News for hackers. Stuff that matters.

zk108

Re:Perplexity launches AI agents for work on Friday November 06, 2026 @01:50PM (#50396611)
what is it that these lot do again?
Second this: creatine feels like such a hack, it helps me do more and stay focussed. Flying from SF to London and working across timezones has often messed my sleep often, I can cope so much better with creatine.
Re:The Compiler Is Your Best Friend, Stop Lying to It on Wednesday December 31, 2025 @04:26PM (#46448524)
dragon book >> tiger book
Re:Google Antigravity on Wednesday November 19, 2025 @03:58AM (#45977318)
Until they stop supporting it and rug builders using it in typical Google fashion
Re:Transpiler, a Meaningless Word (2023) on Thursday November 13, 2025 @08:22AM (#45914601)
"Programming languages are not just syntax; they have semantics too. Pretending that you can get away with just manipulating the former is delusional and results in bad tools."

So eloquently put, what starts off as just simple syntactic conversion usually snowballs into semantics very quickly.

Re:Algebraic Types are not Scary on Wednesday September 17, 2025 @08:03AM (#45274783)
...and a monad is just a monoid in the category of endofunctors of some fixed category

Just kidding but algebraic types are a great abstraction paradigm,

Re:Typed languages are better suited for vibecoding on Monday August 04, 2025 @05:12AM (#44783490)
*Typed languages are better

Fixed it

Show HN: Getting full-text scientific content into LLMs+Agents is stupidly hard

https://www.valyu.network/blogs/deepsearch-v2-updates

Most APIs don’t return actual content. You get metadata, maybe an abstract, maybe a snippet...never the thing itself. And if you want proper sources like arXiv, PubMed, or major publishers? Good luck. You’re stuck scraping tens of millions PDFs or semantic scholar and building your own ingestion pipeline.We hit this building agentic workflows and RAG backends. What we needed wasn’t “search”, it...

( Read More... | 2 comments | 3 points | on HN )
Re:Claude can now search the web on Friday March 21, 2025 @01:13PM (#43438349)
We’re giving away free credits to try out our platform — no card required. If you’re building with AI and need quality data, we’d love your feedback!
Re:DeepSearcher: A Local open-source Deep Research on Thursday February 27, 2025 @11:10AM (#43195678)
(founder here) We are working on that problem of providing deeper level of search especially on proprietary datasets (think reference works, books, papers etc.). Started off with Arxive papers )We are working on that problem of providing deeper level of search especially on proprietary/ copyright datasets (think reference works, books, papers etc.). We are working with a number of large publishers on this.

We started off with Arxive papers to test out the product- would love to get feedback :)

https://exchange.valyu.network/

Cofounder here- let us know what dataset provenance you want us to trace. Or datasets you need a hand curating- build highly performant models with responsibly sourced diverse, quality datasets quickly.