23 New Startups in AI Drug Development Space Launched 2023–2025
A snapshot of emerging startups applying AI across the drug development pipeline—from molecular design to clinical decision support.
AI began entering biomedical research in the 1970s through early initiatives like SUMEX, which explored how computing could support scientific reasoning. By the late 1980s, applications such as MULTICASE and ADAPT introduced AI-driven tools for toxicity prediction, marking its first use cases in drug development.
Joshua Lederberg at a Laboratory Instrument Computer teletype in 1974, during the launch of SUMEX—Stanford’s pioneering AI infrastructure for medical research. (Courtesy: U.S. National Library of Medicine)
By 2022, AI-generated drug candidates began entering preclinical and clinical pipelines, moving from theory to application. COVID-19 accelerated interest in computational methods, and the release of AlphaFold showed how AI could crack complex biological prediction problems. Around the same time, transformer-based large language models—popularized by tools like ChatGPT—made AI more visible across research and industry.
Regulatory agencies are responding as well—in 2024, the FDA’s Center for Drug Evaluation and Research (CDER) formed an AI council and released draft guidance on AI use in drug development workflows.
In this article: Protein and Peptide Engineering Platforms — Model Simulation and Physics‑Based Tools — Generative and Foundation Model Ecosystems — AI‑Enhanced Drug Discovery Workflows — Multi‑Omics and Precision Diagnostics — AI‑Driven R&D Support and Optimization Platforms — Autonomous Scientific Research and Lab Automation
Today we’re taking a look at a new wave of startups that entered the AI drug discovery scene over the past two years, between 2023 and early 2025.
Protein & Peptide Engineering
Xaira Therapeutics
Location: San Francisco, CA, USA
Launch Year: 2024
Xaira Therapeutics, a biotechnology company with $1 billion in funding, is focused on advancing drug discovery by integrating AI research, data generation, and therapeutic development.
Leveraging the work of co-founder Dr. David Baker at the University of Washington, Xaira utilizes AI models such as RFdiffusion and RFantibody to design novel proteins and antibodies targeting complex biological challenges. Its computational framework covers the entire drug development process, from biological discovery and molecular design to clinical development.
Xaira’s data platform integrates diverse, high-resolution datasets to train models and generate hypotheses. It aims to reveal disease mechanisms and improve predictions, especially where existing data is limited.
The company uses an iterative pipeline that refines its models based on experimental results, helping to accelerate the development of potential therapeutics.
Abiologics
Location: Cambridge, MA, USA
Launch Year: 2024
Abiologics is a Flagship Pioneering-founded company developing a new class of suprabiological therapeutics called Synteins. These are synthetic proteins composed of amino acid building blocks beyond the 20 naturally occurring L-amino acids, including D-amino acids. The company uses generative AI and deep learning to design these structures.
According to Abiologics, Synteins are intended to overcome the limitations of recombinant protein technologies. The company states that its approach enables scalable production of ultrastable synthetic proteins with enhanced bioavailability and tissue penetration, including access to solid tumors and the gastrointestinal tract.
Abiologics combines computational protein design with automated laboratory systems to accelerate the discovery-to-validation cycle. It aims to reduce both time and cost compared to traditional biologic development. One of its current candidates is a Syntein-based GLP-1 therapeutic in development as an oral treatment for metabolic disorders.
Abiologics has received a $50 million initial financing from Flagship Pioneering to advance its platform and therapeutic pipeline.
VRG Therapeutics
Location: Budapest, Hungary
Launch Year: 2023
VRG Therapeutics is advancing the development of next-generation cell and gene therapies and peptide-based pharmaceuticals. The company has built a proprietary miniprotein-based drug discovery platform, AI-MPRO, which integrates two core technologies: ISEP (Individual Sequence Enrichment Pattern) and CREATe.
ISEP uses directed evolution to generate a peptide library of over one million variants, drawing from both natural and AI-designed scaffolds. These variants are screened using advanced wet-lab techniques and next-generation sequencing to find high-affinity, highly selective candidates. CREATe supports this process through AI-guided scaffold selection and machine learning-based optimization, converting peptides into miniproteins—compact, stable therapeutic molecules that offer the precision of monoclonal antibodies along with small molecule benefits (better stability, lower immunogenicity, and possible oral bioavailability).
This integrated approach is described as enabling rapid, targeted drug discovery with high specificity across diverse targets. VRG highlights a pipeline directed at unmet needs in autoimmune disorders, inflammatory diseases, and oncology. It is currently incubating two lead candidates: VRG-145 for migraine prevention and another program targeting central nervous system disorders.
As of 2024, VRG Therapeutics has raised $11 million in funding, including a $5.5 million Series A round.
Reticular
Location: San Francisco, CA, USA
Launch Year: 2024
Reticular is a startup focused on developing interpretable AI for protein engineering to enhance drug discovery. The company aims to address what it sees as a key problem in AI-driven drug discovery—limited control and transparency in protein AI models like AlphaFold. Reticular’s approach targets the manageability of "black box" models through mechanistic interpretability techniques—an approach shown to work with large language models like Claude and ChatGPT.
Reticular’s technology is designed to let pharmaceutical companies precisely direct protein AI models to produce biologically relevant designs and minimize reliance on costly trial-and-error experiments. By unlocking information within these models, Reticular seeks to let users control outputs as easily as prompting a chatbot, potentially making protein design more predictable and efficient.
A notable feature of its tech is described in a proof-of-concept, which describes the identification of the first interpretable features in protein models, reportedly allowing exact manipulation of biological functions. The method leverages sparse knowledge extraction, even with limited validation data, drawing parallels to advancements in AI interpretability research.
Reticular is currently piloting its technology with early-stage biotechs, with plans to scale rapidly, aiming to unlock untapped potential in biological models for reliable and cost-effective drug discovery.
Up to now Reticular has obtained $500k in the pre-seed round.
Model Simulation & Physics‑Based Tools
Nanograb
San Francisco, CA, USA 2023
To read the rest of this article, upgrade to a BiopharmaTrend Pro subscription.
Gain full access to all of our deep dives and content archives.
