andrew_zhong

Systematic testing of browser agents today is not easy: testing on real websites is flaky, rate-limited and potentially expensive (e.g. using proxies or bypassing Captcha), while static-HTML benchmarks lack state and…

HTML-to-Markdown converters produce clean, readable content for both humans and LLMs — but the DOM structure is lost along the way. You can always feed Markdown to an LLM to extract structured information, but that…

We've been building data pipelines that scrape websites and extract structured data for a while now. If you've done this, you know the drill: you write CSS selectors, the site changes its layout, everything breaks at…

While working with LLMs for structured web data extraction, we saw issues with invalid JSON and broken links in the output. This led me to build a library focused on robust extraction and enrichment: - Clean HTML…

We created LightFeed to transform any websites into lightweight and focused news feed. Use LLM to sort and summarize posts using user's custom prompt. For example, you can get anything related to "Games, science or open…

The new Llama 3 8b in on par with 22b models, better but could be 10x cheaper than GPT3.5 AI builders, if you are using Llama 3 in backend, where do you host it or what API do you use? (For production usecases with good…

Hi folks, I am working on an AI newsfeed app. It uses LLM to summarize and filter stories on any website or RSS. When exporting this custom AI feed to custom RSS (so that users on RSS readers can subscribe them), I am…