Head camera view: two hands in grey gloves lift a cardboard box off a warehouse shelf.

Illustrative time 00:00:12

Four first-person illustrations: lifting a box, tying rebar, driving a screw and scanning a parcel. The points represent hand movement; they do not show an acceptance assessment.

For warehouses, logistics, construction and workshops

Your team knows how the work gets done.

Mirentor proposes recording hands-on tasks from the perspective of the people doing them, to help train robotics models. See what participating could involve for your business and your team.

Join the waitlist How it works

Mirentor proposes recording hands-on tasks from the perspective of the people doing them, to help train robotics models. See what participating could involve for your business and your team.

Nine cameras, nine tasks

Nine cameras for a closer look at the work.

The counter adds simulated time from nine cameras. It does not represent real recordings, accepted hours or earnings.

  1. Camera 1: two hands in gray gloves lift a cardboard box off a warehouse shelf.
  2. Camera 2: two hands in gray gloves tie two rebars with wire and pliers on a building site.
  3. Camera 3: one hand steadies a pine board while the other drives a screw with a drill.
  4. Camera 4: one hand holds a parcel while the other scans it with a barcode scanner.
  5. Camera 5: two hands take a small box off a shelf and lower it into a plastic crate.
  6. Camera 6: two hands push a pallet jack loaded with boxes down a warehouse aisle.
  7. Camera 7: one hand holds a steel ruler on a sheet of cardboard while the other cuts along it with a utility knife.
  8. Camera 8: one hand steadies a board while the other sands it with a sanding block.
  9. Camera 9: one hand holds a wire while the other tightens a screw in a breaker panel.

Accumulated simulated timeIllustration ×1 00:00:00

01 · What labs are buying

Experience is in the details.

A written procedure describes the steps. First-person video shows how they are carried out: the hands, the tools and the adjustments each task requires.

It is also known as egocentric video. Researchers can use it to study physical tasks and help train robotics models.

Packing orders, handling materials or using tools: your team knows the work and the details of each task. Mirentor aims to bring that experience to robotics labs.

Three public facts from the field

  1. More than1 millionHours of video

    One robotics lab pretrained its model on more than 1 million hours of mostly head-mounted video.

    Source: the lab’s model page (opens in a new tab)
  2. Half a millionHours of real-world data

    Another lab trained its model on half a million hours of real-world data, recorded with devices that people wear.

    Source: the lab’s blog (opens in a new tab)
  3. Logistics, restaurants, factories, officesWork settings

    One humanoid robot company lists logistics, restaurants, factories and offices among the tasks it collects.

    Source: the company’s announcement (opens in a new tab)

An illustration of how movement is studied

  1. Head camera view: two hands in gray gloves lift a cardboard box off a warehouse shelf.
    1 · The videoThis is work as a head camera sees it.
  2. 2 · The handsHands, tools and objects help explain the task.
  3. 3 · The pointsThe points represent hand movement in this illustration.
  4. Two robot hands hold a cardboard box on a shelf, seen from above.
    4 · The robotVideo can provide examples for training robotics models; it does not, by itself, teach a robot to perform the task.

Learn from the best.

02 · How it works

Four steps to your setup day

From the waitlist to the first recorded hour, this is the whole process.

Camera kitStep 04Not to scale

The head cameras we bring on setup day. Illustrative drawing, not a specific model.
  1. Join the waitlist

    Fill out a short form: your business, your city and how many people could wear a camera. We open the next city where the most businesses are ready.

  2. Intro call

    On a call, we explain the program, answer your questions and check whether your operation fits.

  3. Contract

    If your operation fits, you sign a contract and we set the date for your setup day.

  4. Setup day

    We bring the head cameras, prepare your site and train your staff. We set up the first site in each city in person. Then a local city manager runs that city.

Start with step 01

03 · What counts

Only accepted hours count.

A recorded hour is not necessarily an accepted hour. Acceptance depends on the program’s review criteria.

The example distinguishes tasks from breaks; it does not replace acceptance criteria.

Time card Hypothetical shift. It illustrates durations, not approved hours or earnings.
  1. 07:00 Unloading a truck 1 h 30 min Task
  2. 08:30 Stacking boxes 1 h 30 min Task
  3. 10:00 Break 15 min Break
  4. 10:15 Using a power drill 1 h 45 min Task
  5. 12:00 Lunch 1 h Break
  6. 13:00 Packing orders 3 h Task

The example totals 9 hours: 7 hours and 45 minutes of tasks and 1 hour and 15 minutes of breaks. Task time is not the same as accepted hours.

