Live data from Hacker News

Reduce your LLM agent costs by 90% with structure-preserving HTML compression

github.com

1–10 of 13 posts

Re: Reduce your LLM agent costs by 90% with structure-preserving HTML compression

#2
I built Emmetify to solve a costly problem in LLM automation: when your agents need to process HTML (like for web automation or scraping), they waste tokens on all those and attributes.

The solution converts HTML into a compressed format that preserves the complete DOM structure. This means your LLM agents can still generate accurate XPath/CSS selectors, click buttons, fill forms, and follow links - just at 1/10th of the token cost.

It works by converting HTML into Emmet notation - a concise abbreviation syntax widely used in web development. LLMs understand this syntax just as well as HTML due to its presence in training data.

Post reply on HN