⛵Dhow Dispatch | Dr. Aadeel Akhtar Spent a Decade Building Hands for Humans. Then the Robots Came.

In partnership with

Download Dhow - the community-powered private markets platform helping Muslim investors discover opportunities, learn, connect, and back Muslim founders building the future.

Our $10,000 allocation spots are fully filled, and Dhow Horizon Fund I has now made its first investment in Lore, an AI-powered entertainment intelligence platform. The minimum commitment for new investors is now $25,000, and access will become more limited as we continue deploying the fund. Reserve your allocation here.

🚀 | TL;DR

Dr. Aadeel Akhtar traces his interest in prosthetics to age seven, when he met a girl with a limb difference while visiting Pakistan. He eventually built an academic path across biology, computer science, electrical engineering, and neuroscience around the problem.

In 2014, Juan Suquillo in Ecuador tested an early 3D-printed prosthetic prototype. Suquillo made a pinch with his left hand for the first time in decades. Akhtar has repeatedly described that moment as the point when he decided the technology needed to leave academia.

PSYONIC was incorporated in 2015. After years of product iteration, the Ability Hand launched nationwide in 2021 with independently moving fingers, impact-resistant compliant mechanics, pressure sensing, haptic feedback, and a closing time of around 200 milliseconds.

Akhtar appeared on Shark Tank in February 2024, asking for $1 million for 2% of the company. The televised negotiation ended with Lori Greiner, Daymond John, and Kevin O'Leary offering $1 million for 6%.

The company has since expanded far beyond prosthetics. Arm says 300+ prosthetic users and 90+ robotics organizations use PSYONIC technology, while ABB, NVIDIA, and Universal Robots have publicly announced work involving the Ability Hand.

The larger question is whether PSYONIC can turn years of human prosthetic use into an advantage in physical AI, where dexterous manipulation and high-quality real-world interaction data have become major bottlenecks.

🇵🇰 | The Story Starts in Pakistan

Aadeel Akhtar has told the origin of PSYONIC many times, and the story begins long before the company existed.

His parents are originally from Pakistan, while he was born and raised in the United States. During a family trip to Karachi in 1994, when he was seven, he saw a girl around his age who was missing her right leg and using a tree branch as a crutch.

What stayed with him was the gap between their lives.

They shared the same ethnic heritage, yet access to resources and healthcare had produced completely different possibilities. In interviews years later, Akhtar described that encounter as the moment that pushed him toward prosthetics and toward the idea that advanced limbs had to become more accessible.

For a long time, he planned to go into medicine.

He enrolled at Loyola University Chicago as a biology student in 2004 and originally expected to become a physician working with people with limb differences. According to Loyola, he had been valedictorian of his high school and moved quickly through college, completing his biology degree in three years.

Then a computer science class changed the shape of the plan.

During his sophomore year, he took an introductory object-oriented programming course and became interested in the possibility of combining medicine with software and engineering. Instead of choosing between the two, he began stacking disciplines around the same problem.

He earned a B.S. in Biology in 2007 and an M.S. in Computer Science in 2008 from Loyola. He then entered the Medical Scholars Program at the University of Illinois Urbana-Champaign, where he pursued advanced training in engineering, neuroscience, and medicine.

He later earned an M.S. in Electrical and Computer Engineering and a Ph.D. in Neuroscience.

The degree list makes a lot more sense once you look at what he was trying to build.

A useful bionic hand sits at the intersection of biology, neural control, motors, embedded computing, materials, software, ergonomics, and clinical care.

Akhtar was effectively building an education around one object.

🧠 | Building a Career Around the Human Hand

At Illinois, the work became much more specific.

Akhtar joined research efforts focused on upper-limb prosthetics and the problem of making artificial hands easier to control and more informative to the person wearing them.

Myoelectric prostheses use electrical activity generated by muscles in the residual limb.

Electrodes detect that activity, software interprets the signal, and motors in the prosthesis respond.

That can give a user control over opening, closing, or changing grips.

The harder problem is feedback.

Biological hands operate as a closed loop. We move our fingers and continuously receive information through touch and proprioception. We know when we are touching something, how firmly we are holding it, and roughly where our fingers are without staring at them.

