No more diving into complex DOM structures or writing fragile XPath expressions. Just specify what data you are scraping from the web with natural language-like queries, and AgentQL handles the rest.
Anima - OnBrand Vibe Coding — Design-aware AI for modern product teams.
Design-aware AI for modern product teams.
Promoted
Oh this is sick! I can see lots of use cases to run E2E testing - my QA team will love this. How accurate are the results? I know on the website it says there is no black box, I wonder how does prompt engineering work together with DOM selection in this case?
Congrats on launching this awesome product @shuhao_zhang and team!
@shuhao_zhang@tonyhanded Hi Tony! Yes E2E testing is definitely a strong use case for this.
The accuracy is dependent on the page complexity and the query quality, but has been consistently very high from our internal benchmarks. One of our key recommendations is to use semantic context https://docs.agentql.com/agentql... to optimize your queries. The more specific your queries are, the more accurate and consistent your responses will be. We have several tools listed on the website to help you get started if you want to test the accuracy out yourself!
@tonyhanded My pleasure! Let me know if you have any other questions 😊
Report
AgentQL sounds like a game changer for anyone who’s tired of the headache that comes with web scraping! 🙌 The idea of just using natural language to interact with elements is genius! No more wrestling with fragile selectors, which always seem to break when you least expect it.
I'm really curious about how well it adapts to different site structures—does it handle them pretty seamlessly? The thought of writing less rigid code is definitely appealing. Plus, the integration of advanced prompt engineering is intriguing. I can see this being a real productivity booster, especially for teams who rely on frequent web data extraction.
Kudos to you, @shuhao_zhang, for taking on the challenge of traditional web automation limitations! Can’t wait to see how AgentQL evolves and the new use cases it will unlock!
@shuhao_zhang@blankwebdev Short answer: yes.
Supporting reusable queries for varying site structures was definitely a focus of ours. You are right in that many websites do have significantly different layouts while providing similarly structured content - think e-commerce, travel, etc. With the right query, we very are capable of handling these seamlessly. If you have any questions or need help building such a query for your use case, feel free to reach out to us on Discord! I'd be happy to help
This is great, proud of the team, love the concept.
It is helping with something that LLMs are so great at.
Small changes, in this case website/UI, make it painful for humans to manage.
AgentQL does this beautifully.
And that interactive playground is so cool. You just get it. In 10sec or less.
@jakub_czakon Thanks so much! We’re really proud of the team too, and it’s awesome to hear that you love the concept. You’re absolutely right—LLMs are fantastic at handling those small but impactful changes, and AgentQL is all about making that process smooth and effortless.
And we're glad you’re enjoying the playground! Our goal was to make it intuitive and quick to grasp, so it’s great to hear that it clicked for you in just 10 seconds. Thanks for the support!
Congrats on the launch @shuhao_zhang. I tested a lot of data extraction tools tailored on web scraping but AgentQL seems different, fresher, more modern!
@gdc Really appreciate the kind words. Glad to hear it’s making an impression! Looking forward to hearing more about your experience with it.
Report
Hi @shuhao_zhang This is very impressive, I hope it can help users and have a positive impact on its users, and congratulations on the product launch 👏🚀
We benchmark regularly and run the same use-case multiple times in succession to ensure we are measuring consistency as well as accuracy. It's one of the areas I'm most impressed with AgentQL, actually, as we actually score very well on consistency in our evaluations, meaning our results are reliable and repeatable! We also make sure to measure other aspects, such as how often we may be providing answers which are non-existent on the page (indicating a hallucination) in addition to just raw result comparison for accuracy.
As for plans to expand our SDK, definitely! We've got a JS SDK in the works, even as I'm typing this, and would love to hear more from the community on what other areas to support next!
@nan_mtz We are planning to open source the Python SDK and excited to work with the community to bring AgentQL onto more languages!
Report
Great job, guys! I tried using it couple of times — works like magic) I think you should definitely target large companies with market intelligence departments — my working experience (I've been working at world's largest classified ads company and world's top-3 job board company) tells me that your product would be extremely valuable there.
Thanks for your advice, Denis! We'll definitely be a valuable tool in a toolset of such companies. With this idea in mind, we would be grateful if your colleagues could also try us out for their market research and we can see how we can make your life easier!
@denis_shchigelskiy Thanks for the feedback and the kind words! We're glad to hear it’s working well for you. Targeting large companies with market intelligence departments is definitely on our radar—your insights confirm that we're heading in the right direction. Appreciate the suggestion!
AgentQL
AgentQL
AgentQL
AgentQL
Developer marketing examples
AgentQL
AgentQL
panda{·}etl (YC W24)
AgentQL
AgentQL
AgentQL
AgentQL
AgentQL
AgentQL
AgentQL
AgentQL