Browser-Use Agents: What They Can and Can't Do — An Honest Guide
Blog
Guide
"Just Browse Any Website" Sounds Too Good. Is It?
Browser-use is one of OpenClaw's most powerful capabilities. It controls a real browser session — clicks, scrolls, fills forms, reads text. No API needed. But it's not magic. Here's an honest breakdown.
What Works Great
Data collection from any website: @andrewjiang collected 4M posts from 100 top X accounts. The agent handled pagination, rate limiting, and extraction automatically. Source ↗
Booking and check-in flows: @armanddp automated flight check-in — finding the flight, completing the form, selecting a window seat. Source ↗
Form filling and submissions: @avi_press filed an insurance claim through the portal using natural language. Source ↗
Shopping and ordering: @dreetje automated grocery ordering with saved credentials and 2FA. Source ↗
Price research and negotiation: @astuyve saved $4,200 on a car via multi-channel automated negotiation. Source ↗
What's Tricky
CAPTCHAs: Simple ones get handled. Advanced reCAPTCHA v3 with behavior scoring can block sessions. This is evolving.
Heavy JS SPAs: Some single-page apps with complex client-side rendering are harder to navigate reliably.
Aggressive bot detection: LinkedIn actively detects automation. Browser-use works but requires careful session management.
Real-time interactions: Video calls, live chats, streaming — not great fits. The agent works best with static or semi-static pages.
When to Use Browser vs. API
Use Browser When | Use API When |
|---|---|
No public API exists | A reliable API is available |
Logged-in sessions needed | High-volume operations |
Visual navigation required | Data consistency matters |
Scraping browser-only data | Speed is critical |
The Bottom Line
Browser-use excels at structured tasks — forms, data extraction, booking flows. It gets less reliable with highly dynamic or bot-hostile environments. For most business automation, it fills the gap that APIs don't cover. With StartClaw, it's available out of the box.