Drug discovery remains one of the most inefficient industries, with over $300 billion spent annually, more than a decade required per drug, and a failure rate above 90%.
Helical, a Luxembourg-based startup founded in 2023, believes the problem is not a lack of AI—but a lack of systems that turn AI into usable science.
The company has raised $10 million in a round led by redalpine, with participation from Gradient, BoxGroup, Frst, and and notable angels including Aidan Gomez (CEO Cohere), Clement Delangue (CEO HuggingFace) and Mario Goetze (pro soccer player).
“Scientists need a structured way to iterate on hypotheses, like in a wet lab—but in silico.”
Originally built as an open-source platform for biological foundation models trained on DNA and RNA data, Helical quickly realized that access to models alone was not enough to drive scientific decisions. “From the beginning, we knew a model platform alone would not be enough,” says co-founder and CTO Maxime Allard. “Scientists need a structured way to iterate on hypotheses, like in a wet lab—but in silico.”
This insight led to its shift toward a “virtual AI lab,” designed to turn fragmented model outputs into reproducible scientific workflows.
The platform connects two environments—a Virtual Lab for scientists and a Model Factory for ML engineers—built on shared data, models, and results. “All scientists iterate—they don’t test a hypothesis once,” Allard explains. “The Virtual Lab enables that iterative process computationally.” The goal is to replace siloed tools and non-reproducible analyses with standardised in-silico experimentation.
“Without reproducibility, AI in biology cannot scale.”
Helical already works with several top-20 pharmaceutical companies, including Pfizer on predictive safety biomarkers and Tanabe Pharma, with early deployments compressing discovery timelines from years to weeks.
“It’s not just about speed,” says Allard. “It’s about making scientific decisions reproducible and defensible.” Reproducibility is central to the platform, especially as regulators like the FDA increase expectations for transparency in AI-driven drug discovery. “Without reproducibility, AI in biology cannot scale,” he adds.
More broadly, Helical reflects a shift in AI value creation—from building models to orchestrating them inside real-world workflows.
While challenges remain due to the multi-dimensional nature of biology, the company’s ambition is clear: to make in-silico discovery a standard engine for pharma R&D, enabling scientists to test, reproduce, and scale hypotheses at computational speed.
Read more articles:
