Give Agent a task to complete.

Give Agent an outcome

Agent completes supported computer tasks from a LowLighter conversation. You describe the result you want, and Agent works through the steps in its own browser workspace or with the supported files needed for the task.

Research and writing

Read relevant pages, collect useful facts, and turn them into a document with the structure you requested.

Documents and presentations

Create or update Google Docs, Slides, spreadsheets, and other supported browser documents.

Forms and routine work

Enter information, move through browser pages, and complete supported multi-step tasks.

Files and PDFs

Open, review, complete, save, upload, or download supported files when the task calls for them.

Follow the work from the Agent overlay

The Agent overlay shows the conversation, current activity, and completed result. You can write a follow-up, correct the request, or stop the task without moving the app you were using into the background.

Agent can continue working while you use another window. The visible Agent overlay remains available when you want to check progress or provide more direction.

Choose how independently Agent works

Full access lets Agent proceed through supported actions within your request. Ask before running asks for your approval at the points covered by that setting. You can take over directly whenever a website needs private sign-in details or another personal step.