The Non-Sci-Fi Story of Brain-Computer Interfaces: A Conversation with Ximing Wang of "Kongshan Ci" | Yunqi Podcast Attent!on
Beneath the Brain, Above the Signal
In recent years, AI has been sprinting forward at a staggering pace. We've used "neural networks" to simulate how the brain connects, employed large models to approximate language, reasoning, and memory, and grown accustomed to understanding intelligence as something that can be trained, deployed, and invoked. Yet the brain itself — the very organ that AI repeatedly borrows as metaphor — remains shrouded in mystery.
From the first brainwave curve recorded in 1924 to Neuralink reigniting public imagination about brain-computer interfaces, humanity's exploration of neural signals has never stopped. The flashiest part captures the most attention: capturing signals, controlling cursors or robotic arms with thought, reconnecting with the external world.
But that's not the only destination worth reaching. Emotion, sleep, cognition — these invisible burdens of daily life — are becoming the real entry points for brain-computer interfaces to reach ordinary people. In this episode of "Attent!on," we want to explore a non-sci-fi story about brain-computer interfaces with you.
Our guest today is Ximing Wang, co-founder of Kongshanci, a brain-computer project under the Yunqi-SJTU AI Angel Fund.
In 2024, as large models and agents dominated the zeitgeist, Ximing Wang — Cambridge economics background, years of experience in medical data and algorithm development — chose to team up with people from "heavier" backgrounds: surgical robotics, medical imaging cloud, long-term expertise in neuromodulation, simulation, and transcranial focused ultrasound.
Together they built Kongshanci, entering a hardware-heavy, long-cycle track: mental and cognitive disorders.
This is a gap that has long existed yet faces persistent bottlenecks. Take depression: traditional medication, psychotherapy, and physical therapy all play roles, but also face structural challenges — long cycles, high individual variation, and difficulty scaling delivery.
What Kongshanci wants to do is connect a more precise "circuit diagram" to the complex brain activity beneath the skull: first making the invisible signals, pathways, and abnormal changes more visible, then using appropriate physical energy for targeted intervention, bringing disordered parts back to relatively stable states.
This is Ximing Wang's slice of understanding brain-computer interfaces: it's not just "reading the brain to control external devices," but transforming invisible emotional and cognitive damage into readable, quantifiable, verifiable signals and energy, then intervening at the right location.
**Along this path, Kongshanci is building a brain-computer interactive neuromodulation platform for mental and cognitive disorders. In other words, they're not chasing cooler sci-fi stories, but providing a more verifiable and accessible intervention path for brains worn down by daily invisible burdens.
Today's guests:
Ximing Wang
Co-founder, Kongshanci
Na Li (Linda)
Managing Director, Yunqi Capital
Scan the QR code above to listen to the full episode
01 Going Deeper Into the Brain
Linda:
Many founders enter the field because of a specific event in their lives, or some dramatic story. What brought the Kongshanci team together to tackle emotional and cognitive problems?
Ximing Wang:
Our team has one major characteristic: everyone deeply empathizes with depression, cognition, and related issues. Some have experienced these struggles themselves; others have watched family or friends caught in them.
Emotional and cognitive issues actually run through a person's entire life. We're not just focused on depression, but on overall emotional quality. Kongshanci will start from one entry point, do it well, then gradually expand our boundaries and capabilities.
Linda:
When you first started planning the company, why did you choose to address this problem through brain-computer interfaces and neuromodulation?
Ximing Wang:
I've been working in this direction for the past 6 years. I'm personally very interested in it, and I see a clear trend.
I used to work more on pure algorithms and data, but I was very convinced that on the hardware side, in the combination of signals and energy, something better would emerge. So when we started planning the company in early 2024, our first internal discussion was about solving this problem through the brain-computer interface approach.
It's just that at that time, the market hadn't fully bought into this larger direction. So when building the product, we narrowed our scope once, moving forward with a sufficiently small unit first, then gradually spreading our philosophy to the industry and beyond.
Linda:
When did you start considering your first fundraising round? How did you connect with Yunqi?
