Imperial College London spinout Cogitat has been acquired by the neurotech company Indivi for an AI foundation model that can translate brain wave data acquired using EEG into read-outs of users' brain states.
Indivi, based in Basel, Switzerland, aims to use the acquisition for an undisclosed sum to transform itself into the world's leading brain AI company, planning initially to use the technology to help pharmaceutical companies detect whether experimental treatments for neurodegenerative diseases such as Alzheimer's are impacting the underlying biology before their effects may show up in behaviour.
Cogitat was launched in December 2020 to develop a brain-computer interface that allows users to control virtual reality games using only the mind. Its early deep learning models could recognise people's behavioural intentions in data from electroencephalogram (EEG) headsets, which are cheaper and easier to use than other forms of brain imaging but yield very noisy data.
The company later extended its AI models to interpret users' levels of attention or distraction. It now aims, with its foundation model NeuroRVQ, to provide a general-purpose model that does for the brain what large language models do for language.
Cogitat, based in the UK and Greece, will function as a business unit at Indivi, helping drive the biotech's transformation into a neuroscience-based AI company that plans to pursue applications across the life sciences and eventually consumer technology.
Cogitat's technology won first place in the EEG Transfer Learning (BEETL) competition at the prestigious AI conference NeurIPS. The team attributes the capabilities of its interpretable AI technology to the quality of training data acquired through carefully designed experiments at Imperial well before the current AI gold rush.
"Public EEG archives are a valuable starting point, but they mostly capture the brain at rest. We design every experiment with a purpose: when we know what the brain was doing at each moment, we know what the model is learning," said Dr Dimitrios Adamos, Cogitat's Chief Technology Officer and co-founder. "A cardiologist learns most from how the heart responds to exercise. We do the same with the brain: we record it while it's at work."
They argue that there are no shortcuts to the insights behind Cogitat's patented technology, which come from hands-on data-gathering and model development. "Our machine learning engineers had to go through the data-collection process themselves to understand what worked. For example, early on, some signals that looked like concentration turned out to be eye movements. Disentangling genuine brain activity from everything else in the signal is a long, laborious process," said Dr Allan Ponniah, Cogitat CEO and co-founder.
Guilhem Dupont, CEO and Founder of Indivi, said: "This acquisition marks a new chapter for Indivi as a neuroscience-based AI company. The brain is the most complex organ in the body, and remains one of the hardest to understand and measure. Cogitat's EEG foundation model changes that, decoding brain activity and revealing signals far closer to the underlying biology than ever before.
"That opens up exciting opportunities: initially the potential to provide drug developers earlier data on how new treatments are working for brain conditions, and, in the long-term, could inform the future development of next generation brain health technologies beyond life sciences."
Dr Ponniah from Cogitat added: "Every brain is unique and speaks its own language. We've built an advanced AI model that works across every brain's unique language without individual training to identify subtle changes in brain activity that have not been easily identifiable before. This addresses one of neuroscience's biggest translational gaps: the failure to detect early changes in brain biology that predict whether a treatment or product will actually work.
"With Indivi we have the opportunity to take our work to the next level, putting the model to work in the real-world, validating the technology and delivering true value."
NeuroRVQ will be presented at the 40th Annual Conference on Neural Information Processing Systems (NeurIPS, 6-12 December 2026)