TipAgents that click for you: AI that uses the browser
AI can now control a web browser like a person: clicking, filling in forms and gathering information. Here is how to use it safely.
Both OpenAI and Anthropic now offer “computer use”: the AI model looks at the screen and clicks, types and scrolls. OpenAI runs the browser in its own hosted environment and requires the user to approve every new website the agent wants to visit.
Anthropic’s version runs in an environment you control and can operate the whole desktop, not just the web. For web-only tasks there is a separate browser tool.
There is also open source: Browser-Use is free (MIT licence) but requires technical knowledge.
For everyday users this shows up in products such as Perplexity Comet and agent modes in ChatGPT and Gemini. Sensible tasks: comparing prices, collecting quotes, filling in long forms.
How to get started
- Start with a harmless task: “Compare the price of X in three shops and make a table.”
- Only approve websites you recognise – the agent should ask first.
- Never let the agent complete purchases or log in with your bank ID without you present.
- Review the result; agents still make mistakes on complicated sites.
More on TipAI
Sources
- OpenAI: Computer use (Agents API) (retrieved 2026-10-15)
- Anthropic: Computer use tool (retrieved 2026-10-15)
- AI Agent Rank: Browser-Use vs Operator vs Computer Use 2026 (retrieved 2026-10-15)
TipAI explains AI in plain language for beginners, families and small businesses. Every article has sources and a "How to get started" section.