Guide 01
When do you need an agent with its own browser and computer?
See when an AI agent needs its own browser and computer, with a decision table, three practical workflows, and a test for whether persistence earns its cost.
Practical guides
A useful task.
A result you can check.
Choose your setup, give the agent a clear brief, and verify what comes back. These guides distinguish suggested workflows from recorded tests.
Guide 02
Run a browser agent locally or on a VM: which fits your work?
Compare running a browser agent locally or on a VM: privacy, availability, setup, private access, costs, persistence, and checks before your first task.
Guide 03
Temporary environments for troubleshooting and testing: a practical browser-agent workflow
Plan a temporary browser-agent workflow: reproduce faults, inspect server logs, preserve findings and authorize cleanup separately from paid infrastructure.
Guide 04
Monitor server health and logs from a separate agent computer
Use a separate agent computer to investigate server health and logs, with read-only access, redaction, practical checks, tested alerts and recovery planning.
Guide 05
Collect invoices from billing portals without giving an agent your passwords
Collect billing-portal invoices with human sign-in, limited access, a download ledger, duplicate checks and durable files—then verify totals before filing.
Guide 06
Get an evidence-backed competitor report from live websites
Compare competitors from live websites with dated source URLs, a repeatable evidence table, clear uncertainty labels, and a browser-agent prompt you can copy.
Guide 07
How to keep control when a browser agent needs you to sign in
Learn how to watch, message and take over a browser agent for sign-in, return control safely, and understand what account content may reach the model afterward.
Guide 08
Organise an inbox with a browser agent: a real supervised test
See BotHearth’s real Gmail test: 51 minutes across two runs, reported results, and a cautious workflow for labels, filters, privacy, human review and undo.
Guide 09
Audit your live website for AI-answer visibility and search discovery
Use a browser agent to audit crawl access, HTML answers and evidence, then check Search Console and measure AI citations without promising better rankings.
Guide 10
Choose a model for browser work and recover when a task stalls
Choose a model for browser work by task, cost and account access. Learn how to switch connections, save results and recover stalled BotHearth tasks safely.