Ximing Wang:
At the very beginning, we pooled some money ourselves with many friends to get started. The company was founded in October 2024. After completing the demo, we received our first angel round.
Later we were fortunate to meet Michael Mao (note: Michael Mao, Founding Managing Partner of Yunqi Capital). After about 30 minutes of communication, he was quite firm on this direction. Then colleagues from Yunqi learned more, and we gradually revealed our entire matrix philosophy — including how brain-computer interfaces are positioned for emotion and cognition, and how they combine with AI — which gained further recognition from Yunqi, leading to our later Pre-A round. In the recent Series A round, Yunqi also continued to double down.
Linda:
What was your impression of Michael in those 30 minutes?
Ximing Wang:
Gentle and refined. Michael was probably the first investor we met with a heavier technology label. His thinking dimensions and perspective were different from traditional medical device investors. So communication flowed quite smoothly. I'm not from a particularly traditional medical device background either, so I could better express what we originally wanted to do. For example, globalization, or moving toward the consumer end in the future — these aren't particularly conventional paths in traditional medical devices. But emotion and cognition themselves have strong daily-life attributes; they happen at home, in the small moments of life.
02 Demand and the Gap
Linda:
From the perspective of medical data, where do issues like emotion, cognition, and sleep sit within the entire healthcare system?
Ximing Wang:
Many people initially think of them as a specific category of disease, but actually, they're more like "capillaries" throughout the healthcare system — truly present in any scenario.
I once had a major sports injury — torn knee ligament, plus fractures and bone cracks. The psychological trauma during surgery and recovery was enormous. You keep replaying the injury, with lots of worry, frustration, and anxiety.
So these issues are actually everywhere in the healthcare system and need more recognition and acknowledgment. The lack of attention in the past was partly due to awareness gaps, and partly because solutions weren't systematic enough or effective enough.
Linda:
For depression specifically, where do the real treatment bottlenecks lie?
Ximing Wang:
About 30% of depression patients have treatment-resistant depression and don't respond well to medication. Another figure: roughly 40% of depression patients in China need to visit three or more hospitals before achieving relative relief. This really reflects the industry's past situation: because treatment methods weren't efficient enough, many people felt they were knocking on doors everywhere with nowhere to turn.
We entered this industry partly because we observed this. We spent a year embedded in hospitals, observing patient situations in different clinics, conducting extensive interviews with patients and doctors. For example, whether they felt their diagnosis was the problem or the treatment was; whether treatment duration was too long or side effects too severe. These were foundational homework for our systematic understanding.
Linda:
Where are the ceilings for medication, psychological counseling, and physical therapy respectively?
Ximing Wang:
Traditional medication has two relatively consensus issues. First, the effective rate isn't that high, roughly 30% to 40%. Second, side effects still exist — weight gain, appetite-related issues, or other side effects. The medication cycle is also quite long, possibly several months to a year.
Psychotherapy has developed rapidly in recent years, but its core problem is delivery consistency. It heavily depends on experience, training, and long-term stable relationships.
Physical therapy itself is an effective method, but it also had major problems in the past. For example, some treatments require 6 to 10 weeks of repeated hospital visits. Going to the hospital every day already discourages many people. And after 6 to 10 weeks, the effective rate is also roughly 30% to 40%. This means patients invest lots of time and energy, but may not get cured, and could even see symptoms worsen due to disappointment.
Linda:
Beyond depression, do cognitive disorders, sleep disorders, and anxiety face the same supply-side dilemmas?
Ximing Wang:
These directions share two common points. First, if we call this category demand-side, then its supply-side is severely insufficient. Good solutions that can meet this demand are extremely scarce. Second, they all relate to the deep brain. The deeper you go, the higher the research difficulty, the greater the complexity, and the finer the functional divisions.
This is also one reason we believe this field has sufficient runway for continuous refinement over a long time.
03 Another Imagination
Linda:
When people mention brain-computer interfaces, the first reaction might be Neuralink, craniotomy, electrodes. What is Kongshanci's understanding of brain-computer interfaces? Do you also need to open the skull?
Ximing Wang:
That depends on what stage we're at and what problem we're solving. One important point for us: we're a product company. We'll follow technology development, have colleagues with different backgrounds join, and make excellent products.
