Ci bi mu melGis bi ci topp
Reflexion ak Agent yuy jubbanti seen bopp
Aplikaasioŋ yi
GUIDE ci aplikaasioŋ yi
Computer-using agents interact with graphical interfaces through screen observations and input actions such as clicks, typing, and scrolling.
Déglul tegtal biiIA-nettali · ~2 min
They can target some applications without a dedicated API, but performance depends on access, visual grounding, and task complexity.
A computer-use agent receives a screenshot or other supported interface observation, selects an action such as click, type, or scroll, and observes the resulting screen before acting again. Benchmarks such as OSWorld test open-ended tasks across real desktop and web applications in controlled operating-system environments. Some systems can also use accessibility metadata, which supplies structured properties of interface elements. Visual grounding and action execution can fail, and working through a GUI does not make every application accessible or every task reliable. Use permission limits, verification, and human approval for sensitive actions.
Navigate a web form in a test environment and verify the resulting page state.
Use a desktop benchmark task to edit a document and confirm the file was saved.
Combine screenshots with accessibility metadata when an environment makes it available.
Run actions in a restricted environment and require confirmation for consequential operations.
Computer-use systems and benchmarks continue to evolve. Compare models on the operating systems, applications, observation modes, and action space relevant to the intended task; benchmark scores do not establish universal task success.
Bulletin bu amul fayda
Ñatti jaar-jaari IA yuñ firndeel bis bu nekk ci ayu-bis bi, ñu bind leen ci Àngle bu leer. Gratis ba fàww, amul yëgle.
Benn email bis bu nekk ci ayu-bis bi. Benn klik nga dindi abonemaa bi. Dunu musa jaay wala séddoo sa adres.
Nattal sa bopp
Feedback ci saasi ci bépp tontu, ak sertifikaa buñ mëna séddoo ak ID buñ mëna saytu soo jàllee ci njàng mi.
Jàppale njàngum IA te doo fay. AI Understanding mbootaay la buy def te Yàlla tax — amul yëgle, amul paywall, musul am. Defal ab sarax
Computer-using agents interact with graphical interfaces through screen observations and input actions such as clicks, typing, and scrolling. They can target some applications without a dedicated API, but performance depends on access, visual grounding, and task complexity.
CUA dafay gis nataali ekraŋ yi ba noppi def ay jëf yu niro ak yu nit ñi di def, lu ci melni klike ak tëye butoŋu.
Ndax dafay def ni jëfandikukat nit, CUA mën na dawal aplikaasioŋ yu yàgg wala yu amul API.
Ginaaw jëf bu nekk, ndawu liggéey bi dafay jël nataalu ekraŋ bu bees, daal di nekk benn bouclage perception-action.
Jëfandiku ordinatër ak Operatëru AIU_AAR 10_ dañu siiw ci jëfandikoo ordinatër.
Jox ab ndawu liggeykat mu yor ab ordinatër dëgg amna risk, kon isolement ak surveillance dina wàññi loraange yi.
Weyal di jàng
Tann nañu yeneen njiit ngir topic bii
Ci bi mu melGis bi ci topp
Reflexion ak Agent yuy jubbanti seen bopp
Aplikaasioŋ yi