Most prosthetic systems historically handled the command side much better than the feedback side.

The Illinois research group used surface electromyography, pattern recognition, vibrotactile feedback, electrotactile feedback, and skin-stretch feedback to study how users could both command a prosthesis and regain some sense of what it was doing.

The work connected Akhtar with collaborators including Levi Hargrove at the Rehabilitation Institute of Chicago, John Rogers' group on flexible electronics, and the Range of Motion Project, a nonprofit providing prosthetic care in underserved communities.

This period shaped the company in ways that matter today.

PSYONIC grew out of years spent testing how people actually interact with artificial limbs, where current devices fail, and what users care about after the novelty of the technology wears off.

Akhtar has said early customer and clinician interviews were especially important.

One complaint kept surfacing: expensive multi-articulated hands could break during ordinary life.

A rigid finger could hit the side of a table, snap, and send the user into a warranty or repair process.

That is a very different design constraint from making a robotic hand perform an impressive laboratory demo.

The product therefore had to become fast, light, repairable, resistant to impact, useful across common grip patterns, and compatible with clinical systems prosthetists already used.

The human body set the form factor.

Everyday life set the durability test.

🌎 | The Prototype That Changed the Plan

In the summer of 2014, Akhtar traveled to Quito, Ecuador, with the Range of Motion Project to test an early bionic hand on Juan Suquillo, a former Ecuadorian soldier who had lost his left hand decades earlier.

The prototype looked nothing like a finished medical device.

Akhtar has described it as roughly three times the size of a normal hand, with wires running into breadboards, power supplies, computers, and the wall.

Contemporary reporting from Ecuador described a myoelectric system that used muscle signals from the arm to control the prosthesis.

Suquillo was fitted with the device and eventually pinched the artificial thumb and index finger together.

Loyola's later account records his reaction:

Akhtar has said Suquillo had gone so long without the hand that he had to relearn the motion of pinching.

The commercial implication hit Akhtar quickly.

A successful academic prototype could become a paper, a conference presentation, or a line on a CV.

A product could reach people.

Akhtar returned to Illinois convinced that the technology had to be commercialized.

PSYONIC followed in November 2015.

The company won the University of Illinois Cozad New Venture Competition that year, and Akhtar later won the Illinois Innovation Prize. Public records and company materials also show early support through National Science Foundation grants.

The decision involved a major personal tradeoff.

Akhtar remained inside a medical training path and had already invested years in becoming a physician-scientist. In 2017, after finishing his Ph.D. and the first year of medical school, he took a leave to work on PSYONIC full time.

That move turned a research project into a manufacturing problem.

The question became whether PSYONIC could repeatedly build a hand that clinicians could fit, patients could trust, insurers could reimburse, and engineers could improve without making it too expensive to reach the people it was originally designed for.

🦾 | Building the Ability Hand

PSYONIC went through eight major versions before announcing a nationwide Ability Hand launch for September 2021.

That iterative history matters because the final product reflects compromises that only appear once hardware leaves the lab.

The current public product page lists the Ability Hand at roughly 490 grams.

All five fingers can flex and extend, while the thumb can rotate electrically and manually.

The hand supports multiple grip patterns, connects over Bluetooth, charges over USB-C, and carries an IP64 rating for dust and splash resistance.

Older technical literature from PSYONIC gives a useful view into the mechanics.

A 2022 paper described six degrees of freedom driven by brushless motors with field-oriented control.

The fingers could close 90 degrees in roughly 200 milliseconds.

Pressure sensing across the fingertips, finger pads, and lateral surfaces fed information into a vibration unit, creating a simple form of touch feedback for the user.

The practical value is easy to understand.

Most people can pick up a paper cup without consciously thinking about grip force.

A prosthetic user may have to rely heavily on vision to estimate whether the hand is touching an object and how hard it is squeezing.

PSYONIC's contact sensing gives the user another channel.

In small studies published by the company's researchers, subjects were more successful at handling fragile cups and hollow eggshells when touch feedback and contact reflexes were enabled.

Durability became another core design choice.

PSYONIC patented compliant linkage mechanisms that allow a finger to flex under impact instead of transferring the entire load into rigid plastic joints.

Public patent records also show work around sensored brushless motors, tendon-driven systems, and the broader prosthetic-hand architecture.

The clinical workflow matters as much as the mechanics.

The Ability Hand is not a consumer gadget bought from a normal checkout page.

A prosthetist evaluates the residual limb, creates the socket and suspension system, positions electrodes, configures controls, and trains the user.

The hand integrates with established prosthetic control systems rather than requiring PSYONIC to replace the entire clinical stack.

The company also designed around reimbursement.

PSYONIC describes the Ability Hand as an FDA-registered Class I medical device and says it is covered by Medicare.

That language matters.

FDA registration and classification are different from a premarket FDA approval. Most Class I devices are exempt from the 510(k) premarket-notification process while remaining subject to applicable regulatory requirements.

The affordability story also deserves context.

On Shark Tank in 2024, Akhtar said the Ability Hand sold for roughly $15,500 and cost about $1,800 to manufacture at the time.

Comparable high-end prosthetic systems can involve much larger total equipment and clinical costs.

The amount ultimately paid by a patient depends on the broader prosthetic system, fitting, insurance, and reimbursement.

The deeper product story sits beyond the specs.

The Ability Hand was built to survive normal human life.

It had to deal with table edges, water, charging cycles, fragile objects, repeated use, and the simple reality that a device attached to someone's body has to be comfortable enough to wear every day.

Those constraints later became surprisingly relevant to robotics.

🦈 | A $50 Million Ask on Shark Tank

On February 23, 2024, PSYONIC appeared on Season 15, Episode 15 of ABC's Shark Tank.

Akhtar entered with retired Army Sgt. Garrett Anderson, an Ability Hand user who had lost his right forearm while serving in Iraq.

Akhtar asked for:

$1 million for 2% of PSYONIC.

That implied a $50 million valuation.

It was an aggressive ask for a hardware company that, by Akhtar's own on-air numbers, had generated roughly $2 million of lifetime sales at the time of filming.

The pitch became a useful public snapshot of the company at that stage.

Akhtar said the hand sold for about $15,500, cost approximately $1,800 to manufacture, and had more demand than PSYONIC could satisfy.

Production capacity was around 100 hands per year.

His goal was to scale toward 500 and eventually 1,000.

The Sharks focused heavily on fundraising.

Akhtar explained that the company had already raised millions through equity and grants and had also turned to equity crowdfunding through StartEngine.

Mark Cuban questioned why a company with strong demand and ambitious technology had not attracted enough traditional venture capital to eliminate the need for crowdfunding.

Robert Herjavec raised a similar concern.

Akhtar's answer revealed an interesting problem.

Healthcare investors often looked at the robotics side as outside their mandate.

Robotics investors could see the prosthetics market as too specialized.

PSYONIC sat across categories that were usually financed and evaluated separately.

Kevin O'Leary first countered with $1 million for 10%.

Akhtar pushed back on dilution.

Kevin O’Leary eventually partnered with Lori Greiner and Daymond John on a combined $1 million offer for 6% of PSYONIC. The equity was split evenly, with each Shark receiving 2% through a combination of common and advisory shares. Akhtar accepted the offer on air. The structure mattered because it allowed PSYONIC to bring in three Sharks while preserving the economics around the company’s stated valuation more carefully than a straightforward 6% common-equity sale would suggest.

Akhtar accepted on air.

PSYONIC later explained publicly that the contemplated structure involved a mix of common and advisory shares designed to preserve the company's stated valuation framework.

As with every Shark Tank pitch, the televised handshake was subject to diligence after filming.

Public secondary reporting later indicated that the deal appears to have closed, but a clean primary-source announcement documenting every final term is difficult to find.

For that reason, the safest description is the one everyone saw:

A $1 million on-air offer for 6% (split 2% each via a mix of common and advisory shares).

The appearance did something more important than settle a valuation debate.

It put a technical prosthetics company in front of a mainstream audience.

And by then, another customer was already starting to emerge.

