Light can now switch individual, genetically chosen cells on and off in a living brain — millisecond by millisecond. The biology is extraordinary. The software that will carry it to patients mostly doesn't exist yet. This page is where we start thinking about who builds it.
The breakthrough is real, and it is not magic. Optogenetics controls light-sensitized cells with light. It does not read minds, and we will never say it does.
The hard part is moving from a lab bench to a human life. Between a working experiment and a person who benefits sits a decade of calibration, training, measurement and maintenance — most of it software.
Nobody owns the connective tissue. Biologists own the biology, hardware teams own the device. The software that makes the whole system usable is everyone's problem and no one's job.
Honesty about evidence is the only durable advantage. In a field prone to overclaiming, saying exactly how mature each result is earns the trust that wins the serious work.
You earn the right to build, one real problem at a time. No platform on day one. Start with a single bottleneck a real team will pay to remove.
In 1979 Francis Crick named the problem: to understand the brain, you'd need to switch one chosen type of cell on and off without touching its neighbours — and do it as fast as the brain itself works. Every tool of the era forced a trade-off. Optogenetics is what broke it.
That broken trade-off is the entire reason this field exists — and the reason its software has to be just as precise and just as fast as the biology it serves.
None of them is enough alone. This is why the field took decades — and why making the whole system work together is still the unglamorous, valuable part.
A single protein, borrowed from algae and microbes, that turns a flash of light directly into an electrical signal inside a neuron — safely, and without harming the cell.
Genetic delivery that installs that switch in one chosen cell type, buried among thousands of others that must stay untouched.
Hardware that carries precisely timed light into living tissue without overheating it or disturbing natural activity — in a subject that is awake and moving.
The whole discipline, in plain language, on a single page. This is the small version — there are 40+ published papers behind each branch, and we're turning them into teaching pieces one at a time.
Everything the field can do to a chosen cell comes down to three moves. Worth knowing, because each one needs its own software to be useful to a clinician.
A pulse of light makes the chosen cells fire — on demand, as fast as a hundred times a second. This is how you cause an effect instead of just observing one.
Another kind of light-switch quiets the chosen cells instead — used, in research, to interrupt runaway activity such as a seizure the moment it starts.
A subtler set of switches nudges a cell's internal signalling rather than making it fire — a dimmer, not an on/off button.
The field is moving from pre-set light pulses to systems that watch the brain and react in real time. That shift is almost entirely a software and control-theory problem — and it's the part we find most interesting.
Light fires on a fixed schedule, no matter what the brain is actually doing. Good for proving an effect; blind to the moment.
Sensors read the brain continuously; software decides, pulse by pulse, when and how to respond — inside a window shorter than a tenth of a second.
A closed loop has to sense, decide, and act in under ten milliseconds, over and over, safely, for years. That is not a biology problem anymore. That is the software we care about.
Follow any one of these programs from the lab toward a patient, and the same five layers appear every time — usually built from fragile scripts by people whose real job is biology or hardware.
No customers, no product, no claims. Here is the real work in front of us right now — the kind a founder can start this month.
Working through 40+ published papers and turning each into a plain-language explainer, like the ones on this page. Understanding first.
Writing down what we believe about where this goes and why the software layer is the place to stand — tested against people who know more than us.
Talking to therapy developers, device teams, and clinicians who feel the software gap — to find one specific, budgeted problem worth solving first.
A small, working example on clearly labelled synthetic data — to show what good software here looks like, before claiming anything.
This page exists to start one good conversation — not to sell anything. The next step is a coffee and a hard question, not a contract.