Magic AI's fitness mirror raises $11 million, and a coach with blind spots

A London mirror company just landed one of the year's biggest AI checks

Magic AI, a London startup, has raised $11 million (about £8 million) in a Series B round co-led by the venture firms Beringea and IW Capital, with backing from Active Partners, PXN Group, Accelerate Ventures, Doorway and Ventures Together. That brings the company's total funding to roughly $20 million since it launched in 2023, on top of earlier rounds and an angel round that reportedly included former England cricket captain Sir Alastair Cook and footballer Jesse Lingard. Founder and CEO Varun Bhanot called it one of the largest consumer AI deals in the UK this year, and the company says the money is earmarked almost entirely for scaling in the United States, where it already counts more than 18,000 members across all 50 states.

The product is a wall-mounted mirror with a built-in camera and screen that projects a holographic trainer over your reflection while you work out. Its ReflectAI software claims to track your movement, count reps and correct your form in real time, guiding you through more than 1,000 exercises across strength, HIIT, yoga, Pilates, dance and boxing. Bhanot has said his own motivation was blunt: he built the first prototype after losing a quarter of his body weight working with a personal trainer he could barely afford, and the pitch to investors leans on that same math. "The cost of personal training, which is what we replace, is too expensive," he said. In the US the mirror sells for $1,499, marked down from $1,999, with a Magic+ subscription running around $22 a month, plus optional dumbbells and a bench.

What the money is actually buying

Magic AI says revenue grew 342% over the past year, on prior-year sales of about £10.7 million, and the company has picked up real recognition along the way: a spot on TIME's Best Inventions list in 2024, a King's Award for Enterprise for innovation in 2026, and a ranking as the UK's fourth fastest-growing hardware company from The Sunday Times. It has also poached serious Peloton talent, hiring the company's former international managing director and Kevin Cornils, Peloton's former chief commercial officer, to run its American expansion. That is a specific and telling hire. Peloton spent years learning, expensively, what does and does not get Americans to keep using a piece of fitness hardware, and Magic AI is betting that knowledge transfers.

The awards are worth pausing on, the same way we've flagged with CES's health category before. A Best Inventions nod or a Sunday Times growth ranking is a signal that journalists and judges liked the pitch and the trajectory, not evidence that the underlying tech does what it claims for your knee, your spine or your form on a heavy squat.

What a camera in your living room can and can't see

The technical claim behind Magic AI, and every mirror like it, is that computer vision can watch your body the way a trainer would. It's a reasonable ambition, but the published research on AI-based posture and movement tracking is more cautious than the marketing. A peer-reviewed review of AI in postural management found that while pose estimation systems can identify anatomical landmarks with real precision in controlled settings, that accuracy degrades in the real world. Lighting, background clutter and camera angle all matter, loose clothing and body composition can throw off keypoint detection if the training data wasn't diverse enough, and when one part of your body blocks another from the camera's view, the system has to guess. The authors put it plainly: a system "can be technically precise but clinically irrelevant." That is the gap between counting reps accurately, which a mirror can probably do, and catching the subtle joint angle that turns a normal lunge into a knee injury three months later, which is a much harder problem.

The hardware graveyard this category needs to avoid

Magic AI is also stepping into a category with a genuinely bad recent track record. Lululemon bought the smart mirror pioneer Mirror for $500 million in 2020, at the height of pandemic home fitness demand, then wrote off $442.7 million of that value in 2022 and discontinued the hardware entirely, pushing remaining subscribers toward an app instead. None of that guarantees Magic AI's mirror meets the same fate. Its price point is lower, its software is newer, and it is chasing a US market Mirror never fully cracked. Still, it is the reason investors and buyers alike should treat "AI personal trainer" claims with some skepticism rather than take the marketing at face value.

Our take

An $11 million Series B, in a sector still living down a $500 million writeoff, is a genuinely useful signal that a smaller, cheaper AI mirror can find real customers if it prices and targets carefully. Judged as a cheaper, always available substitute for a generic gym session, Magic AI's pitch holds up reasonably well. Judged as a literal replacement for a trainer's eyes on your form, the science on computer vision still has real gaps, especially for the subtle deviations that actually cause injuries. Buy it, if you buy it, as an accountability tool you'll actually use, not as insurance against getting hurt.

Health Trend Wire holds no position in Magic AI and was not paid for this article. This is general information, not medical or financial advice.