🤖 | Then the Robots Came

Robotics researchers began buying the Ability Hand because they were struggling with a problem that looked very familiar to a prosthetics engineer.

Industrial automation has traditionally avoided general-purpose hands whenever possible.

A factory can use a two-finger gripper, suction cup, magnetic end effector, or task-specific tool because the environment is structured and the object is known in advance.

That works extremely well for repetitive tasks.

General-purpose robots face a harder requirement.

A robot moving through a home, hospital, warehouse, or flexible manufacturing environment may need to open a drawer, pick up a deformable bag, turn a knob, handle a tool, grasp an irregular object, or change grip halfway through a task.

The environment stops adapting to the robot.

The robot has to adapt to the environment.

That makes the hand one of the hardest components in physical AI.

Human hands pack more than twenty degrees of freedom, dense sensing, compliance, high strength-to-weight, and extraordinary control into a compact form.

Robotic systems can approximate pieces of that performance.

Doing it reliably, cheaply, and at scale remains difficult.

PSYONIC had already spent years optimizing around many of those exact characteristics.

The hand needed to be lightweight because a person had to wear it.

It needed to survive impacts because people hit things.

It needed tactile sensing because users could not rely entirely on vision.

It needed compact motors and onboard control because there was very little room available.

In 2022, PSYONIC publicly released a research version of the Ability Hand.

The company said early research users included Meta, NASA, Apptronik, and Sanctuary AI, along with academic laboratories.

The research API exposed torque, velocity, and position control across six brushless motors and streamed encoder and touch-sensor values over Bluetooth or USB.

The robotics footprint has grown since then.

Arm now says more than 90 robotics organizations use PSYONIC technology, alongside more than 300 prosthetic users.

Arm's case study describes 12 Arm-based microcontrollers coordinating sensing, actuation, and feedback inside the platform.

In March 2026, PSYONIC announced that the Ability Hand had become a native asset in NVIDIA Isaac Lab, NVIDIA's open-source framework for robot learning.

NVIDIA's own Isaac Sim documentation lists multiple PSYONIC hand assets with left/right and size variants.

In June 2026, ABB Robotics announced a collaboration that mounts the Ability Hand on its GoFa collaborative robot and explores dexterous handling using touch, compliant mechanics, and human-generated training data.

Then in September 2026, PSYONIC and Universal Robots announced an integration for UR's Gen 7 platform.

PSYONIC says the Ability Hand can be programmed through PolyScope X using a URCap integration, bringing grip and position control into the same interface used to program the robot.

The shift expands the market without requiring a completely different piece of hardware.

A hand that works on a person can also function as the end effector on an industrial arm, mobile robot, research platform, or humanoid.

PSYONIC did not have to invent a second product category from scratch.

Robotics discovered value in the product that already existed.

🧬 | The Data Hidden Inside a Bionic Hand

The hardware opportunity is easy to see.

The data opportunity is more subtle.

Large language models could train on enormous collections of text.

Vision systems could train on billions of images and videos.

Robots need data about physical interaction, and much of the most valuable information never appears clearly in a video.

Imagine watching someone pick up a paper cup.

The video shows the trajectory of the hand.

It may not reveal which fingertip made contact first, how pressure changed as the cup began to deform, how much torque was applied at each joint, or the tiny adjustments that kept the cup from slipping.

Robotics companies solve this in several ways.

They generate synthetic interaction inside simulation.

They teleoperate robots and record the result.

They use motion-capture gloves.

They collect demonstrations from humans.

Increasingly, they combine several approaches.

PSYONIC has an unusual starting point because a sensorized robotic hand is already being worn by people.

When the same hardware architecture exists on both the human and robot side, the representation gap can become smaller.

PSYONIC calls its approach “real-to-real” transfer.

The company's public NVIDIA announcement describes a workflow in which a person uses the Ability Hand to perform a physical task, producing interaction data from the motors and sensors.

Researchers can use the hand as a simulated asset inside Isaac Lab, train or refine a policy, and then deploy the policy back onto physical robotics hardware using the same end effector.

