Care is a craft. We build the infrastructure around it.
We exist for one conviction: the people who deliver care in person should be made more capable by technology, not replaced by it.
For forty years, software was built for people at desks. The people who deliver care in person, across the floors of a hotel, got almost none of it. The industry’s answer now is to replace them: chatbots for the guest, kiosks for the counter.
We think that is the wrong bet wherever presence is the product. The craft of care cannot be automated, and shouldn’t be. But the work around it, the coordination, the memory, the handoffs, the follow-through, has quietly eaten the time that craft needs. So we automate that work, and give the hours back to the people.
The future we want to build
Caring for people in person is one of the largest kinds of human work, and it never got its infrastructure layer. We intend to build it.
In that future, what a team learns stops vanishing at shift change. It becomes a memory the house owns and compounds, more valuable every year it runs. The judgment a great host holds by instinct can be carried, taught, and shared, instead of walking out the door. None of this makes service less human. It makes room for more of it.
What grounds the design
Our name comes from omotenashi, the Japanese practice of anticipatory care. It is not decoration. It is the design specification, and these are the lines it draws.
Anticipate without surveilling
Read what a guest needs before it is said, from what they have shared and what the team has noticed, never from covert tracking or a profile bought off the open web.
The best care is invisible
If care announces itself, it stops being care. Omote stays entirely behind the floor. It never speaks to the guest, and the guest never meets the software.
Memory belongs to the house
The relationship belongs to the team, not to one star employee’s phone. What the operation learns is shared by whoever is on the floor today. It stays when they leave.
Support judgment, not scripts
The craft is learned, not scripted. Omote carries the context and leaves the call to the person, there to make good judgment easier, never to replace it with a rule.
And a few lines we will not cross. Omote never surveils your team, never lets your data train anyone else’s models, and the memory it builds stays yours. What we refuse to build is as much the design as what we do.
Carlos Marin
I grew up in my family's hospitality businesses, hosting travelers across Venezuela, from the Andes to the Amazon. I learned early that the people who provide care deserve admiration, not replacement.
Since then I've kept finding the same thing inside every great service experience, from Kyoto's ryokans to São Paulo's botequins: someone who remembered, anticipated, and cared. After a decade across McKinsey, the founding team at Sonder, and building AI products at Google, I'm building Omote around that conviction. I also love hosting people for dinner.
We publish what we learn as we build it, in the open. No cadence, no noise.
If you build software and a hard, unsolved problem pulls at you, see Building Omote.