Read/Write Labs launches to build an AI platform that writes code for the brain

A new name and a bigger team

According to the company, its new identity reflects a focus on creating the intelligence layer that can read and write information to the brain through various neural interfaces. The goal is to treat a wide range of disorders and improve cognition. Read/Write Labs is also exploring how recent discoveries in biological intelligence could inform more human-centered AI systems.

The company is led by co-founders Adam Hanina, CEO, and Martin Schrimpf, Chief Scientific Officer. Schrimpf is the creator of the industry-standard Brain-Score and leads the NeuroAI Lab at EPFL. Three scientific fellows and advisors join them:

• Jim DiCarlo, external scientific fellow and an MIT professor of systems and computational neuroscience, whose research on biological vision produces machine-executable scientific models of how the brain works.
• Dan Yamins, external scientific fellow and a Stanford University professor, described as a pioneer in developing AI models of the brain.
• Stephen Wolfram, external advisor, founder and CEO of Wolfram Research and creator of Mathematica, Wolfram|Alpha and the Wolfram Language.

The company also named Katherine Fairchild as Chief Technology Officer and Ben Lönnqvist as Head of NeuroAI. Fairchild was previously Senior Director of AI/ML at Digital Diagnostics, developer of the first FDA-authorized autonomous AI diagnostic system, and recently led engineering at the MIT Siegel Family Quest for Intelligence. Lönnqvist holds a PhD in Computational Neuroscience from EPFL and leads the company's digital brain modeling work. His research has been published at conferences including ICLR and ICML.

The release notes that Professors DiCarlo and Yamins serve in their individual capacities as external advisors, and that institutional affiliations are listed for identification only.

Digital brains for treatment discovery

For brain health, the company develops what it calls digital brains: AI models that predict how real brains will respond to sensory or direct neural stimulation. According to the company, these models make it possible to study brain dysfunction and design treatment candidates in silico. It says this has the potential to replace much of today's slow and expensive in vivo testing, making treatment discovery far faster and broader than currently possible.

Schrimpf said that AI models can now predict the brain's inner workings "with unprecedented accuracy." In his words, this allows "writing precise patterns of activity - 'code' for the brain - to treat disorders, and building AI that is more deeply aligned with human intelligence and values."

Using brain data to guide AI

The company also argues that the same science reaches beyond therapeutics. Read/Write Labs believes the human brain offers an important and underexplored path toward AI that is safer and more aligned with human values. It points out that qualities like compassion are difficult to specify through behavior alone, because very different internal mechanisms can produce the same caring-looking answers.

Rather than relying exclusively on behavioral training and external safeguards, the company says it is exploring how to use human brain data to test and guide computational systems. The aim is AI that is safer and better suited to working alongside people. Hanina described the approach as working in both directions: "using AI to understand and treat the brain, and using the brain to shape AI."

Patents and funding

Read/Write Labs says it holds a broad patent portfolio with over 30 patents pending and granted. It has received funding from the National Institutes of Health, the Swiss Government, and the Wyss Center for Bio and Neuroengineering.

The announcement is a statement of strategy and team rather than a product launch. It gives no timelines, clinical results or availability details, so the company's claims about faster treatment discovery and safer AI remain its own stated goals.