Hi HN,
I built a CLI for uploading documents and querying them with an LLM agent that uses search tools rather than stuffing everything into the context window. I recorded a demo using the CrossFit 2025 rulebook that shows how this approach compares to traditional RAG and direct context injection[1].
The core insight is that LLMs running in loops with tool access are unreasonably effective at this kind of knowledge retrieval task[2]. Instead of hoping the right chunks make it into your c
Article URL: https://arxiv.org/abs/2505.24363
Comments URL: https://news.ycombinator.com/item?id=44291204
Points: 6
# Comments: 0
Article URL: https://pubmed.ncbi.nlm.nih.gov/30101594/
Comments URL: https://news.ycombinator.com/item?id=44290582
Points: 61
# Comments: 65
Article URL: https://angad.me/blog/2025/getting-free-cruise-internet/
Comments URL: https://news.ycombinator.com/item?id=44291630
Points: 63
# Comments: 107
Hello HN!
I've been working on Canine for about a year now. It started when I was sick of paying the overhead of using stuff like Heroku, Render, Fly, etc to host some web apps that I've built. At one point I was paying over $400 a month for hosting these in the cloud. Last year I moved all my stuff to Hetzner.
For a 4GB machine, the cost of various providers:
Heroku = $260 Fly.io = $65 Render = $85 Hetzner = $4
(This problem gets a lot worse when you need > 4GB)
The only downside
Article URL: https://www.nature.com/articles/d41586-025-01880-9
Comments URL: https://news.ycombinator.com/item?id=44292342
Points: 15
# Comments: 0
Article URL: https://arxiv.org/abs/2506.11016
Comments URL: https://news.ycombinator.com/item?id=44290315
Points: 28
# Comments: 11
Article URL: https://coroot.com/blog/opentelemetry-for-go-measuring-the-overhead/
Comments URL: https://news.ycombinator.com/item?id=44290331
Points: 10
# Comments: 4
https://coroot.com/blog/opentelemetry-for-go-measuring-the-overhead/