Learn / Glossary
Words to know
Every term from our lessons, in plain words. Each one links to the lesson that explains it.
- Accepted
- Checked by our team and approved for pay.
From: What you earn → - Accepted hour
- An hour of video our team has checked and approved for pay.
From: Who we are → - AI
- Short for artificial intelligence. Computer programs that learn from examples instead of following hand-written rules.
From: AI training → - AI lab
- A team that builds and trains AI models.
From: Who uses this data → - Automation
- Machines doing work that people used to do.
From: Will robots replace cleaners? → - Autonomous
- Able to run on its own without someone driving it.
From: Robots in cleaning today → - Building client
- The company or person that pays you to clean a building.
From: Getting your building client's OK → - Building services
- The trade that keeps buildings clean and running, like janitorial work.
From: Who we are → - Change order
- A written request to change the Order Form.
From: The paperwork → - Clutter
- Lots of small things in the way, like papers and cords.
From: What robots still can't do → - Consent
- A clear yes from a person before they are recorded.
From: Explaining it to your crew → - Dataset
- A big collection of examples, like thousands of hours of video, gathered to teach a computer.
From: Egocentric data → - Detail work
- Careful cleaning of small or hard spots, like vents, edges and glass.
From: What this means for you → - Dexterity
- Skill with the hands, like a good grip or careful wiping.
From: Will robots replace cleaners? → - DocuSign
- A website where you sign papers online.
From: Your first two weeks → - Egocentric
- Centered on me. Video shot from the worker's own point of view.
From: Egocentric data → - Employment projection
- The government's estimate of how many jobs there will be in ten years.
From: How far away are robots from doing a cleaner's job? → - Feedback
- A note from our team about what to fix in your video.
From: Your first two weeks → - First-person video
- Another name for egocentric video. The camera sees what you see.
From: Egocentric data → - Floor scrubber robot
- A machine that cleans open floors along a set path on its own.
From: Will robots replace cleaners? → - Forecast
- A best guess about the future, based on data.
From: How far away are robots from doing a cleaner's job? → - Foundation model
- A big, general model other teams build on for their own jobs.
From: World models → - Frame
- One still picture. A video is many frames in a row.
From: Vision-language models → - Grip
- How the hand holds something. Robots find soft things like trash liners hard to grip.
From: Vision-language-action models → - Head mount
- A headband that holds a phone at the forehead, pointed down at the hands.
From: Who we are → - Headcam
- A camera worn on the head. In our program it is your own phone in a head mount.
From: Egocentric data → - Human motion data
- Recordings of how a person's body and hands move during a task.
From: Yes, your video trains robots → - Humanoid robot
- A robot built in a human shape, with arms, hands and often legs.
From: Who uses this data → - Instruction
- What you tell the robot to do, like "wipe the counter."
From: Vision-language-action models → - Label
- A note that says what task is happening in a piece of video.
From: Where your video goes → - Machine operator
- The person who sets up, runs and cares for a cleaning machine.
From: What this means for you → - Manipulation
- Robot word for using hands: grabbing, turning, wiping, lifting.
From: Robot training → - Master Services Agreement (MSA)
- The main contract. It sets the rules for every Order Form.
From: The paperwork → - Mediation
- A meeting with a neutral person to settle a dispute before going to court.
From: The paperwork → - Model
- The program that comes out of training. It is what makes the guesses.
From: AI training → - Month-one commitment
- The 200 to 1,000 recorded hours a new customer plans for the first month.
From: What you earn → - Moravec's paradox
- The finding that tasks easy for people, like seeing and grabbing, are hard for computers.
From: What robots still can't do → - Motor control
- Sending moves to a robot's joints so its arms and hands move.
From: Vision-language-action models → - Off camera
- Kept out of the video. Your cleaner pauses or points the phone away.
From: Getting your building client's OK → - Order Form
- The short paper that sets your pay rate and term.
From: The paperwork → - Permission
- A clear yes, best in writing, that lets your crew record at a building.
From: Getting your building client's OK → - Prompt
- The question or instruction you type to an AI.
From: Vision-language models → - Property manager
- The person who runs a building for its owner and often signs the cleaning contract.
From: Getting your building client's OK → - Quality check
- Going back over finished work to make sure it meets the standard.
From: What this means for you → - Recorded Hour
- The agreement's name for an hour of uploaded video that counts toward pay.
From: What you earn → - Review window
- The 24 to 48 hours our team takes to check an upload.
From: What you earn → - Robot learning
- Teaching a robot a task by showing it many examples.
From: Yes, your video trains robots → - Robot scrubber
- A floor machine that scrubs a route by itself.
From: Robots in cleaning today → - Robot training
- Teaching a machine to move and handle things to finish a task.
From: Robot training → - Route
- The path a floor robot follows through a building.
From: Robots in cleaning today → - Service robot
- A robot built to do work for people outside a factory line.
From: Robots in cleaning today → - Setup call
- A call where our team sets up the recording app with you.
From: Your first two weeks → - Simulation
- A computer copy of a room where a robot can practice before trying the real thing.
From: Robot training → - Staged video
- Video made to add time, like wiping a clean spot again. It is not accepted.
From: Explaining it to your crew → - Synthetic data
- Made-up video or pictures created by a computer to use in training.
From: World models → - Teleoperation
- A person steering a robot by remote so it can record the moves.
From: Robot training → - Training
- Showing a computer many examples so it learns a skill.
From: Where your video goes → - Training data
- Examples a computer learns from, like videos of real work.
From: Yes, your video trains robots → - Upload
- Sending video from the phone to our storage online.
From: Where your video goes → - Vision-language model (VLM)
- AI that looks at pictures or video and answers in words.
From: Vision-language models → - Vision-language-action model (VLA)
- AI that sees the room, reads an instruction and moves a robot to do it.
From: Vision-language-action models → - World model
- AI that learns how things move and change, so it can predict what happens next.
From: World models →
Still have questions?
Talk to our team.
Call us and ask anything about pay, your clients or your crew. Or sign up and we will call you.