Crawl AI – A Web-Based Platform for Building Custom AI Assistants

[Maks2204]

https://www.crawlai.dev

[Maks2204]

Hey HN, I’ve been working on Crawl AI to simplify creating AI assistants by combining web crawling, scraping, and model training in one platform. Would love your thoughts or feedback!

[jasonthorsness → Maks2204]

I went ahead and made an account through the mystery login. Clean interface. It asks "What do you want your assistant to do?" - I said "It recommends the best articles for me on hacker news" - now it is reading some blogs about Hacker News, and seems stuck on "Building". How long is that supposed to take? UPDATE - it finished one BUILDING and is on to another, so I guess the answer is "many minutes"... but now I wonder when the whole thing will be done?

[jasonthorsness → jasonthorsness]

It eventually completed! But, when I asked it to recommend me an article, it recommended "https://www.recombee.com/blog/breaking-the-news-the-role-of-..." which has never appeared on hacker news. So, I don't know ¯\_(ツ)_/¯.

My main takeaway from this is the art of displaying progress for long-running tasks needs to come back for a lot of these long-running LLM-powered workflows.

[lxchase → Maks2204]

Is it supposed to go straight to a log in page? If that is your intention, I don't believe people will explore further without any landing page information about your service.

[flessner → Maks2204]

Feedback 1: Have a landing page?

I am not signing up for something I don't know. Before "committing" I want to see capabilities, documentation, pricing. Where's all that?

[uglygoblin]

Feedback: The landing page should include details about what this is and not just a login modal.