Hey everyone! I'm looking to create a custom browser agent similar to **GPTOperator**, something that can run a lot of automated tasks over extended periods. The idea is to set up an agent that can function like an automatic job application tool, accessible by over a thousand users at the same time. I've looked into services like Browserbase and Steel, but their custom plans only allow for about 100 concurrent sessions, which isn't nearly enough for what I need.
I'm especially interested in how to deploy this system for 1000+ concurrent users while integrating an AI loop, since current options don't seem to handle this well. I've also considered using alternatives to GPT Operator because it's generally too broad and expensive, and I'm exploring cheaper models for implementing the agent's flow. If anyone has suggestions for open source options or guidance on using Cursor or similar tools, that would be fantastic!
1 Answer
Running browsers in headless mode is a solid start! It sounds like Browserbase's limitation is more about cost and resource management rather than technical hurdles. Each headless instance consumes a fair amount of memory and computing power. So, if you're scaling for thousands of users, you'd need to invest in substantial resources.
Right now, many organizations are tackling similar problems, and it’s tricky to create an agent that dynamically searches with complex constraints. If your tool amplifies noise, you're likely to face hurdles getting through to job applications since many platforms might begin filtering out mass submissions. Just a heads up!
Got it, thanks for the insight! I understand the limitations with headless browsers now. It looks like I might need to consider custom infrastructure to handle the scale I want. It's all pretty complex, but I'm determined to find a way to make it work.