Grace Brown's first robot arm moved on three hydraulic axes. She built it in a high-school engineering class, the kind of assignment that quietly reroutes a life. Brown already liked mathematics and physics. The arm made those abstractions physical. Arduino boards, MATLAB and code followed. So did a recognition that engineering could be less like solving for an answer in a textbook and more like pulling an object out of imagination.
She kept pulling. Brown built a 3D printer. Then an electric skateboard. At the University of Melbourne she studied mechatronics, the useful borderland where mechanical, electrical and software engineering overlap. She later joked that mechatronics is for the “indecisive engineer.” It was a broad education, and at times broader than she wanted. She wanted a workshop. She wanted to make things.
During Melbourne's pandemic lockdowns, that appetite met a more intimate design brief. Living in student accommodation, far from her family in Brisbane, Brown imagined a machine that could do two modest things: ask how her day was going and offer a hug. The first version of Abi was partly a technical workout and partly company. The prototype had none of the business plan that would later gather around it.
“At heart, I'm a problem solver and a builder.”Grace Brown, reflecting on why engineering fit
Carry the robot into the room
When restrictions eased, Brown began looking beyond her own experience. The question was no longer whether she could build a companion robot. It was whether anyone else would want one. She contacted 40 care homes around Melbourne. An early Abi traveled in a basket.
There is something clarifying about that image: a founder carrying an unfinished humanoid into the place where the premise will either hold or collapse. Hardware companies can hide for years behind development cycles. Brown chose exposure. The prototype did not need to look inevitable. It needed to produce honest reactions.
Those visits shaped an operating idea that still distinguishes Andromeda. A social robot cannot be evaluated only by its component list. Motors matter. Cameras matter. Models, memory and reliability matter. Yet a technically capable machine may still stand too close, interrupt at the wrong moment or project the wrong kind of presence. In a shared human environment, behavior becomes part of the infrastructure.
The field loop
Brown had found the beginnings of a company. In 2022, she co-founded Andromeda with Yan Chen, a university friend and electrical engineer. Startmate provided an early structure and investment. Galileo Ventures followed. An accelerator run by Amazon Web Services and MassRobotics brought the team to the United States. Brown also deferred her studies for a year without initially telling her parents. They thought she was finishing her degree while she was learning how to finance a robotics startup.
The concealment makes more sense in the context Brown has given it. Her mother, a Chinese immigrant to Australia, built small businesses of her own but wanted the security of a conventional job for her daughter. Brown had watched her improvise and persist. She absorbed the persistence, then applied it to a path that hardly resembled security.
A character with actuators
Abi looks unlike the metallic humanoids that dominate robotics demonstrations. She is bright, compact and deliberately character-like. The reference Brown returns to is Baymax, the inflatable companion in Big Hero 6. Pixar is another touchstone. Animators can make a pencil line seem shy, curious or delighted. Brown wants Andromeda to produce that emotional legibility with actuators, eyes, gestures and voice.
That framing changes the org chart. After Andromeda's 2025 financing, Brown said early San Francisco hires would include character animators with studio experience. The instinct is more practical than whimsical. An animator studies timing and anticipation, the weight of a glance, the difference between a movement that invites and one that startles. In social robotics, those choices influence whether a machine gets used.
Brown calls trust a North Star. It is easy to reduce that word to warmth and color, but her version has mechanical consequences. How should Abi enter a room? Where should she stand? What should happen after the model misunderstands a person? Does the machine recognize that someone is distracted and adjust? These are manners translated into product requirements.
Inside Andromeda, Brown's own role has traveled the same path from components toward systems. Early work included CAD drawings, motors, electronics and chatbot programming. As the company grew, her work became capital, recruiting, customers and coordination. She and Chen joke about the division: she envies the technical play in his job; he sees a CEO who spends her day talking.
Her generalist training remains useful. A mechatronics engineer learns to care about interfaces. A motor, a circuit and a controller can each work perfectly and still fail as a machine. Companies have the same problem. Brown's task is now to make engineering, animation, care-home operations, manufacturing and capital move as one system.
Learning the other half of the job
The transition did not arrive with a tidy promotion. In one early account, Brown remembered staying up all night watching Y Combinator videos, trying to learn the vocabulary of startup building. She understood the employability of technical skill as a kind of downside protection: if the company failed, she could find engineering work. That calculation gave her room to take the risk.
It also gave Andromeda a distinctive early rhythm. Before there was a full-time staff, the team met on Saturdays because that was when everyone was available. Brown split her own schedule between the physical build and an expanding orbit of investors, mentors, academics and prospective customers. In 2023 the company hired its first engineer. By late 2024 it had a warehouse, 15 staff, more than A$100,000 in revenue and paying customers.
Growth changed what good hiring meant. Brown has described early robotics employees as generalists who can cross boundaries and assume ownership. Later, when the systems become more sophisticated, specialists matter more. She has also talked about “anti-selling” roles: showing candidates the difficult, unvarnished version of startup work so that mission and expectations align before they join.
That directness appears in her public writing too. Brown speaks easily about doubt, exhaustion and the gap between an ambitious plan and the next solvable problem. But she rarely lowers the stated destination. The contrast is part of her operating personality: the target can be extreme while the work remains concrete. Today may involve a fundraising call. Tomorrow may involve a gesture that lands half a second too late.
The long road past the demo
By 2024, Andromeda had paying customers, a Melbourne warehouse and a growing team. Brown appeared on Forbes Australia's inaugural 30 Under 30 list, then on the 2025 Forbes 30 Under 30 Asia list for industry, manufacturing and energy. Recognition made the story easier to tell. It did not make robots easier to manufacture.
The more revealing milestone arrived in September 2025: a A$23 million Series A led by Forerunner Ventures, with Main Sequence, Rethink Impact and other investors participating. At the time, Andromeda was producing eight Abis per month and laying plans for much greater capacity. Brown moved to San Francisco, near investors and a deeper pool of software and animation talent, while manufacturing continued in Australia.
By mid-2026, she said the California team numbered just under 10. Melbourne remained important enough for Andromeda to move into a 2,500-square-metre facility. Dozens of robots were deployed, mostly in Australia, with the U.S. presence growing. The two-city shape is a compromise familiar to physical-technology companies: go where talent and capital cluster without abandoning the shop floor that taught the product how to exist.
“For us, it's not coming. It's already here.”Brown on the robotics industry's future tense
Brown is not coy about scale. She has spoken of a billion-dollar company, then raised the target to a trillion. She imagines companion robots in homes, schools and care settings, and has said she wants to see more Abis on the street than dogs being walked. Ambition at this size can sound weightless. Her more credible case is made in smaller units: a knock before entering, a well-timed pause, a face remembered, a machine that recovers gracefully from a mistake.
Those details point to the wager beneath Andromeda. Humanoids will not earn a place in everyday life simply by lifting more weight or walking faster. They will be judged by how they alter a room. Brown has spent a decade moving from hydraulic arms to that softer, harder frontier. The work still begins as it did in school: imagine an action, connect the systems, and see whether the object in front of you behaves the way a person hoped it would.