My View on Commercializing Home Robotics

Price against labor, not against appliances.

$10k is expensive for an appliance, but cheap compared with the labor Memo could replace. A weekly cleaner at $150-200 costs roughly $8-10k every year. Looked at that way, Memo costs about the same as one year of weekly cleaning. The challenge is framing it well: consumers need to compare Memo to the recurring cost of household help, not to the cost of other appliances.

I would sell the body and rent the brain: price the hardware near cost, offer financing, and charge a subscription for the skill library and service. Sunday's assets move in opposite directions - the hardware depreciates from the day you buy it, while the intelligence should appreciate every week. A one-time purchase doesn’t cover that. It also fixes the inevitable problem of mismatched consumer expectations. A subscription sets the expectation that Memo will keep improving. Early capability gaps become part of a product that's getting better rather than a promise the hardware failed to deliver.

The first 1,000 homes are a sampling decision, not a sales decision.

The general instinct is to choose the first households based on enthusiasm for the product and willingness to pay. I'd choose them based on what Sunday needs to learn. The first 1,000 homes are effectively helping choose the environments the next model learns from. I’d treat these homes more like a sampling problem. Before selecting them, I’d define the variables most likely to affect performance - layout, home size, clutter, appliances, pets, kids, flooring, laundry setup - and score applicants on both commercial value and learning value.

Then build the cohort deliberately: enough high intent customers to test demand/willingness to pay, but enough variation to expose Memo to the environments it needs to handle. I’d track performance by household type as the beta grows and keep filling gaps where the model is weakest.

There’s an interesting tension: the households that most want Memo (affluent, urban, dual income) may represent a relatively narrow set of homes. I'd deliberately include harder environments, even if that means a less perfect beta experience. The goal with the sample isn't just happy customers, it's making future robots meaningfully better.

Longer term, I think one of the biggest markets could be aging in place. People have a limit on what they'll pay to avoid chores, but they'll pay far more to avoid losing their independence. A robot that helps someone stay in their own home longer solves a much bigger problem than convenience. It probably isn’t the market to start with - first prove the product is safe and reliable in everyday homes; but it’s a market I’d build towards.

Reliability is the business model.

I strongly believe the service model can make or break this business. Nobody has figured out how to service a heavy robot inside thousands of homes, and sending a technician every time something goes wrong gets expensive very, very quickly.

I’d focus on three things early:

• Make Memo easy to diagnose and repair. The robot should know what failed, and major components should be easy for a technician to swap quickly.
• Give every robot a service P&L. Track technician time, travel, parts, shipping, remote support and downtime by robot, hardware generation and market. At 50 robots, service problems are manageable. At 5,000, they can destroy the economics.
• Prioritize density over reach. Start with a handful of cities where technicians can service several homes a day rather than spreading customers across the country. I’d rather have 1,000 robots concentrated in a few markets than 1,000 spread across the US.

I’d feed service data directly back into the hardware roadmap. Rank failures by how often they happen × what they cost to resolve. Sunday is already using beta to learn what the robot can do. I'd use it just as aggressively to understand what breaks, how often it breaks, what it costs to fix, and how those numbers improve with each hardware generation. Before scaling nationally, I'd want to know exactly what it costs to keep one Memo working in one home for one year.

Getting from $20k to $10k is as much a business problem as an engineering one.

A lot of the path from $20k to $10k should come from scale: supplier negotiations, volume commitments, manufacturing improvements, and designing components for cheaper production. That's where I'd want a very clear cost down roadmap by component, with an owner, volume threshold, and negotiation date attached to each major opportunity.

The hand is particularly interesting. It's expensive, there isn't an obvious supplier, and it's central to what makes Sunday different. When something is both a major cost and a major differentiation, owning it starts to make sense.

Memo's wheeled design should also be a real economic advantage. Fewer actuators, less power and less complexity than a robot that has to walk and constantly balance should make $10k much more achievable.

The other question I’d answer early is how much of the supply chain needs to be American. Sunday’s positioning around trust, privacy, and American homes is valuable, but it may come at a real cost per robot. I’d make the tradeoff pretty concrete. Build two versions of the BOM: the lowest cost supply chain we can realistically build, and the version that meets whatever “American-made” standard we think matters to customers. Then price the gap component by component - not just purchase price, but tariffs, freight, lead times, and working capital. If the difference is big, we need to fully understand whether customers will actually pay for it. The goal would be to decide where American sourcing creates real value and where we’re just paying for the narrative.

Beta should prove the business, not just the robot.

Sunday now has confidence that the technology works. I’d use beta to answer three different questions: