Maxitech is proud to back Weave Robotics, the Y Combinator S24 team behind Isaac 0, a robot already folding laundry for paying customers in the Bay Area—and building a compounding learning loop that makes home autonomy finally feel inevitable.

 

Why has home robotics been stuck

The biggest gap in robotics is not intelligence in the lab. It is reliability in the last mile.

Homes are chaotic compared to warehouses or factories: lighting changes, floors are uneven, clothes come in infinite shapes and textures, and the “inputs” are whatever someone pulls out of a hamper. A system can look impressive in a controlled demo and still fail the first time it meets a real household.

That is why home robotics has plateaued at narrow, task-specific devices. The hard part is not building a robot that can fold laundry once. It is building one that keeps working across thousands of real loads of laundry, that learns from its failures, and improves without requiring a technician in the room.

 

Meet Weave Robotics

Weave started with a bounded, high-frequency task that forces real-world reliability: laundry folding. The average household runs multiple loads per week, and the “done vs. not done” outcome is obvious—perfect conditions for fast iteration. Isaac 0 is a stationary, home-ready folding robot designed to deliver value immediately. It is already running in Bay Area households and in commercial laundry settings, where each fold generates the data required to improve performance over time.

Weave’s core insight is pragmatic: instead of waiting for perfect autonomy, Isaac 0 runs autonomously wherever it can. The robot starts as useful on day one, and the corrections feed the next model update.

 

What makes Weave different

Two decisions separate Weave from most physical AI teams.

Hardware discipline. Isaac 0 is stationary, stable, and built with as few moving parts as it takes to do the job well. Fewer components mean fewer failure modes, faster iteration, and a cost structure that general humanoid platforms will struggle to match in the near term.

A compounding data advantage. Embodied AI improves fastest where robots are already deployed. Every successful fold, and every corrected fold, becomes training data. The flywheel compounds precisely because Weave is shipping.

Every week in the field creates concrete learning: more edge cases captured, more remote interventions converted into training examples, and more reliability shipped back to customers.

 

Why we invested

Maxitech invested in Weave because they are building embodied AI the only way it becomes a durable business: in the real world, with real users, under real constraints. Demos do not produce reliability; deployments do. Weave’s approach turns every “messy” household edge case into a measurable input for improvement—and that is how physical AI crosses from impressive to inevitable.

We also underwrite founders. Weave is led by Kaan Doğrusöz and Evan Wineland, who have been building together since their time at Carnegie Mellon (they met in 2015, and have been close collaborators ever since). Both are Apple veterans, and Kaan worked alongside Evan in Apple’s ecosystem as they shipped in one of the most operationally demanding product environments in the world. That shared history shows up in how Weave operates today: clear division of responsibilities, high trust, and an obsessive bias toward deployment-quality engineering. This category rewards operators who can build, manufacture, deploy, support, and iterate—not teams that only demo.

 

Looking ahead

Robotics is entering an operational era. The teams that win from here will be the ones already in homes and businesses, already collecting task data, and already iterating on it every week.

Over the next 24 months, we expect Weave to expand its footprint in commercial laundry environments, where throughput, consistency, uptime, and serviceability are non-negotiable. Early deployments in real-world laundry operations create the exact feedback loop embodied AI needs: varied garments, high volume, tight SLAs, and clear metrics for speed and quality. In physical AI, the lead is built in production: the longer Isaac runs in demanding commercial settings, the more performance data Weave can capture, the faster reliability improves, and the harder that advantage becomes to replicate—before that same capability rolls downstream into additional use cases.

Learn more at weaverobotics.com.