The robot is the cheap part
In an installed robot cell the arm is 25–40 % of the cost. Mapping, programming, integration and supervision make up most of the rest.
Cyberstrata is a marketplace for work done by robots. A customer posts a job and a deadline. We plan it, bring robots from fleet owners together with the operators who know them, coordinate the work around the clock and prove it was done. Fleet owners earn from robots that would otherwise stand still.
Robots keep getting cheaper. Making one do a specific job still takes months of setup.
So nobody rents a robot for a single job, and the robots that already exist stand still between contracts.
In an installed robot cell the arm is 25–40 % of the cost. Mapping, programming, integration and supervision make up most of the rest.
Rental marketplaces list robots and leave the job to the renter. Rented humanoids are booked for shows, with an operator, because they cannot yet work alone.
Formic, Robonnement and Skyline Robotics sell robot hours and outcomes. Each must buy every robot it uses, which takes debt and grows slowly.
Robots sit idle between contracts, at certain hours or out of season. How much capacity is free, and when, is our first open question.
The customer buys an outcome. We plan it, staff it from several fleets and prove it was done.
What, where, by when, and what counts as done.
A fit check decides which robot types can physically do which parts. Logistics decide how many robots, transport, elevators, charging and supervision.
Fleet owners whose robots are able, free in that window and nearby, each with its own operators.
The owners' operators run their own robots. We coordinate across fleets, around the clock.
The robots' own logs, such as coverage maps and m² cleaned, plus spot checks on site decide acceptance.
Owners are paid and we keep a commission. Track records update and the job plan is saved, so the next run is faster.
90 planned robots clean a 45-floor tower in 4 hours. All 400 would still be riding the elevator at 06:00.
One owner runs 400 cleaning robots; another needs a tower cleaned overnight. The elevator, not the cleaning, sets the pace. Record time comes from planning, not from more robots.
| Plan | Robots | Trips each way | All robots up | Cleaning done | All robots down | Night, h | Robots left free |
|---|---|---|---|---|---|---|---|
| All 400 robots | 400 | 134 | 04:42 | 04:57 | 11:37 | 13.6 | 0 |
| 90 robots, planned | 90 | 30 | 23:30 | 00:30 | 02:00 | 4.0 | 310 |
45 floors × 2000 m²1000 m²/h per scrubber3 robots per elevator trip3 min round tripcleaning starts when all robots are upEstimateinputs to verify with makers
Fewer robots, well scheduled, finish first. When each robot starts as soon as it arrives, the planner's fastest plan uses 60 robots and ends after 2.6 h.
A robot counts only if it can do the part, is free in the window and can be moved to the site. Cleaning fleets often already work at night.
The first night in a building is slow: floors must be mapped and elevators connected. The saved plan makes every repeat faster.
Rental platforms leave the job to the renter. Service operators must buy every robot.
We sit between customers and fleet owners and bring owners revenue, instead of adding a cost layer on top of their robots.
| Category | Examples | What they do | What they leave open |
|---|---|---|---|
| Rental marketplaces | Sharebot, RoboShare (US) | Owners list idle robots; renters book by the day, week or month | The renter works out fit, integration and operation |
| Humanoid rental | AgiBot, Qingtian Zu (China) | Humanoids rented with an operator, from about 3000 yuan per day | Displays, film shoots and education, not work |
| Robots as a service | Formic, Robonnement, Skyline Robotics | Operators buy, run and maintain robots and sell hours or outcomes | Every robot bought by the operator: capital-heavy, slow to grow |
| Skill app stores | Unitree UniStore, OpenMind, UR+ | Apps and add-ons for robots | Mostly one maker, or fixed recordings |
| Cautionary case | READY Robotics | One operating system across robot brands; raised $41.5 million, shut down in 2024 | Sat on top of makers' software as an extra cost |
Research and a plan, no customers yet. Five assumptions decide whether this is a business.
| No. | Assumption | How we test it | Status |
|---|---|---|---|
| A1 | Building owners and facility managers buy a robot-delivered job by outcome, for example per m² | 10 interviews; first paid jobs | Untested |
| A2 | Fleet owners have robots free in the windows customers need, and take jobs through us | 10 interviews; robots listed | Untested |
| A3 | Our planning makes jobs faster or cheaper than an owner working alone | Planned against actual time in pilots | Untested |
| A4 | Robot logs plus spot checks are accepted as proof of work | Acceptance in pilots | Untested |
| A5 | A 10–20 % commission fits inside the job margin | Pilot pricing | Untested |
Two ways in: post a job, or list a fleet.
Describe a building, a time window and what counts as clean. We return a plan: which robots, how many, how long the night takes and what it costs.
Tell us which robots you run, where, and when they stand still. We bring qualified jobs with a plan, a contract and payment. Your operators run your robots.