It uses the latest and best computer use model, gemini 2.5 computer use. There is no better computer use model right now, and you can try it out free.
Sign up now and get 1000 free credits to try it out.
What BrowserForge does differently is run agents in real Chrome instances and focus on goal-oriented automation. You describe what you want done (as you would to a teammate), and the agent navigates the site like a human would, clicking, typing, selecting dropdowns, waiting for dynamic content, extracting structured data, and completing multi-step flows.
A few things we’re building around: - Persistent authenticated sessions: log in once, reuse saved cookies/auth state for future runs. - Monitoring + reliability: agents can retry steps and handle common UI variability (and when something needs human help, that’s surfaced instead of silently failing). - Integration-ready: trigger runs and receive results via API + webhooks.
Concrete examples people use this for: - Log into an admin portal, download a daily report, extract key numbers, and forward them to your systems. - Monitor a marketplace for new listings / price changes / removals and alert when conditions are met. - Fill out multi-step forms (applications, record updates, ticket creation) inside authenticated SaaS tools.
It’s early, and it won’t work on every site without friction (things like 2FA/CAPTCHAs can require manual intervention), but for a lot of browser-based ops it’s already useful.
I’d love feedback from HN on: What browser workflow would you automate first? Where do existing approaches (Playwright/RPA/etc.) break down for you? What would you want for debugging/auditability when an agent runs? If you want to try it, it’s here: https://browserforge.ai (CTA: Get Started Free).
grantsingleton•1mo ago
I use BrowserForge to automate getting the latest research papers onto my other project arxivexplained.com.
You can auth a browser and the AI will be able to use that authenticated session.
There is an AI prompt generator so you can just describe what you want it to do and it will write a prompt that brings in all the tools needed.
You can run the browser agents on a schedule, from a trigger, or manually.