Why We're Building Synaptrix And Why It Matters

Some problems in the world can wait. Brain-computer interfaces (BCI) aren’t one of them. If we don’t build the right kind of BCI company now, someone will build the wrong kind. That’s why Synaptrix exists: we are singularly focused on building a scalable, computation-first non-invasive BCI platform, unlike any before.

Some implants will do things that non-invasive BCI will never do (cure blindness for example). That work matters and I respect it. But it will not be the way billions of people interact with machines, doctors, or their own autonomy. The reality is in the numbers. According to the WHO, over 3 billion people live with some neurological condition right now. Less than a fraction of a tiny fraction will ever receive invasive implants. That is not a failure of medicine. It is just how humans actually behave.

My clearest takeaway from the decades of work done in brain–computer interfaces is that the future of neurotechnology will be determined by general methods leveraging computation. Synaptrix is built on the contrarian thesis that only computation, not incremental hardware improvements, will drive transformative advances. The reason is simple: human biology is fixed, but computation compounds. Every few years, compute grows exponentially cheaper and more abundant, while the fidelity of non-invasive sensors grows marginally. Yet most neurotech research continues to behave as if computation were fixed, and the only way forward were to build better electrodes or sensing modalities.

For a short while, this hardware-first approach appears to work. A new optical headset promises cleaner signals; a new EEG amplifier claims lower impedance. Researchers extract a few extra microvolts of clarity and celebrate progress. But, as history shows, this progress plateaus quickly. The noise floor of biology remains. The skull does not become transparent. Time spent engineering around these constraints is time not spent learning how to interpret the signals that already exist.

When computation is applied at scale, however, the ceiling moves. In machine learning, this pattern has repeated many times. Medicine did not get revolutionized by better X-ray film stock, but instead when neural networks learned to see pneumonia shadows like radiologists. And language did not hit GPT-4 scale because someone built a better microphone, but because transformers made the data legible. Every domain that once depended on handcrafted expertise has been overtaken by methods that scale with compute.

Neurotech is no exception. The non-invasive signals we record (EEG, EMG, fNIRS, DOT, etc.), though limited by resolution and bandwidth, likely already contain everything needed to infer intent, affect, and pathology. What is missing is not new hardware, but new understanding. The lesson that AI learned a decade ago now arrives for neuroscience: the future belongs to systems that learn from massive computation, not to those that rely on human intuition about how the brain should look or behave.

This is why Synaptrix exists. We will not build toys or incremental accessories. Our goal is simple: a non-invasive BCI device that costs less than a phone, running novel deep learning models that make the brain fully legible. Such a system would let people interact with their surroundings in ways once thought impossible—telekinesis through thought alone. It would let paralyzed individuals move their wheelchairs as easily as they once moved their limbs, restore speech to those who have lost their voices, and one day enable the diagnosis and even treatment of all neurological diseases. This approach is how we scale neurotech to the entire world.