Paradromics puts its Connexus brain implant into a speech-restoration patient

Why it matters

Paradromics is no longer only proving that Connexus can record brain signals in humans. It is testing whether a fully implantable BCI can become a long-term communication product for patients who cannot speak.

Paradromics founder and CEO Matt Angle has put the company's Connexus brain-computer interface into its first long-term speech-restoration patient, a Michigan woman with a motor neuron disease who has difficulty speaking, Business Insider reported.

The procedure took place in early June 2026 at University of Michigan Health as part of an FDA-approved clinical study, according to Business Insider. The operation lasted about four hours and was led by Dr. Matthew Willsey, a University of Michigan neurosurgeon. The patient, whose name has not been disclosed for privacy reasons, is now home recovering, Angle told the outlet.

That makes the milestone current, but it also needs precision. Paradromics previously completed a first-in-human Connexus recording in 2025 during an acute procedure at the University of Michigan, according to the company's announcement. The new step is different: Connexus is now implanted in a patient for an FDA-approved, long-term study aimed at restoring communication.

Angle has spent a decade pushing Paradromics toward this clinical moment. The near-term bet is narrower and medical: prove that a fully implantable BCI can read speech-related brain signals safely and clearly enough to give people with severe motor impairment another way to communicate.

The product is a speech system, not a biological cure

Connexus is not designed to repair a patient's vocal cords, mouth muscles, or nerves. The therapy Angle described to Business Insider is a decoding system. The patient attempts to speak. The implant records neural activity associated with speech. Software then translates those signals into text or synthesized speech on a computer.

"We will work with her in a training and rehabilitation environment to teach her how to speak through a computer so that she can attempt to say things," Angle told Business Insider. "The device can read out the representations of speech in her brain. And then our hope is that the computer will then be able to complete speech on her behalf."

The hardware is small but invasive. Angle told Business Insider the brain implant is about the size of a dime. It sits on the brain's surface and uses 421 platinum-iridium microwires that insert into brain tissue; each wire is less than half the diameter of a human hair. Willsey told the outlet that Connexus also includes extension leads running under the skin down the left side of the neck to a transceiver beneath the left clavicle. That chest device communicates wirelessly through the skin with an external receiver.

Paradromics' Connect-One clinical study page describes the same architecture in patient-facing terms: brain signals travel from the Connexus brain interface to a chest implant, and a receiver worn on the chest sends signals to a computer. The study page says Connexus is designed to turn signals from the brain area that controls speech movement into text, a digital voice, or commands for a phone or computer.

The study will test more than whether the device turns on

The first visible result will not be a product launch. It will be data. Paradromics said the Michigan patient will be evaluated over six years, Business Insider reported. Angle said the first year will focus on data collection, including safety, words per minute, vocabulary size, and how many bits of usable information the system can move during a conversation.

That last metric is the one Paradromics wants the market to watch. On its platform page, Paradromics says its current BCI platform has demonstrated an information transfer rate of 200 bits per second, a company claim based on preclinical work rather than this patient's performance. The clinical question is whether that engineering claim can survive contact with a real user, a real recovery timeline, and daily communication needs.

The Connect-One study is also recruiting under practical limits. Paradromics says potential participants may qualify if they are adults ages 22 to 75, have a progressive disease such as ALS or an injury such as a brainstem stroke that affects muscles or nerves, have trouble speaking, can consent and communicate well enough to share their experience, can read and understand English, have used assistive communication devices, and live within four hours of Ann Arbor, Michigan. The device remains investigational and is not for sale.

Neuralink set the public clock. Paradromics is choosing a different first proof point.

The unavoidable comparison is Neuralink, which put its first implant into Noland Arbaugh in January 2024 and made computer control the public demonstration of BCI progress. Paradromics is competing in the same intracortical BCI category, but the clinical emphasis here is speech.

That distinction is strategic. Cursor control is legible to the internet: a patient can play chess, browse, or post. Speech restoration is harder to demo cleanly and harder to measure with one clip. It requires vocabulary, rate, reliability, latency, training, and a patient interface that works when communication is not a stunt but a need. If Paradromics can produce meaningful speech performance, the result would give Angle a stronger medical wedge than attention alone.

The company's timing follows a regulatory sequence. Paradromics received FDA Investigational Device Exemption approval in November 2025 to begin the Connect-One Early Feasibility Study. The early June implant is the point where that paper clearance became a patient reality.

The enhancement question is already inside the medical one

Angle is careful to frame Paradromics around therapy. The first use case is communication for people with severe motor impairments. But the company also says its platform may eventually support direct AI interaction, advanced prosthetics, and treatments for mental health and other neurological conditions.

That is the tension every serious BCI founder now has to manage. The same interface that could help a patient speak through a computer could, over time, become a higher-bandwidth input layer for software, AI systems, prosthetics, or robotics. Angle did not duck that in his interview with Business Insider.

"Anyone who's building interfaces with the nervous system is building both the capability to restore capabilities and enhance capabilities," he said.

The more immediate risk is clinical, not philosophical. Brain implants involve surgery, anesthesia, infection risk, long-term device reliability, privacy questions, and dependency on hardware and software that must work when a patient needs it. Angle told Business Insider that any surgery involving general anesthesia or implantable devices carries "non-zero risk," including infection, and said Paradromics designed Connexus to last a long time to reduce the need for repeat surgeries.

The next few weeks should show whether Connexus can begin producing useful signals from the Michigan patient after recovery and training. The larger test will take longer: whether Angle's decade-old premise, that high-data-rate neural interfaces can become durable medical products, can clear the distance from a successful implant to a communication system a patient can actually rely on.