We're not a completely non-invasive brain-computer interface company. We're also continuously investing in and observing invasive directions, but haven't formed products at this stage yet, so we won't disclose externally. Our style is: we basically only talk after it's done.
We don't think brain-computer interfaces should be completely divided into invasive and non-invasive. What's more important is: what problem does it actually solve? Is it solving emotion and cognition-related problems, or motor intention problems? The purpose determines the technical path forward, and also determines the relatively optimal solution at the current stage.
Linda:
When is invasive needed, and when is non-invasive suitable?
Ximing Wang:
Take playing piano as an example — fingers moving at high speed continuously, this relates more to motor intention and demands very high temporal resolution. But if you switch to emotion, a person switching between ecstasy, sadness, calm, and rage isn't changing at high speed like finger movements. Emotion is relatively more stable; you don't need to frantically interpret every second's changes. So its solution is different from the skull-opening approach for reading motor intention.
Linda:
In relatively accessible language, what problem is Kongshanci actually solving?
Ximing Wang:
We can look at it in stages. In stage one, we hope people recognize: we're a company using brain-computer interface methods to treat depression, providing solutions with high efficiency and strong recognition from doctors and patients.
In stage two, we hope people recognize our capabilities across the full energy spectrum. We call it the brain-computer interactive neuromodulation platform. We'll combine different signals with different energies, focus on the mental and cognitive disorder domain, and make excellent products and solutions.
Linda:
You just mentioned that medication and psychological counseling each have bottlenecks. But the general public might have a more fundamental question: taking medicine to regulate chemicals, we can relatively understand that. But magnetic, electrical, ultrasound — these physical energies, why would tapping on the brain improve mental illness or sleep problems? What actually happens in the brain under physical energy stimulation?
Ximing Wang:
The development of this discipline is somewhat the reverse of pharmaceuticals. Many neuromodulation technologies originally stemmed from observations or accidental phenomena, with scientific investigation following afterward.
So far, human understanding of the brain is still in very early stages. The mechanisms behind many phenomena are very complex and not fully understood. But over the past decade or two, understanding of how these physical energies affect the brain has continuously been refreshed.
**Take magnetic stimulation as an example. The most basic principle can be understood like this: **a large current passes through a coil instantaneously, generating a pulsed magnetic field. After this magnetic field passes through the skull and brain tissue, it induces an electric field locally. What actually drives changes in the neural membrane is this induced electric field. Researchers then observe neuronal firing, neurotransmitters, EEG rhythms, and deep brain region activation — it involves both electrophysiology and chemophysiology, ultimately possibly acting together to improve symptoms.
Some of the latest research continues advancing this understanding. For example, some studies have found that pulsed magnetic stimulation may enhance synaptic plasticity in neurons.
04 From Papers to Clinical Answers
Linda:
Research in this field progresses quickly, so why hasn't clinical translation been ideal?
Ximing Wang:
The reason it's not ideal is that we believe this is a systematic engineering problem. Take depression as an example — it involves diagnostic accuracy, degree of personalized treatment, device performance itself, execution precision and efficiency, and prognosis management. If any link has obvious shortcomings, the final treatment effect will suffer greatly.**
Most pure technology perspectives may focus more on major breakthroughs at a single point, like improving efficiency at one technical point, but overlook the systematic engineering problem.
So in our team, clinical is also a very important R&D department, not just for running clinical trials. It's part of comprehensive capabilities. Many teams domestically and internationally still approach this in a point-based way trying to solve problems. From the start, we hoped to spread more evenly. For example, early in the company's founding, we invested in simulation capabilities because they provide great help for overall system performance.
Linda:
Your first clinical patient gave the team a lollipop on day five of treatment. Can you tell that story?
Ximing Wang:
This has always been a case we're very proud of. The team spent a long time on R&D. Although we had confidence in the product itself, before actually treating the first patient, everyone was actually nervous. The first non-patient was myself — I completed the entire treatment, so I clearly knew what the experience was like.