The proposed model calculates payment from accepted hours, not shift length.

04 · The money

One accepted hour, three shares

Labs pay us by the accepted hour. We split each accepted hour three ways: your business, your workers and us. It is extra income for work your team already does.

Diagram: one accepted hour, 60 minutes long, splits three ways: your business, your workers and Mirentor. The amount of each share is pending.

1 accepted hour

  • $

    Your business

    Income for work your team already does.

  • $

    Your workers

    You set this share. We guide; we do not force.

  • $

    Mirentor

    We bring the cameras, train your staff and sell the video to labs.

The pay rule

If your workers get nothing, the program does not work.

  1. Workers get more
  2. They record more hours
  3. Everyone earns more

It is a loop: the more your workers get, the more hours they record, and the more everyone earns. Then it starts again.

05 · For your team

“Are we training the robots that replace us?”

It is the first question your workers will ask you. It is fair, and it deserves a straight answer.

We do not make promises about the future of jobs. This is what is true today:

  • Joining is voluntary. Workers who want it go first.

  • Every accepted hour is paid. We split it three ways, and one share goes to the workers.

  • The explanation comes before the camera. You and our team explain the program before anyone records.

  • Anyone can stop recording at any time.

  • The video trains robots. It is never used to rate or monitor anyone’s work.

06 · Fit check

Does your operation qualify?

Four questions to find out if your business qualifies. At the end you will see the result and the next step.

Already sure? Go straight to the waitlist

Question 1 of 4. What kind of work does your team do?

Labs want real work: moving boxes, handling loads, using power tools.

Question 2 of 4. How many workers could wear a camera?

Count only people who work with their hands. The best fit is 100 to 500.

Question 3 of 4. Will your workers get a share of each accepted hour?

An accepted hour is time spent working, with hands in frame. You set your workers' share. We guide; we do not force.

Question 4 of 4. Where is your operation?

We open one city at a time. Latin America first.

Join the waitlist

Operation profile

01Industry
02People
03Team pay
04City
Result

07 · Where we start

One city at a time

Latin America first. We open one city at a time, starting with these four.

Choose a city

Dot map of Mexico, Central America and South America with four cities marked: Mexico City, Panama City, Medellín and Lima.
First site
We set it up in person in every new city.
After that
A local city manager runs that city.
Next city
We choose it by how many businesses are ready there. Joining the waitlist adds to your city's count.
United States
The program also accepts US businesses, but Latin America comes first.

A short summary, not legal advice.

Your city is not on the map? Join the waitlist anyway.

08 · Before you decide

Common questions

What does my business have to do?

Your team keeps doing its usual work. Workers who join wear a camera while they work. You set what your workers get for each accepted hour, and you host us on setup day. We bring the cameras, train your staff and sell the video to robotics labs.

What does it cost me?

Nothing. There is no cost to your business, and we bring the cameras at no cost.

How much does my business get?

Labs pay for every accepted hour, and that payment is split three ways: your business, your workers and us.

Your share per hour $

See the split

Who decides what my workers get?

You do. We guide; we do not force. If workers get nothing, the program does not work: the more they get, the more hours they record and the more everyone earns.

We pay workers directly.

Do my workers have to join?

No. Joining is voluntary, and workers who want it go first. You and our team explain the program before anyone records.

Rules by country

What counts as an accepted hour?

An hour of video of someone actually working, with their hands in frame. Breaks do not count.

See what counts

Which cities come first?

Panama City, Medellín, Lima and Mexico City. We open one city at a time and choose the next by how many businesses are ready there. US businesses can join too, but Latin America comes first.

See the cities

What happens after I join the waitlist?

We schedule an intro call. If your operation fits, you sign a contract and we set your setup day. After the first site in your city, a local city manager runs that city.

Another question? Ask it on the intro call.

Join the waitlist

09 · Last step

Join the waitlist

Tell us about your business and your team’s tasks. Registration does not confirm acceptance or a start date.

Standby

All fields are required, except WhatsApp.

Your operation
Location

Enter your operation’s location. Registration does not confirm availability in that city.

Contact

With the country code, for example +1.

Demo form. It does not yet send or save data.