Yet another startup is coming out of the woodwork claiming it’s built a home robot that might actually be useful.
This time it’s San Francisco-based Tangible Robots and its dexterous home assistant named Eggie. Its design is clean and minimalist, with smooth white panels covering much of its structure and exposed black joints at the wrists and elbows. Its head is rectangular with two front-facing cameras for eyes.
The startup revealed itself to the world with a 26-second video showing its flagship cleaning up spilled coffee. In the clip, Eggie grabs a white mug with one hand while its other hand wipes the countertop with a red cloth.
“The world was build around humans,” the startup said on X. “Eggie doesn’t look human, Eggie interacts like us. Dexterous. Mobile. Compliant. We’re building Eggie to be the smartest robot to ever walk on Earth.”
On its website, Tangible says Eggie is built around three engineering pillars: dexterity, compliance, and whole-body control. It’s trained on real-world data gathered from hundreds of environments and thousands of interactions. The startup says it’s taking a full-stack approach to tackling general-purpose autonomy, with everything from its hardware to control systems to artificial intelligence developed in-house.
Tangible Robots is led by CEO Bipasha Sen, who says she left an MIT PhD program to pursue building robots for homes. She co-founded Tangible with fellow researchers Shankara N. Vaidyanathan and Benjamin Soria. Vaidyanathan leads core robotics and control systems work, while Soria heads mechanical and software engineering.
The trio began building prototypes while incubated at Founders, Inc., an SF-based hardware-first incubator. Tangible has raised about $4 million to date across two seed-stage rounds, according to Traxcn. A pre-seed investment closed in late 2024 followed by a $3 million round in 2025, led by India-based Blume Ventures and Micelio Technology Fund.
Tangible says its differentiator is its focus on context-rich manipulation, using humanlike mechanical hands, compliant arms, and real-world training data to tackle the messy and unpredictable. The team argues that the next big breakthrough toward general-purpose robotics will come from AI that understands the 3D world, how things feel, and how to move safely and effectively.
While many robotics firms rely heavily on synthetic data for training, Tangible focuses on real-world data collection. One of its methods involves engineers wearing a sensor-packed partial exoskeleton that captures human-level touch and force information during tasks like cleaning and picking up objects.
The startup is recruiting research and engineering talent with expertise in imitation learning, reinforcement learning, dexterous manipulation, and vision-language-action (VLA) models. They’re also seeking mechanical, software, and control engineers, advertising $1-2 million total compensation for top candidates.
For now, Tangible says it’s focused on refining autonomy through large-scale, real-world training rather than commercialization. Their emphasis on human data collection is similar to Silicon Valley-based Figure AI, which secured more than $1 billion in funding earlier this year as it rushes its bipedal humanoids into homes and industry.