BioRender launches Leo AI agent for scientific research figures

Toronto-based software startup BioRender has launched Leo, an AI agent specifically designed to assist researchers in creating accurate scientific figures. Unlike general-purpose generative AI tools such as ChatGPT, Midjourney, or DALL-E, which often produce inaccurate biological structures or inconsistent legends, Leo is built to function more like a trained medical illustrator.

BioRender co-founder and CEO Shiz Aoki stated that general AI tools treat scientific figures as standard illustrations, leading to hallucinations and visual errors that can undermine scientific credibility. Leo aims to address this by guiding users through a process of planning, sketching, and building images. Instead of jumping directly from a prompt to a final picture, the agent asks questions and provides sketches that researchers can subsequently refine by hand to ensure precision and adherence to scientific nomenclature.