Flourish One is a $3,555 home robot you teach chores with your phone

Flourish, a startup operating across San Francisco and Paris, opened pre-orders for its first 50 Flourish One robots on September 29, priced at $3,555 each.

• The pitch: a home robot for repetitive chores, trained by demonstration rather than programming.
• The hardware: a six-wheeled mobile base with two arms, rather than a bipedal walking design.
• The compute: runs on a Raspberry Pi, a deliberate cost-saving choice over custom robotics hardware.

The catch: it's not a general-purpose robot yet. It's built for simple, specific, repeatable tasks, not open-ended household work.

Flourish's core pitch centers on how quickly an owner can teach the robot a new task, without any coding.

Founder Antoine Marcel described the process directly: "When you are training your robot to water your plants, you just take your phone, move your phone, and it will move the arms." The robot mirrors the phone's movement to learn the motion, typically within about 30 minutes per task.

That teaching method is central to Flourish's whole approach: rather than shipping a robot pre-programmed for a fixed list of chores, each owner trains their own robot for their own home's specific layout and habits.

Marcel has been direct about why a single, universal skill set doesn't work well for home robotics: "The way I tidy my apartment is not the same as you do in your house." That's the reasoning behind per-home, cloud-based AI training rather than a one-size-fits-all skill library baked in at the factory.

By the numbers: each arm carries a 1.5kg payload capacity, uses two-fingered grippers with wrist-mounted cameras, and the robot runs on roughly 12 hours of battery life for arm operation, with a small lidar sensor handling obstacle detection around the home.

Flourish One stands about 1.1 meters tall and weighs roughly 20kg, built on a telescopic body structure that gives it a reach of about 89 centimeters. The six-wheeled base was a deliberate choice over legs.

Marcel explained the reasoning: a wheeled, balanced mobile platform costs less to build and is inherently safer in home environments than a bipedal design attempting to walk around furniture, pets, and children. That tradeoff sacrifices some of the visual "humanoid" appeal rival robots lean on, in exchange for lower cost and fewer ways to tip over or injure someone.

What it can't do: the robot isn't waterproof and isn't suited for tasks like washing dishes or other complex, multi-step chores. Flourish is positioning it narrowly, for well-defined, repeatable tasks rather than as a do-everything home assistant.

Flourish One's most striking feature might just be the price tag. At $3,555, it undercuts prominent rivals by a wide margin.

The comparison: 1X's bipedal Neo robot is priced around $20,000 for US home delivery, positioning it as a premium, aspirational home humanoid. Flourish is explicitly chasing the opposite end of the market, prioritizing an actually attainable price point over matching Neo's more humanlike walking design or broader task ambitions.

That price gap reflects a real strategic bet: rather than competing directly on capability with well-funded bipedal humanoid projects, Flourish is trying to prove a narrower, cheaper robot can still deliver enough real value to build a customer base and iterate from there.

Flourish One's grippers and wrist cameras exist in a category where safety standards are still genuinely unsettled. Consumer environments are widely considered harder to design for safely than factory floors, since homes contain children, pets, cluttered surfaces, stairs, and unpredictable situations a warehouse robot never has to navigate.

Why that context matters here: the International Organization for Standardization is currently updating its safety requirements for personal care robots, a framework that's over a decade old, but the proposed update doesn't set hard limits, testing methods, or enforcement mechanisms specific to home robots working around people.

Consumer safety standards for robotic manipulators are close to nonexistent, and legal liability for a home robot causing injury remains largely untested in court.

Some companies in the space, including 1X Technologies, have responded by deliberately limiting what their robots are physically capable of, restricting their ability to handle heavy, sharp, or hot objects regardless of what the hardware could technically do. Flourish's own limitations, no dishwashing, not waterproof, narrow task scope, read as a similar kind of self-imposed caution rather than a simple technical shortfall.

Flourish One's wrist-mounted cameras and cloud-based training model raise a separate question apart from physical safety: privacy. Robots with always-on or frequently active cameras aren't necessarily built with the same security hardening as dedicated home security cameras, and a mobile camera that moves through multiple rooms creates a different kind of exposure than a fixed one.

The tradeoff is baked into the product itself: Flourish's phone-based, per-home teaching method is precisely what makes the robot flexible and personalized, but it also means task demonstrations and ongoing camera footage likely get processed through Flourish's cloud infrastructure rather than staying entirely on-device.

That's a common pattern across the teleoperation-trained robotics industry broadly, where cloud processing is what makes rapid, flexible learning possible in the first place. It's also exactly the kind of data flow that makes some potential buyers hesitate before letting a camera-equipped robot loose in their living room.

Flourish's pre-seed funding round was led by Families Fund, a firm founded by Kartik Sathappan, a detail that lines up neatly with the company's stated focus on busy parents and household chores as its initial target use case, rather than a broader enterprise or industrial robotics play.

Building a robot around a Raspberry Pi rather than custom silicon or a more powerful industrial computer is itself a notable engineering decision. Raspberry Pi boards are inexpensive, widely documented, and easy to source at scale, all useful traits for a startup trying to keep hardware costs low enough to hit a sub-$4,000 price point.

The tradeoff: a Raspberry Pi is nowhere near as powerful as the custom compute stacks found in higher-end humanoid robots, meaning heavier processing, like real-time complex vision or advanced autonomous reasoning, likely happens in Flourish's cloud rather than on the robot itself. That architecture choice reinforces why cloud connectivity isn't optional for Flourish One; it's core to how the robot actually functions day to day, not just how it learns new tasks.

With only 50 units shipping before Christmas 2026, Flourish is treating this initial batch as a proving ground rather than a mass-market launch. How well those first robots actually perform in real homes, on real chores taught by real, non-technical parents, will shape whether Flourish scales production or has to rethink its approach.