PSYONIC demonstrated the idea with a pipetting task at NVIDIA GTC 2026.

A human user wearing the Ability Hand manipulated a pipette and dispensed liquid.

The same type of hand was then shown on different robotic systems performing the task.

The demonstration is illustrative rather than a peer-reviewed benchmark, but it shows the architecture the company is pursuing.

ABB's collaboration gives the thesis more external weight.

ABB explicitly says it is exploring whether touch and motion data generated through human prosthetic use can help train robots for delicate, variable tasks that remain difficult to automate.

That creates the most important unanswered question around PSYONIC's robotics strategy:

Does human-generated Ability Hand data measurably improve robot learning?

The public record does not yet provide a clean independent benchmark comparing policies trained with PSYONIC human data against equivalent policies trained without it.

That benchmark matters.

A proprietary dataset becomes valuable when it is hard to reproduce and materially improves outcomes.

Human prosthetic data could become powerful because it comes from uncontrolled real-world environments and because the hand contains tactile sensors.

Other approaches may close some of that gap through simulation, teleoperation, vision-based imitation, or purpose-built sensor gloves.

The market is already testing that question.

Proception pairs a 22-degree-of-freedom robotic hand with a sensorized glove designed specifically to collect human manipulation data.

Genesis AI has also built human-like robotic hardware and collects data from sensor-equipped gloves.

Linkerbot says its LinkerSkillNet platform contains hundreds of dexterous manipulation skills.

Multiple teams are betting that the future of robotics depends on solving hands + data together.

PSYONIC enters that race with a source of human interaction that began as a medical product rather than a data-collection device.

📈 | A Real Business Entering a Much Bigger Race

PSYONIC is still an early hardware company, but its public SEC filings show that it has moved beyond the research-project stage.

The company reported:

2023 revenue: $1.90M

2024 revenue: $3.33M

2025 revenue: $5.35M

That works out to roughly 76% revenue growth from 2023 to 2024 and another 61% from 2024 to 2025.

What began as a relatively narrow prosthetics business now overlaps with a heavily funded robotics category.

In China, Linkerbot completed a financing in 2026 at a reported $3 billion valuation and told Reuters it would target $6 billion in a future round.

Reuters reported that the company was producing close to 5,000 dexterous hands per month and expected to reach roughly 10,000 per month.

Linkerbot also claims a large library of real-world dexterous skills.

In California, Proception raised an $11 million seed round led by First Round Capital with participation from Y Combinator and BoxGroup.

Its Proception Hand has 22 degrees of freedom, while the matching Proception Glove captures human manipulation data using a sensing architecture designed to transfer behavior into the robotic hand.

Genesis AI launched in 2025 with a $105 million seed round co-led by Eclipse and Khosla Ventures to build foundation models for robotics.

The company has also developed human-like robotic hardware and a system for collecting human manipulation data.

Then there are vertically integrated humanoid companies such as Tesla, Figure, Apptronik, 1X, and others that have strong incentives to develop their own hands.

The hand is strategically important enough that some of the largest robotics companies may prefer to own the full stack.

That leaves PSYONIC with several possible businesses inside the same product.

It can remain a supplier of dexterous end effectors.

It can deepen integrations with industrial and research platforms.

It can build software around manipulation.

It can create proprietary datasets.

Or it can combine those pieces into a platform that becomes increasingly difficult to replace.

The public evidence strongly supports the hardware and integration business today.

The data and software layers remain earlier.

That distinction matters for investors because hardware adoption does not automatically translate into software economics or a defensible data moat.

The company still has an unusual asset:

years of real-world field use with human beings.

Prosthetic users impose a level of durability, ergonomics, and unstructured-environment testing that most robotics products do not receive.

If that experience compounds into better hardware and better data, the path becomes more interesting than either prosthetics or robotics viewed separately.

🧭 | The Dhow Perspective

Dr. Aadeel Akhtar started with a human problem.

He spent years learning how people control prosthetic limbs, how those devices fail, what patients need from them, and what it takes to move a medical device from a research setting into everyday life.

That work created technical assets that became valuable in an entirely different market.

As AI moves further into the physical world, the bottlenecks change.

