If Browse AI struggles on a site: test Reddit, Facebook groups, and other bot-hardened targets with a small run before committing credits, users report high failure rates and gibberish output there. When a run errors repeatedly, try the other execution mode (extension vs AI Studio) before assuming the robot is misconfigured. And keep expectations realistic on aggressive anti-bot sites, no-code does not mean no-blocking.

Context: Web (Trustpilot, Browse AI reviews): real users report scraping reliability problems on hard targets. One reviewer found Reddit pages and Facebook groups either failed outright or returned gibberish (strings of random characters) in both the Chrome extension and AI Studio modes. Another reported errors on 98% of attempts, with successful runs only scraping the first bit of info. These are user reports, not official docs, so treat them as field experience.