5 (March 2019 - April 2020) was the transition to a one-person virtual clinical development company and when the epalrestat and aripiprazole repurposing trials were launched at Mayo Clinic for the treatment of PMM2-CDG and NGLY1-CDDG, respectively. Perlara 2.0 (May 2020 --> The first decentralized and distributed biotech

Nanograb is a computational drug discovery company that uses AI to generate the best combination of binders to treat different diseases. Our product allows drugs to be targeted to very specific areas of the body.

b-12 is an AI copilot that plans how to create and test new molecules for chemists. We're like a GPS for chemistry, helping pharma companies plan the recipes for new drugs, accelerating early-stage drug discovery from years to months. Andres built the first AI agent that can autonomously think and make molecules in a robotic lab, published in Nature Machine Intelligence (750+ citations), and won the best paper award at the NeurIPS AI for Science conference. Zlatko is a 3x National Chemistry Olympiad champion, competed at IChO level, and medicinal chemist who has worked at Roche and leading lab robotics companies. He has an extensive network of hundreds of chemists across Europe. We are publishing a groundbreaking paper in Nature demonstrating our AI's ability to optimally plan chemical recipes. We run paid pilots with big pharma companies showing that our platform planned a set of experiments that achieved full conversion in a single attempt, eliminating the usual 8-12 iteration cycle.

Parallel Bio created a human 'immune system in a dish' to discover drugs and immunotherapies more likely to work in patients. Our platform has all of the same elements as a human immune system, meaning you can test drugs and vaccines as if you were testing them in actual patients from the start. The biggest reason why 95% of new drugs fail is that they were tested in mice - but with our human platform, pharma companies will know which drugs will successfully treat patients, speeding up the process and reducing the cost. To date our platform has successfully been used to test 12 drugs, and vaccines against 8 different diseases. The company is founded by Robert DiFazio, a former Stanford R&D director with a PhD in immunology, and Juliana Hilliard, a former System1 Biosciences R&D lead with an MSc in bioengineering, both of whom have led drug discovery programs for years.

We're building Quinn at iollo, an AI scientist that analyzes multimodal biological data (multi-omics, imaging, etc.) to discover new biology in hours, work that typically takes teams of scientists years. Quinn translates these discoveries into accelerated development of novel therapeutic strategies and interventions for extending healthspan.

Intracellular delivery is the most problematic step in T cell immunotherapy discovery, development and manufacturing. Indee Labs is developing a non-viral intracellular delivery system based on microfluidic vortex shedding (µVS) to rapidly deliver nucleic acids, proteins and gene editing complexes to patient and donor T cells with minimal perturbation.

An AI-maximalist pharma platform. We use agents to buy drugs, run clinical trials, and sell for a profit. 10x faster and 20x leaner than incumbents. Sourcing Agent: Mines public/private databases and unstructured global data to surface overlooked preclinical candidates. Scientific Agent: Runs comp bio models (ESM-3, RFdiffusion, Boltz-2, AlphaFold) to assess safety and efficacy in silico. Commercial Agent: Analyzes FDA incentives, pricing dynamics, TAM, competitive landscape, and payer alignment. Clinical Agent: Runs digital twin simulations, evaluates CRO/CMC risk, and builds trial plans to Phase 1. BD Agent: Summarizes and delivers company-tailored pitches. Built by 2 Stanford CS students who have built 3 startups together before.

Technology finds information and delivers physical products to us almost instantly. mRNA does the same for biology. At HelixNano, we see a world where treatments are cheap, progress is decentralized, and you can modify biology at the push of a button. By focusing on mRNA improvements from first principles, we’re getting there faster than we thought. Experiments that used to take months or years can be done in weeks and cost thousands instead of millions. The rate limiting step for our progress is now scientific curiosity. Rigid hierarchies and top-down decisions will miss the magic. Some of our most powerful technologies started out as side projects of junior team members. At HelixNano, every individual scientist can have a massive impact on the company's direction. And the more diverse our team, the faster we’ll find the really cool ideas. We've started rolling out some of our new tech — moving a COVID-19 vaccine candidate for the immunocompromised towards the clinic, and testing an intervention with massive environmental impact. But we’re scientists at heart, and team focus is solidly on the horizon.

Macromoltek: Revolutionizing antibody design. Description: Macromoltek, a computational de novo drug design company, rapidly produces accurate and credible antibody designs. We have built a proprietary platform that enables design against difficult targets inaccessible by traditional methods and have are already designing antibodies for large biopharmas and smaller biotechs.

Gilgamesh is a pioneering, clinical stage mental health science-focused biotechnology company. We have a disciplined focus on developing innovative new chemical entities (NCEs) leveraging a unique combination of medicinal chemistry, intellectual property strategy, neuroscience & neurobiology, and drug development expertise. We are an experienced team, with proven success in biotech exits and inventions.

PostEra is building a modern 21st century biopharma. We're using our advances in machine learning to accelerate Medicinal Chemistry and bring more cures to patients. PostEra advances small molecule programs through partnerships with biopharma while also advancing its own internal pipeline. We've raised $26M from top investors and closed $1Bn in AI partnerships, signing multi-year agreements with Amgen, Pfizer and the NIH. PostEra also launched and led the world's largest open-science drug discovery effort; COVID Moonshot.

Numerion Labs is an AI-native company accelerating the discovery of life-saving medicines through the development and use of cutting-edge machine learning algorithms. The company unites computational chemistry, structural biology, and medicinal chemistry to pioneer the next generation of AI-driven drug discovery platforms.

Miracle is the control room for clinical trials. Our platform consolidates data from multiple trial systems into real-time dashboards, delivering automated insights that empower clinical-stage biopharmaceutical companies to streamline operations, reduce manual workflows, and accelerate data-driven decision-making. By transforming complex trial data into actionable insights, Miracle helps biopharma teams speed up clinical trials, and ultimately, improve patient outcomes.

Adventris makes cancer vaccines. Our platform solves the problem of targeting poorly immunogenic oncogenes. Our first "off-the-shelf" vaccine targets KRAS, the most common cancer oncogene. In the long run, we envision a world where every adult receives our pan-cancer vaccines annually – preventing the majority of cancer deaths. If you are interested in learning more about our approach to treat and prevent cancer, please reach out to us at contact@adventris.com

Angstrom AI builds GenAI-based molecular simulations to substitute wet lab experiments in the pre-clinical drug development pipeline. We are a team of 2 PhD's and 2 Professors from the University of Cambridge who decided to start a company together after we realised how to combine breakthoughs in our research in quantum-accurate models of physics and generative AI models. Our Biotech/Pharma clients can verify the efficacy and safety of new drug candidates using our computer simulations, which match the accuracy of wet lab experiments, but are over 100x faster. We achieve this accuracy by constraining our genAI-based simulations to obey the laws of physics, avoiding the hallucinations seen in other GenAI technologies. Since joining YC, Angstrom AI has developed the first physically accurate gen-AI based simulation of multiple molecules interacting. We have published the first molecule water solubility results with accuracy within the error range of wet lab experiments. We have also kicked-off a 150K pilot project with a pharma company to apply our tech to estimating solubility in their drug development pipeline.

At Pirouette, we have developed a low-profile, disk-shaped, rugged injection device with a patient-centric focus on affordability, portability, and usability. The patented device, based on extensive patient input, is designed to make administering an injection as easy as pushing a button. This allows individuals with limited or no training to easily and intuitively perform an injection on themselves or others.

Reticular helps pharma companies discover drugs with AI models like AlphaFold by making them steerable, just like you can prompt LLMs. Today, limited validation data means companies spend millions on failed experiments trying to steer these models through trial and error. We’re piloting our AI interpretability technology with early-stage biotechs and scaling rapidly. Just a week after our pivot, we identified the first interpretable features ever found in protein models, allowing precise control over biological functions. Nithin and John met competing in Biology Olympiads before spending 4 years as roommates at MIT publishing ML/bio research in NeurIPS and Nature. We believe biological models encode far more information than anyone is currently using - our goal is to unlock this potential.

We have developed an AI platform enabling to accelerate the discovery and the design of genomic therapies such as Cell Therapies, RNA Therapies and DNA Therapies.

It takes 5 years for pharma to develop long lasting drug injections for chronic diseases like cancer and diabetes. Persist uses AI-driven automation to reduce formulation development time down to 2 years, a ~50% reduction.

Proteins are the most advanced nanotechnology we know of. At Adaptyv, we’re building a fully automated protein foundry to allow you to synthesize and test any protein you design.

Matrubials is an anti-infectives company, focused on human milk peptides as therapeutics to address antimicrobial resistance and modern diseases. Currently the global rise in severe bacterial infections and the drying pipeline of effective treatments are burdensome for healthcare and economics. Sculpted by evolution, milk, first food for the newborn, provides an exemplary solution to address the nutritive and protective challenges an infant is faced with. The anti-pathogenic molecules from milk can be developed into therapeutics geared towards all-age diseases. Our first indication is bacterial vaginosis, a bacterial dysbiosis related inflammation that results in recurrent infections, disruption of homeostasis and long-term clinical complications in women of reproductive age, including but not limited to preterm births, fertility issues and increased rates of STDs. Our next inflection point is the evaluation of rapidly acting and selective candidate peptides in pre-clinical host impact and early safety studies, which can aid clinical and portfolio development towards topical applications.

Olio labs uses AI to develop combination therapeutics that consider the thousands of interacting proteins in your body rather than targeting just one or two. Their lead combinations target obesity and are more effective with fewer side effects than Ozempic, the fastest growing drug of all time. Their custom ML built from real-world expertise and cutting edge AI searches trillions of combinations to find the perfect one.

