My robotics company that doesn’t build robots

A few months ago ago I was watching demos from the big humanoid companies and had a realization: these robots are incredible, but they have nowhere to actually go.

The robotics industry has a deployment problem. Companies raise BILLIONS to build hardware, then test it in pristine labs that look nothing like real homes. When it's time to deploy, they don't have relationships with property owners, they don't have operational infrastructure, and they definitely don't know where the cleaning supplies are kept.

We decided to build the boring part. We partner with robotics OEMs and give them what they actually need: real environments to deploy in, operational infrastructure to support it, and the hyper-specific data that makes robots actually useful.

Right now we're running cleaning operations across homes in CA. All with humans, but we have deals in place to be the first to deploy robots for the job when the OEMs are ready. We know they won't be super good at cleaning, so we will be integrating them with human cleaning teams to do simple tasks (wipe the counter, pretty much). But every turnover generates training data.

A few things that surprised me:

1. The data that matters isn't always "how to fold a towel." It's "where does THIS towel go in THIS bathroom." Hyperlocal beats general-purpose.
2. Rental property managers don't really care if a robot cleans their unit. They care if it's clean by 3pm. Reliability > novelty.
3. The unsexy operational knowledge is actually the moat.

We're trying to build the "picks and shovels" for the robotics gold rush, except the picks and shovels are cleaning schedules schedules and a database of where people keep their TV remotes.

Anyone else building in the robotics space? Curious how others are thinking about the deployment gap.