Briefly Bio Launched to Help Solve the Scientific Reproducibility Crisis

Science is in a reproducibility crisis, according to officials at recently launched Briefly Bio, which is headquartered in London. They add that in preclinical research, it’s estimated that over 50% of efforts to reproduce experiments fail, costing the industry over $50 billion each year.

To help solve the problem, the company has developed software that makes lab work more reproducible by helping scientists capture and share their work clearly and consistently, say the Briefly Bio co-founders Katya Putintseva, PhD, Harry Rickerby, and Staffan Piledahl.

With its launch, Briefly Bio has secured a $1.2 million pre-seed funding round. It was led by Compound VC, with participation from NP Hard, Tiny VC, and angel investors across tech and biotech.

Biological experiments have become increasingly complex. With this growing complexity, details that are critical to shared understanding are often undocumented and lost, the co-founders continue, pointing out that this makes scientific collaboration inefficient.

They list such points such as lab scientists struggle to reproduce and build on top of each other’s experiments; data scientists do not have the necessary context to analyze the data produced in their labs; and automation teams lack all details to build robotic labs.

Shared language for experiments

Briefly Bio decided to tackle these problems by creating a shared language for experiments that is consistent across scientists, and clear for any collaborator to understand. Their software uses AI to convert existing experiment descriptions into this consistent format, while automatically filling in gaps and spotting errors. This helps capture the value of every experiment that is run and enables scientists to learn from each other’s work, notes a company spokesperson.

Before founding the company, Putintseva, Rickerby, and Piledahthe worked together at drug discovery startup LabGenius, where they helped build its ML-driven antibody discovery platform.

“Scientific methods are a bit like software code, they are a set of instructions that define how an experiment should be run,” comments Rickerby, CEO. “The majority of this ‘code’ is incomplete, since writing up each experiment completely takes a huge amount of effort. Now, with LLMs, there’s a way to make these methods consistent without imposing on a scientists’ workflow. As Github helped software engineers collaborate and build on each other’s code, we think Briefly can help scientists and engineers do the same with their experiments.”

With AI and high throughput experimentation, there is an opportunity for huge improvements in the efficiency of scientific discovery, he states. Hundreds of billions of dollars are being invested through startups and big pharma to take advantage.

To realize this potential, science needs more consistency and transparency in how these datasets are generated, since the value of any model is a product of the data it has been trained on. Briefly Bio is building this necessary layer of infrastructure to accelerate scientific discovery in biology, says Rickerby.