Verge uses artificial intelligence and human data to develop better drugs faster. Verge has an end-to-end technology-driven drug discovery and development platform, featuring one of the field’s largest and most comprehensive proprietary patient genomics datasets from human tissue. Verge uses machine learning to mine this data and develops these insights into new drugs using it’s human-centric biology and chemistry platforms.

SelectIQ addresses the primary bottleneck in drug development: recruiting patients into clinical trials. We analyze electronic health records to help doctors find clinical trials for their patients, and to help trial sites find patients for their studies. 72% of doctors want to inform their patients of clinical trials, but only 5% do. We're on a mission to close that gap, ultimately bringing safer, more effective drugs to the patients who need them.

Lavo Life Sciences runs simulations of drug molecules on computers. Pharma companies use these simulations to guide their experiments and ultimately save time and money in the lab. This will de-risk and expedite clinical trials and FDA approval.

90% of drug candidates fail, delaying help for those in need. We’re on a mission to change that. Better drug target discovery powered by the Newfoundland founder effect.

Our mission is to enable pharmaceutical and biotech companies to bring more antibody therapies to patients. Using AI and massively parallel experimentation, we design antibodies that precisely bind the disease target at the right location, while minimizing manufacturability and toxicity risks. We are a well-funded, revenue-generating, bilingual company of wet- and dry-lab scientists, and are founded by AI and protein design experts from Harvard University.

We are a drug discovery company developing a new generation of therapeutics that embrace the complexity of disease. Currently available approaches for creating drugs work on the principle of finding one drug-one protein 'magic bullets'. However, diseases such as cancer and autoimmunity are often the result of several dysregulated proteins across many distinct biological pathways. We are building a computational-experimental platform to design therapeutics that can target several disease-causing proteins at once. By designing multi-specific drugs, we are able to create therapeutics that are more efficacious and safer than existing medicines.

Abalone Bio is tackling challenging undrugged targets underlying diseases affecting millions, focusing first cell-specific antibody drugs to treat obesity and metabolic disease without the GI side effects that cause 25% of GLP-1 drug patients to quit after 1 year. + High throughput experimental measurement uniquely leverages AI/ML: We’ve engineered cells to measure antibodies for pharmacological activity, not just structure or binding like others, 100 million at a time, 100X+ the throughput of others. With our large, proprietary activity datasets, we uniquely leverage ML to both unlock the discovery of rare hits and generate optimized hits for challenging targets, starting with G-protein coupled receptors (GPCRs) + Proven success: We have developed antibody agonists (activators) for 2 out of the 8 G-protein coupled receptors ever drugged by biotech. + Pharma traction: We’ve secured 3 partnerships with $3M in revenue and $125M in downstream value. + Externally validated science: We’ve been awarded $7M in non-dilutive grant funding for platform and program development.

Sixfold is focused on the commercialisation of its patented drug delivery systems for RNA therapeutics such as siRNAs for gene silencing and mRNAs for gene expression.

zPREDICTA is developing organ-specific, physiologically-relevant 3D cell culture models providing a 1-to-1 reconstruction of human tissues. This is the only available 3D culture approach that comprehensively accounts for the major components of the cellular and extracellular tissue microenvironment preserving the critical interactions between a tumor and its surroundings. Clinical study data demonstrate the high correlation between drug efficacy measured in our platform and the clinical response. The ability of the zPREDICTA platform to reliably predict clinical outcomes offers superior compound attrition management, empowering researchers to exclude ineffective compounds while selecting agents with a high probability of success in the clinic. Our approach is compatible with any drug class, multiple tissue/cell types and a wide variety of readouts. Our customers are using these models for efficacy screening of anticancer compounds, including immuno-oncology agents, evaluation of mechanisms of drug resistance, the rescue of failed drug candidates, assessment of off-target toxicity, and a multitude other applications.

Coral integrates molecular, functional, and clinical factors to create new clinical laboratory tests that predict optimal treatments for autoimmune diseases. We have also identified a unique subpopulation of patients who do not respond to current therapies and are working towards addressing this unmet medical need.

Who we are? We write software that understands the activity of individual cells as opposed to all cells. This approach is novel because we are no longer looking at average signals and can be precise. It is like a police person trying to identify the speeding car, they won’t find it if they look at all the cars at once, rather they need to look at the speed of each individual car. This is the beauty of single cell genomics, we can now identify that speeding car. Why now? The biggest barrier for entry into the single cell genomics market are cost and data analysis. To that end, we have reduced the cost by 10-fold and built the computational infrastructure for scalability. Although single cell sequencing has only been around for about five years, we come from one of the top single cell genomics labs and were the first to process and analyze millions of immune cells. Thanks to our software solutions, we have been able to bring big data solutions to the immune system. Where are we and where are we going? We have generated a database of over 1 million immune cells spanning healthy, cancerous, and autoimmune immune cells and are growing through paid partnerships. Harnessing out database and ML software, we can take a data-centric approach for better target discovery. Fortunately, as our database continues to grow, our models will continue to learn for more precise, unbiased, and automated single cell genomics. Thank you!