The first depression patient showed no visible change in the first two days. When she arrived, she still seemed quite severe, unwilling to communicate with others. After treatment, she'd hide in a corner, not speaking. When the hospital offered water, she'd very politely decline — overall a quite closed-off state.
But on day four, she suddenly washed her hair, combed it, and put on lipstick to come for treatment. The entire treatment room was shocked. Because past therapies usually take 6 to 8 weeks, but she showed obvious changes on day four.
We asked further and found she had also cleaned up her home, which she hadn't touched in half a year, and had specially dressed up and changed clothes to come for treatment. This case was extremely motivating for the team. Whether colleagues working on electronics or algorithms, everyone was very energized.
05 The Product Power of Brain-Computer Interfaces
Linda:
After devices enter hospitals, how do doctors view your product?
Ximing Wang:
Doctors feel it's very satisfying to use.
This is actually what we spent the most time discussing when defining the product. In traditional industries, others also make products, but often there are problems of thinking about one thing but not two, or two but not three.
What we hope to achieve: first, patients feel the efficiency is sufficient; second, doctors also find it easy to use. If patient outcomes are good but each use requires several people spending hours, then scalability and recognition face major challenges.
So we try to make the product as optimal as possible.
Linda:
Hospitals should care a lot about safety. How do you convince them?
Ximing Wang:
The bottom line of medical devices is safety. Many colleagues on our team come from the surgical robotics industry and other medical device industries, with sufficiently high standards to begin with. We do extensive rigorous testing and validation to ensure product safety.
Only the second step is clinically validating effectiveness. Safety is an indispensable condition in the medical device environment.
The same goes for sleep devices. Many people overlook the safety of sleep intervention itself. For example, intervening at wrong times could actually cause disruption. Some teams may not recognize this problem and act slightly more aggressively. We tend to first ensure sufficient safety, then discuss how to achieve extreme effectiveness.
Linda:
You just mentioned sleep products. What kind of product is it?
Ximing Wang:
The sleep product is something our entire team really likes. Internal testing is complete, and results are quite good.
Our path is completely different from what's seen on the market. It's a product targeting sleep, analyzing EEG and then using electrical intervention to improve overall sleep quality. The entire experience is very imperceptible — wearing it all night is like wearing an eye mask.
Many people hearing "electrical stimulation" might wonder if it stings, but it's actually not like that.
Linda:
How is this different from sleep monitoring technology in hospitals now, or some consumer-grade sleep products on the market?
Ximing Wang:
It's actually completely different. What we do relates to the deep brain, and has much to do with deep brain functional divisions. Each brain region's mechanism is different; although they may seem connected, internal division of labor is actually very clear.
Take sleep as an example. We combine real-time EEG feedback with dynamic parameter adjustment in electrical stimulation regulation. You can understand it as not just choosing when to stimulate, but also intensity, frequency, and parameters — there's a lot of customization inside. So it's actually quite "medical," a different concept from common solutions in the consumer market.
What we do is targeting specific deep brain nuclei and specific waves during particular sleep stages, using induction to improve sleep. Many products on the market are more about relaxation, soothing, or showing users some EEG fluctuation charts — they're not solving problems at the same level as what we're addressing.
06 From Electrical Stimulation to Ultrasound, Going Deeper
Linda:
You have both hospital-facing medical-grade products and home/personal-facing consumer-grade products. How consistent is the underlying technology between them?
Ximing Wang:
We currently have three products: one already capable of clinical deployment and providing solutions at the hospital end; one is the next-generation ultrasound brain-computer interface technology we're optimistic about; and one is a home-facing product combining EEG and electrical stimulation.
They share several common points.
First, understanding of the brain itself is an important prerequisite for all development. You need to know exactly what to treat and where to target before doing subsequent series of things. Otherwise it's like building a big hammer and banging everywhere without knowing what problem to solve.
Our development logic is the reverse: first being very certain that a certain direction can work out, then systematically making it into a product.
Second, hardware, software, algorithms, and system delivery capabilities are interconnected. For example, how to design better electronics, how to design better structures, how to design interaction between doctors and devices, how to do simulation algorithms beneath the skull. Light, sound, electricity, and magnetism have different paths, but share commonalities in simulation and system design.
