Verified purchaser
Could be good but needs more work
BrowserAct could be great if they implement the following:
1. Allow users to specify specific pieces of data, from specific DIVs/elements on a page, targeting by css selector or xpath. Most competitors have this. The "ai just explain what you want" thing doesn't work all the time, and it's frustrating if you are a technical user who can see the data you need is always in a particular div class if you are scraping pages that follow a similar template.
2. Need a way to tell the ai "do not make anything up on your own if you can't find the content I want you to scrape". Sometimes on pages I was scraping where a field was blank or not found, it would literally write fake content that was similar sounding but not actually on the real live page... This is bad.
3. Need a way to start with an ai agent where you describe what you want to scrape, but then convert to a workflow mode where you have control over each specific step in the drag and drop builder. The team said they are working on this.
4. Finally, they badly need a real help center site with tutorials and documentation instead of the janky blog on their website, and a better system than discord for handling support. Discord is noisy with a bunch of users complaining about their very specific individual problems, and support was not super effective when I tried to open a ticket with the discord ticket bot.. Someone replied but they never really fully answered my question or understood it. Seemed like they didn't have time to read the full problem summary I wrote.
Overall BrowserAct feels like a somewhat quality product, nice easy ish to understand, but they have some work to do to be ready for prime time! Hope to be able to use this product in the future.
Claire_BrowserAct
Nov 10, 2025Thank you so much for taking the time to provide this incredibly detailed, valuable, and constructive feedback on BrowserAct! We are genuinely encouraged that you see it as a "somewhat quality product" and are hopeful about its future, as your insights are critical for helping us get "ready for prime time." To address your excellent points directly:
(1) We completely agree on the need for element-specific targeting. We are currently planning a visual element selection feature that would allow users to specify data from DIVs/elements, and we wanted to ask if this approach (visual selection, perhaps with CSS/XPath support) aligns with your expectations.
(2) You have highlighted a critical issue with AI hallucination, which is a continuous challenge we are actively working to solve. To help us pinpoint this, could you share if you experienced the "fake content" issue primarily in the AI Agent mode or the Workflow mode? We are prioritizing improvements to force the AI to return a null value rather than fabricating data when it cannot be found.
(3) We are thrilled to confirm that the "AI Agent to Workflow" conversion mode you suggested is already in active development, as we agree it's essential for combining ease of use with granular control.
(4) Finally, your feedback on our documentation and support system is spot on; we know this is a major weak point. We are accelerating the development of a proper, structured help center to replace the blog and are simultaneously evaluating better support systems to manage technical tickets more effectively than Discord allows.
Thank you again for this incredibly helpful review; it provides a clear confirmation of our development priorities.