Model quality matters. Compute matters. Training data matters.

Eventually a robot has to touch something.

It has to pick up the cup, turn the handle, hold the tool, and use the object without crushing it.

PSYONIC has spent more than a decade working on that final interface between intelligence and the physical world.

The opportunity now is much larger than the one Akhtar originally set out to address.

The challenge is equally larger.

The next few years will show whether PSYONIC remains one of many dexterous-hand suppliers or whether its unusual path through prosthetics gives it a durable advantage in physical AI.

Some of the most interesting startup opportunities emerge when years of accumulated expertise collide with a market that suddenly becomes much larger.

✅ | Bottom Line

Aadeel Akhtar met a girl with a limb difference in Pakistan when he was seven.

The experience eventually shaped an academic path through biology, software, electrical engineering, neuroscience, and prosthetics research.

An oversized prototype in Ecuador turned the research into a company.

Years of iteration turned the prototype into the Ability Hand.

Prosthetic users forced PSYONIC to learn about durability, touch, speed, weight, repair, reimbursement, and the messy reality of using advanced hardware every day.

Those same lessons are now relevant to robotics.

ABB is testing the hand on industrial cobots.

NVIDIA has integrated PSYONIC assets into Isaac Lab.

Universal Robots has built the hand into its Gen 7 ecosystem.

Arm says more than 90 robotics organizations use the technology.

The next phase comes down to proof.

PSYONIC has to show that the same human use that shaped its hardware can generate data that meaningfully improves robotic manipulation.

It also has to keep advancing while competitors bring far more capital and manufacturing scale into dexterous hands.

That is what makes this story worth following.

Aadeel Akhtar spent more than a decade building better hands for humans.

Then the robots came.

Interested in investing alongside us?

At Dhow, we pay close attention to founders who have spent years building real technical depth before the market fully catches up to them. Aadeel Akhtar’s story is a good example: more than a decade of work on prosthetics has now placed PSYONIC inside one of the most important emerging problems in physical AI. As robotics moves from controlled environments into the real world, dexterity, sensing, and manipulation data are becoming increasingly valuable. The companies that solve those bottlenecks could become foundational to the next generation of intelligent machines. Join the movement, share this with a friend (or two).

Speak naturally. Send without fixing.

Wispr Flow turns your voice into clean, professional text you can send the moment you stop talking. No rough transcription you have to clean up. Actual polished text, ready for email, Slack, or any app.

Speak the way you think. Go on tangents. Change your mind mid-sentence. Flow strips the filler, fixes the grammar, and gives you text that reads like you spent five minutes writing it.

Start flowing free

If you enjoyed this, you’d be doing us a HUGE FAVOR by just clicking the ad above & below - we appreciate it!

The agentic era needs a different CRM. That’s Attio.

Parallel, Turbopuffer, and Wordsmith run their entire GTM motion on Attio, with agents that chase every buying signal, build pipeline, and move deals forward, 24/7.

Try Attio for free

Sources:

1. Loyola University Chicago: Aadeel Akhtar founder profile and PSYONIC history.

2. University of Illinois: Akhtar's prosthetic-control and sensory-feedback research.

3. PSYONIC: Company history and Ability Hand product specifications.

4. MEC Proceedings: Technical research on touch feedback and contact reflexes using the Ability Hand.

5. Google Patents: PSYONIC patents covering compliant finger mechanisms, sensored motors, and prosthetic systems.

6. ABC / Shark Tank: Season 15, Episode 15.

7. CNBC: Reporting on Akhtar's Shark Tank pitch and on-air negotiation.

8. Arm: PSYONIC Physical AI case study.

9. NVIDIA Isaac Sim: PSYONIC Ability Hand robot assets.

10. ABB Robotics: PSYONIC collaboration around human-generated data and robotic dexterity.

11. Universal Robots: PSYONIC Gen 7 integration.

12. SEC: PSYONIC Form C-AR annual filings.

13. Reuters: Linkerbot funding, valuation, and manufacturing scale.

14. Proception: Proception Hand and Glove.

15. TechCrunch: Genesis AI funding and physical-AI strategy.