Brett Kagan is Chief Scientific Officer at Cortical Labs, a company building the world’s first commercial biological computer that fuses lab-cultivated neurons from human stem cells with hard silicon.
What is Cortical Labs?
Cortical Labs started with the simple idea that you can unlock a whole range of capabilities if you are able to interact with the brain. When it comes to intelligence, the brain is the benchmark for general intelligence. Despite trillions of dollars spent on artificial general intelligence, there’s no evidence that anything other than biology can achieve it. So we asked, why reinvent what you can harness?
Using synthetic biological methods, we can generate neurons at scale and interact with them through electricity, which is the shared language of brains and silicon. This not only allows us to explore intelligence but also opens doors for disease modeling, drug discovery, and fundamental neuroscience research. What makes this approach so powerful is its iterative nature. Each step builds on the last. We develop tools to study the brain, which leads to early forms of rudimentary intelligence. In the process, we gain deeper insights into how these systems function, feeding into the next iteration. What starts as a valuable commercial tool today lays the groundwork for even greater potential tomorrow.
A weekly dispatch featuring exclusive interviews with deep tech founders & a roundup of the most important deep tech news.
Tell us about your journey to Cortical Labs.
My background spans neuroscience, informatics, and psychology. This has been a lifelong interest, or at least an adult lifelong interest. But even with a broad background, this is not something any one person can do alone. It requires a multidisciplinary team.
When Cortical Labs was founded, we brought together an electrical engineer, people with computer science backgrounds, general programmers, and myself with a neuroscience background. We were able to build on each other’s expertise in a way that is difficult in an academic setting. Academia tends to create silos, where deep expertise in one area does not always lead to meaningful interdisciplinary collaboration. Startups, by contrast, thrive on that kind of exchange.
Likewise with funding opportunities. Australia’s funding environment for academic research is pretty poor at the moment. For a while the effective rate of early career researcher grants has been around 7%. You’re required to generate expressions of interest, of which only 50% are allowed to progress to the grant proposal stage. It’s a lottery, and not an ideal way to assign new research. So one thing I hope we’re doing at Cortical Labs is reducing the skepticism towards companies doing deep, rigorous research. We’re showing that it can be done separate from a university or spinout and still deliver practical benefits very quickly.
How does Cortical Labs sustain its multidisciplinary approach?
We’re still a relatively small team with about 22 people. No one on the team has the same specialty. There are some clusters, especially in cell culture biology, but even those are made up of just two-person teams with overlapping disciplines. Our team includes specialists in drug discovery, stem cell generation, and neuron production, as well as experts in electrical engineering, front-end and back-end hardware development, FPGA development, informatics, and statistics. It’s a diverse group with a wide range of skills. Keeping everyone aligned can be a challenge, but it’s a fun one.
It’s an incredible team. People are genuinely excited to come to work. More often than not, I find myself telling people to go home, slow down, or not come in at 1 a.m. to run an experiment. I rarely have to encourage anyone to spend more time in the office or lab. It’s a privilege to work with people who are that driven.
Can you tell me about your core offering?
At our core is CL1, which is a biological computer. When we began Cortical Labs we started with the question, “What do you need to interact with brain cells?” You need a way to communicate with them, and you need a way to keep them alive. The CL1 is designed to do both. It uses a microelectrode array (MEA), which is a grid of small electrodes that brain cells can grow on. When coated with extracellular matrix proteins, these devices support the growth of various cell types, and we are exploring three different approaches.
We culture classic monolayers, which are semi-flat layers of cells. We also work with more complex organoids, which are small three-dimensional structures. Beyond that, we are pioneering what we call bioengineered intelligence, where we use discrete structures to position and control the connections between cells. This is at the cutting edge of the field.
On the life support side, the CL1 has pumps that move media through the cells, preventing them from sitting in stagnant liquid. Media is the nutrient-rich liquid that keeps the cells alive. The system includes filtration to remove waste, gas exchange to maintain the correct oxygen and CO2 levels, and temperature regulation to keep everything stable.
The CL1 simplifies real-time interaction with the cells. It converts analog signals into digital form, processes those digital signals in a simulated environment, and then sends the neuron activity back as electrical stimulation. The entire system operates in under a millisecond. And it is easy to use. A simple Python API allows users to control the hardware with minimal coding.
What applications do you envisage for CL1?
A wide range of people are interested in CL1, from basic neuroscience researchers to those working in drug discovery, disease modeling, pharmaceuticals, and biotech. We also see interest from machine learning researchers and those exploring alternatives, particularly those who recognize the limitations of current AI models like LLMs.
Beyond that, we have people who want to use it for all kinds of unusual and creative applications. Some have contacted us about mining crypto, others for robotics. Some want to use it to create music, and some projects are under confidentiality agreements.
All of these ideas may seem to have different levels of merit on face value, but the reality is that we do not yet know what will work. We are always happy to support people as they explore. When silicon computing first emerged, people likely saw it as just a big calculator for spreadsheets. The first person who wanted to build a music synthesizer or create video games was probably seen as eccentric. Yet look at where we are now.
One of our core values is that anyone can be a scientist. Our goal is to help people become scientists, do the best science possible, and build the best innovations they can.
Will this technology hit the “mainstream” soon?
Of course, for something to be “mainstream” we still have lots of work to do and the reality is we can’t do it alone. The CL1 is built to be as affordable as possible and will become more accessible over time. Early users will gain access this year in 2025. Our belief is that if we can provide people with the tools to rapidly iterate on new ideas and explore the limits of what is possible, we can bring this technology to the mainstream much quicker than would ever be possible alone.
If you look at all growth of all new technologies, lowering the barrier to entry precedes massive technological growth and adoption. If we can enable that, I believe we’ll see some incredibly exciting applications arise.
You have previously demonstrated organoid intelligence playing Pong. Any new developments in gaming or other applications?
We’ve been exploring several new possibilities, and there will be exciting updates to share. We will also be publishing updates and research papers. Rigorous science takes time, so we do not rush the process, but we expect to share breakthroughs on the information patterns used to interact with neurons before the end of the year. At the same time, going through the peer review process requires patience, and we are committed to doing it right. Some of it is still under wraps, but much of our focus has been on building the foundation for CL1.
The benefit of building this foundation is that now we can work much quicker. For example, it originally took 12 to 18 months to develop the Pong environment on our first prototype. With our new system, we rebuilt that environment internally in just a few weeks. A PhD-level intern then created a more complex environment in about three weeks—something that would have taken 18 months before.
CL1 was completed at the end of last year, and since then, progress has accelerated. We’ll be revealing more later this year on what we’ve been able to accomplish.