A: Hi there — great questions. Happy to clarify.

1.) Multiple businesses / workspaces
Currently, BrowserAct does not support fully isolated workspaces under one account. However, you can create separate workflows for each business and organize them accordingly. Many users manage different business pipelines this way. If strict separation is required, some teams use separate accounts.

2.) Training AI agents with prompts
BrowserAct is workflow-based rather than prompt-trained. Instead of “training” an agent, you design a structured workflow once and reuse it. This provides higher stability and repeatability compared to prompt-only approaches. Each node is configurable, and our documentation explains how to structure workflows step by step.

3.) Avoiding duplicate data
If duplicate entries appear on the same page, BrowserAct typically returns unique extracted results. Additionally, you can apply filtering logic inside your workflow, such as extracting only results that meet certain conditions such as specific dates, keywords, or identifiers. For more advanced deduplication, many users handle this in their database or automation layer after export.

4.) MCP integration
Yes, BrowserAct supports MCP. You can convert a workflow into an MCP tool and expose it via an MCP server. This allows other AI agents or automation systems to call BrowserAct as an execution layer within larger workflows.

You’re welcome to join our Discord community for more hands-on guidance and advanced setup discussions.

Please feel free to send over any additional questions anytime.