Order from Lab 803632

Computer use trajectories

Recorded sessions of real employees doing real work in business software (ERP, CRM, accounting, ticketing, spreadsheets, internal tools): every click, keystroke and scroll with its timestamp, the screenshot before and after, the DOM or accessibility tree, the goal, and whether the goal was reached with proof. One unit is an episode (one person, one goal, one continuous attempt), failures kept and marked, with commercial training rights and the recorded employees' consent.

How answering works

Any number of people fill one order. The lab receives one dataset, in its own columns.

Your agent reads the order
It looks at what you hold and says which columns you can fill, and what is missing.
Mora checks what you send
Every record: no copies, no personal data, not already public, not on Mora from someone else.
The lab pays per record it accepts
The lab named the price. You fill as much of the order as you can.

The columns the lab wants

You do not need every column. Mora says which ones each dataset fills.

Column Type What it must hold

episode_id id One person, one goal, one continuous attempt

operator_id id Pseudonymous id of the person, the same across their episodes

application / app_version / platform label Which business software, which version, web or which OS

screen_w / screen_h / scale_factor number Screen geometry, needed to trust click coordinates

goal text What the person set out to do, in words

goal_source label Written before by the operator, taken from a ticket, or back-filled by an annotator or a model

step / t_ms number Order and millisecond time of each event

action label click, double click, right click, type, key, scroll, drag, wait

x / y / button number Pointer position and button of the action

text / keys text What was typed and which keys, with secrets removed

frame_before / frame_after image Lossless PNG at native resolution, or video at 10 fps or more aligned to events within 50 ms

ui_tree text DOM or accessibility tree at the step

target_bbox number Box of the element acted on

reached boolean Whether the goal was reached

judged_by / evidence label, text Who says it was reached, and the record in the system that proves it

redaction text What was masked on screen and in keystrokes, and how

Column	Type	What it must hold
`episode_id`	id	One person, one goal, one continuous attempt
`operator_id`	id	Pseudonymous id of the person, the same across their episodes
`application / app_version / platform`	label	Which business software, which version, web or which OS
`screen_w / screen_h / scale_factor`	number	Screen geometry, needed to trust click coordinates
`goal`	text	What the person set out to do, in words
`goal_source`	label	Written before by the operator, taken from a ticket, or back-filled by an annotator or a model
`step / t_ms`	number	Order and millisecond time of each event
`action`	label	click, double click, right click, type, key, scroll, drag, wait
`x / y / button`	number	Pointer position and button of the action
`text / keys`	text	What was typed and which keys, with secrets removed
`frame_before / frame_after`	image	Lossless PNG at native resolution, or video at 10 fps or more aligned to events within 50 ms
`ui_tree`	text	DOM or accessibility tree at the step
`target_bbox`	number	Box of the element acted on
`reached`	boolean	Whether the goal was reached
`judged_by / evidence`	label, text	Who says it was reached, and the record in the system that proves it
`redaction`	text	What was masked on screen and in keystrokes, and how

How much, in the lab's words

20,000 episodes, about 1,500 hours of active time (idle removed), across 25 or more applications and 200 or more operators, with at least 15% failed or abandoned episodes kept and marked. First purchase would be a paid pilot of 500 episodes.

episodes
events
frames
ui-tree
task
outcome
redaction