Saltar al contenido
Atlas de IA

Agentes que operan computadoras

Ver la pantalla, hacer clic y escribir como una persona

Agentes y uso de herramientasIntermedio #37
entradaTextoAcción

El texto completo se presenta en inglés; el título y el resumen están traducidos.

QUÉ SIGNIFICA ESTA CAPACIDAD

Takes a task to be done through a user interface — add this column to the sheet — and outputs a sequence of mouse and keyboard actions, with screenshots as the observation at each step. It needs no software API, so it can drive legacy systems that expose none. Unlike an autonomous agent its actions land on a graphical interface rather than code or APIs.

Cómo se consigue técnicamente

The model treats screenshots as observations and emits actions such as click coordinates, scrolling and typing, advancing in a repeated screenshot–action loop. Training data comes from recordings of human operations and trajectories annotated on synthetic tasks, and evaluation runs on task sets over real websites and desktop apps. To reduce risk, deployments usually run in isolated virtual machines or constrained accounts with confirmation points on sensitive actions.

Productos representativos

5

Organizaciones relacionadas

Usos típicos

  • Testing and accepting web flows
  • Data entry into legacy systems without APIs
  • Automating repetitive back-office operations
  • Operating interfaces on behalf of users with accessibility needs

Límites y dificultades

  • Clicks based on screenshot pixel coordinates miss as soon as resolution or page zoom changes
  • Captchas, pop-ups, long scrolling and lazy loading make it stall or repeat ineffective actions
  • With no reliable intermediate feedback, a wrong click corrupts real accounts and data, so isolated execution is mandatory

Conceptos detrás