Home›Companies›Blank Bio
Blank Bio

RNA intelligence for precision medicine

Blank Bio is an applied AI research lab focused on increasing the success rates of clinical trials. We do this by training RNA foundation models that learn the patterns that shape disease progression and patient response to treatment. We aim to help pharma make more informed decisions in clinical trials by capturing the biology that makes each patient’s tumour unique. We’re a technical team of AI scientists and engineers from companies including Recursion, Deep Genomics, DeepMind, and Amazon, and institutions including Memorial Sloan Kettering Cancer Centre, Stanford, and the Vector Institute.
Active Founders
Jonny Hsu
Jonny Hsu
Founder/CEO
Working on foundation models for RNA therapeutics. I was an early employee at Valence Discovery, a biotech startup that spun out from Mila. I led product and strategy. Valence was acquired by Recursion (NASDAQ: RXRX) in May 2023. I stayed on afterwards to help build out the AI research engine of Recursion called Valence Labs. Biochemist by training | Waterloo graduate
Building AI for the most designable molecule in the world: RNA Published at top ML & Bio conferences including best paper at Foundation models for science @ Neurips. Built self-supervised learning applications for RNA, small molecules, and phenomics. Playing long term games in ML for Biology. One of the first employees at Deep Genomics. PhD from University of Toronto.
Ian Shi
Ian Shi
Founder/CTO
Building foundational models for RNA therapeutics at Blank Bio. Just defended my PhD in CS at the University of Toronto. Published at ICML, NeurIPS, Nature Comms; used 4 GPUs to pre-train an mRNA foundation model that beat competitors using months of compute on thousands of H100s. Now scaling our models to unlock new possibilities in the RNA therapeutic space. Previously at Amazon building software, interned at Pinterest and Amazon as a research scientist in between.
Company Launches
Blank Bio - Biological AI Models to Predict Patient Outcomes
See original launch post

TL;DR:

Today, we are excited to introduce Ladon, a state-of-the-art RNA foundation model that predicts disease trajectory and treatment response in patients. There is immense complexity in the individual biology of every patient. Two patients can have the same diagnosis and receive the same treatment, yet have very different outcomes: one responds well and lives for years, while the other’s disease continues to progress. Understanding and resolving this heterogeneity to predict outcomes is how we unlock clinical trial abundance and ultimately accelerate our delivery of medicines for patients. Ladon is our first step towards this future: one where we can learn more from every patient enrolled so that we can run smaller, faster trials with higher success rates. We built it to give drug developers quantitative support on billion-dollar clinical trial decisions.

Ask: Do you know someone in clinical development or translational medicine at a pharma or biotech? We’d love an introduction. Please email us at founders@blank.bio

uploaded image

—

The Problem

Drugs are getting easier to design, but the bottleneck is still the clinic: 95% of cancer drugs that enter clinical trials never get approved. With more therapies to test than ever before, clinical trial decisions become harder and harder. This is in large part due to patient heterogeneity. Two patients can have the same diagnosis and receive the same treatment, yet have very different outcomes: one responds well and lives for years, while the other’s disease continues to progress. Fundamentally, the challenge with making clinical trial decisions is that we don’t have good enough tools to reliably predict therapeutic response and patient prognosis.

Our solution

  • Ladon is a 3B parameter model trained on millions of RNA-seq samples that span decades of data across labs, sequencing machines, cell lines, animal models, humans, and more.
  • We are state-of-the-art, beating gene expression baselines, established biomarkers, and existing RNA foundation models at predicting patient prognosis and therapeutic response in various settings spanning 120+ oncology tasks, including external validation on 8 retrospectively analyzed clinical trials.
  • More importantly, we demonstrate that Ladon-derived predictions are also robust and generalize across tissue and clinical trial cohorts. In the technical report, you’ll see that every evaluation is measured in an out-of-distribution setting like a held-out clinical trial, another cohort, or another measurement platform.
  • We’ve built it with interpretability in mind. We also show that Ladon is best-in-class at inferring cell composition of bulk RNA-seq samples to make insights actionable.

For pharma and diagnostic companies

All Ladon needs is RNA-seq from tumor samples. If you have banked samples, we will pay for the RNA-sequencing costs! Here are some potential applications:

  • Help with go / no-go decision making: will a Phase 2 response turn into a survival benefit in Phase 3?
  • Contextualize single-arm trial results: build synthetic control arms to see how patients would’ve done on standard of care
  • End trials earlier: adjusting for prognostic signal at baseline can increase trial efficiency and end trials months earlier. Get to market faster!
  • Develop novel signatures: find prognostic and predictive signal that actually hold up in the real world. We can also show that your biomarker is truly predictive by adjusting for prognostic baselines

Our ask

  • Please reach out to us if you’d like to get access to the model
  • You can read our blog and technical report for more technical details
  • If you know any clinical development or translational leaders across pharma or biotech, we’d love to get an intro

https://youtu.be/fIS6qyKh7L8

Previous Launches
Building foundation models to design better RNA therapeutics, starting with mRNA design.
Blank Bio
Founded:2025
Batch:Summer 2025
Team Size:6
Status:
Active
Location:San Francisco
Primary Partner:Ankit Gupta