So experience from the first product greatly helps the second and third products. Because from the start, we viewed every technical path and every product with systematic thinking.
Linda:
You're also advancing ultrasound brain-computer interfaces. Most people's understanding of ultrasound comes from B-ultrasound in hospitals. How can this become a tool for treating brain nerves and cognitive disorders?
Ximing Wang:
Ultrasound is a very interesting energy. It can both image and regulate, and can even do imaging and regulation simultaneously. This is also a very important characteristic of it as a brain-computer interface technical path.
In application, ultrasound broadly involves mechanical effects and thermal effects. Thermal effects can be understood as energy accumulation over a certain time, while what we mainly utilize now is its mechanical effects, then doing imaging and regulation.
Ultrasound waves cause cells to undergo periodic compression and stretching, driving cell membrane deformation, possibly also pulling cytoskeleton, enhancing cytoplasmic flow. More importantly, it may open mechanically sensitive ion channels on cell membranes. That is, when ultrasound produces mechanical force on cells, such ion channels open, further forming some mechanism mechanisms in neuromodulation.
Ultrasound can also combine with many things — microbubbles, modified proteins, or some particles. This involves cavitation effects. For example, inertial cavitation instantly generates local pressure changes, which is a key effect in blood-brain barrier opening and drug delivery.
If regulation solves "how to influence the brain," then imaging solves "how to see the brain." As an imaging tool, ultrasound shares commonalities with the familiar B-ultrasound. B-ultrasound is good for viewing soft tissue, with high repeatability and relatively simple operation. But applied to the brain, the biggest difficulty is the skull.
Linda:
Why does ultrasound defocus when penetrating the skull? Was this a previous technical difficulty?
Ximing Wang:
Yes. The skull is inherently a protective structure; it doesn't easily let various external energies enter the body. When ultrasound waves pass through the skull, relatively unpredictable deflection, defocusing, and deformation occur. Actually, it's not just ultrasound — light, sound, electricity, and magnetism all have certain unpredictability when passing through the skull. For example, light has diffraction, scattering, and deflection; other energy forms encounter similar problems.
But with technological advances, such as CT imaging getting better and better, we can first use CT to very finely reconstruct skull structure, then do physics-based simulation based on this structure. This makes it possible for ultrasound energy from different directions to still precisely focus on the same point after passing through the skull.
Behind this involves many technology stacks: transducer design, drive circuits, matching circuit control, medical imaging understanding, physics simulation, and tuning between simulation results and the real world. Why we can go very deep and achieve very small focusing is essentially the result of multiple technologies working together.
Our current path is to start with focusing, then gradually build up imaging capabilities. Ultimately, we hope to achieve real-time imaging and regulation, forming a closed-loop ultrasound brain-computer interface architecture.
Linda:
Medical and consumer are completely different R&D and commercialization rhythms. How do you balance them? Will the consumer end feed back to the medical end?
Ximing Wang:
We believe consumer will definitely feed back to medical. Consumer-side reputation and recognition will feed back to medical-side recognition. But this also places very high demands on team organization.
Linda:
Thank you Ximing for today's very systematic sharing and popular science. One deep impression after listening to this conversation is that Ximing isn't talking about a product feature, but about a direction he truly believes will happen.
I can also feel that Kongshanci is a team with strong empathy. They truly understand the troubles people are experiencing, and are willing to use technology to respond to these real needs. Such a team is vibrant, and has conviction and pursuit. On behalf of Yunqi, I sincerely wish Ximing's Kongshanci team continued progress in pushing forward this long-term and important endeavor.
Finally, to reintroduce: Kongshanci is a project invested in by the Yunqi-SJTU AI Fund. This is an early-stage frontier technology fund we launched with Shanghai Jiao Tong University in 2025, focusing on artificial intelligence, embodied intelligence, brain-computer interfaces, nuclear fusion, AI for Science, and other emerging and future industries.
We pay attention to these fields not because they're already ready for explosive growth today, but because we believe that technologies that truly change human lifestyles often begin quietly accumulating and growing before they become mainstream.
Subscribe to the Xiaoyuzhou "Attent!on" channel to listen to the full episode.
