From Brain Implant to Lifelong Companion: Michel Maharbiz Unveils Epia Neuro

April 22, 2026 | Michel Maharbiz walked to the podium with an announcement and an apology last week. “This is our Hello World talk,” he told the audience Tuesday’s Bioelectronic Medicine Forum, the first public presentation of Epia Neuro, the stealth-mode neurotechnology company he co-founded with Gil Mandelbaum. After years of incessant investor pitching in private, Maharbiz said, this was the first time he’d be speaking about the venture openly. 

The apology: the talk would be “a little pitchy.” What followed was part company reveal, part field critique, part vision statement and a pitch for nothing less than a new paradigm for how brain-computer interfaces are built, deployed, and kept alive in patients’ bodies over a lifetime. 

Coming Out of Stealth 

Epia Neuro, Maharbiz revealed, has been operating quietly for roughly three years. It now employs 45 people and is backed by a combination of venture funding and pharmaceutical investment. The founding team includes Maharbiz, who until April 1st of this year served as full-time CEO of iota Biosciences, a neural interface company acquired by Astellas in 2020, and Gil Mandelbaum, who Maharbiz credited with patiently building the company under the radar. iota, within Astellas, remains independently active with clinical trials ongoing and a technology he described as “novel and working and incredible.” 

Epia’s advisory and scientific board includes Dr. David Lin, Dr. Mark Richardson, Prof. Bernardo Sabatini, and more from Massachusetts General Hospital, Lenox Hill Hospital, Beth Israel Deaconess, and elsewhere. The company, he emphasized, is “totally independent” from any corporate parent. 

The Patient at the Center: A Stroke Story 

Rather than leading with technology, Maharbiz opened with a hypothetical patient profile. Consider Sarah, he said, a 58-year-old architect who suffers a stroke. After the acute phase, she enters a six-to-nine-month rehabilitation window, but her improvement eventually plateaus. Whatever motor function remains at that point is, in large part, what she will live with for the rest of her life. 

Epia is focused specifically on upper-limb impairment following stroke, a condition that leaves patients with severely limited arm and hand control, profoundly disrupting daily life. Maharbiz estimated that roughly 120,000 patients per year in the United States enter the post-rehab plateau phase he described, with perhaps 60,000 immediately eligible for a device like the one Epia is building. “We built this company to solve this problem,” he said. 

The Device: Skull-Based, Outpatient, and Designed to Last 

The centerpiece of Epia’s technology is a skull-mounted neural interface — implanted in a one-hour procedure that does not pierce the dura mater (the protective membrane surrounding the brain). Maharbiz stressed that the system is designed from the ground up for scalability: it should be operable at any hospital, not just specialized centers of excellence, and is intended, eventually, to qualify as an outpatient procedure — a claim he acknowledged was bold but said was built into the device’s core specifications. 

The implant sits flush beneath the scalp, invisible once installed. It supports wireless read-write operation, has an “extremely long” designed lifetime, and recharges through a wearable headset. Crucially, it includes an optional deep brain stimulation (DBS) capability, though Maharbiz was careful to frame this as a secondary feature, not the anchor of the business proposition. 

During rehabilitation, patients pair the implant with a simple motorized glove, not an expensive, bulky exoskeleton, he said, but a lightweight, disposable-if-needed device meant to fit into real daily life.  

Intent that Originates in the Mind, Not Fingers 

The core algorithmic insight of Epia’s system, Maharbiz argued, is that the decoder is not trying to read individual finger movements or fine-grained motor commands. It is trying to detect intent, a higher-order signal that he said is both more achievable and more clinically powerful. 

“It’s not trying to decode your finger or anything. It’s trying to understand a rich context.” That intent signal then drives the glove’s actuators, abstracting the hard computational work to the software layer. The system trains on two-week loops over more than a year, continuously learning each patient’s unique neural signatures. 

This abstraction is deliberate and, Maharbiz suggested, strategically important. As AI capabilities improve over the next decade, a system anchored in intent detection can ride that wave without hardware replacement. “If you do the neural side right—that abstraction—then the system that wraps around it can basically just keep growing with the state of technology around the patient,” he said. 

Why Assisted Living, Not Rehab Improvement, Is the Business 

Maharbiz argued that anchoring the business model around improving rehabilitation outcomes is a recurring strategic mistake in the stroke neurotechnology field.  

“If your focus is on improving rehab outcomes, you have a really rocky road ahead of you,” he said. Study design becomes treacherous; enrollment criteria are poorly defined; clinical translation is uncertain. Companies that have tried this approach, he argued, have struggled not because their technology failed but because their value proposition was hard to prove at scale. 

Epia’s approach is different: the company’s anchor proposition is assistive function: a device that patients will use for the rest of their lives to help them grip, move, and navigate daily tasks. That creates a predictable, reimbursable product. DBS-based therapeutic intervention during the acute rehab window is available as an option, and Maharbiz said Epia’s study designs are built to evaluate it, but the business does not depend on it. 

“I can look a patient in the eye and say, ‘I don’t know how your recovery journey is going to turn out for the next six to nine months. But I can tell you that this glove will do what this glove does.’” 

The Bigger Vision: From Stroke to Cognitive Decline 

Maharbiz outlined the company’s longer-term ambitions beyond an assistive device for stroke patients.  

Imagine, he said, that Epia’s system works as intended, and Sarah, two years post-implant, is using her glove. The system has been continuously learning her intent signatures. Now, as a function of her age, not her stroke, she begins experiencing cognitive fog, working memory lapses, the early signs of neurodegenerative decline. 

A system that has spent two years learning to detect motor intent, Maharbiz argued, is positioned to detect anomalies in cognitive intent as well, deviations from established patterns that signal early cognitive dysfunction. And with that detection comes the infrastructure to deliver timely, low-cognitive-load nudges and interventions that research suggests can meaningfully slow or mitigate decline. 

“If I have a system that has lived with this patient, detecting intent for two years ... that same abstraction … can be deployed to detect issues with intent that have to do with cognitive decline.” 

The vision is a longitudinal brain interface that journeys with a patient across decades and disease states — starting with stroke and expanding, in time, to cognitive conditions. 

Timeline and Regulatory Path 

On near-term milestones, Maharbiz sketched a brisk schedule. First-in-human studies are expected to begin later this year at Lenox Hill Hospital in New York. FDA submission is targeted for February 2027, with stroke indication first, followed rapidly by the cognitive decline program. 

He also made a brief but notable aside when asked about the consumer path. If Epia achieves PMA approval and its cognitive claims are validated in study data, he observed, and the device costs around $5,000 and takes under an hour to implant—people will draw their own conclusions about how broadly it might one day be applied.  

What Differentiates Epia from Prior Attempts 

The audience had some questions when Maharbiz was done. When asked directly about the field’s previous efforts in stroke—including Northstar Neuroscience, which he said failed not because the technology didn’t work but because of a botched clinical trial on a public company timeline—Maharbiz resisted framing Epia as simply a more complete version of the same approach. 

“It’s not about scalability in the way you mean,” he said. “It’s something more profound.” Companies focused on improving rehab outcomes are, in his view, chasing a noble but commercially fragile goal. The heterogeneity of the stroke population, the difficulty of demonstrating improvement in a defined window, and the challenges of enrollment all compound. Epia, by anchoring first on a lifelong assistive proposition, sidesteps that trap entirely. 

Another question from the audience asked whether less invasive approaches might still capture enough intent signal. Maharbiz conceded the superiority of a wearable. “If a wearable or something like that can do 70% of what an implantable can do, the wearable’s going to win every time.” But he said he has not found that wearable. Until he does, the implant is the answer.