Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Hister(github.com/asciimoo ↗)
    discuss
  2. Show HN: Claude and Codex rate limits as AIM away messages(diewithme.co ↗)
    discuss
  3. Aesthetic Governance, Mega-Events and the Urban Poor(doingsociology.org ↗)
    1comments
  4. The Shortcut Became the System(christophermeiklejohn.com ↗)
    discuss
  5. Classic ML to Cope with Dumb LLM Judges (2025)(softwaredoug.com ↗)
    discuss
  6. Launch HN: Skillsync (YC W26) – AI chat sessions made portable across agents(skillsync.com ↗)
    discuss
  7. Show HN: Paste a song and get 5 license-free cousins that sound like it(mockfreeli.org ↗)
    discuss
  8. Show HN: Lightning strike VFX in Three.js, a step-by-step breakdown(threevfx.com ↗)
    discuss
  9. AODM: AI Optimized Data Markup(github.com/fucaspark ↗)
    discuss
  10. There Is No AI Agent(zak.im ↗)
    discuss
  11. Show HN: A toy implementation of ELF relocation weird machines(github.com/scriptod911 ↗)
    discuss
  12. We Need a Way to Prove Personhood Online(noemamag.com ↗)
    discuss
  13. Financial Times – The era of AI warfare has arrived(ft.com ↗)
    1comments
  14. Zuckoff app detects nearby Meta 'perv glasses'(nypost.com ↗)
    discuss
  15. From Specialist Agents to Distributed Skills over MCP(devblogs.microsoft.com/agent-framework ↗)
    discuss
  16. EU president: AI agents escaping their environment preview what's coming(the-decoder.com ↗)
    discuss
  17. Ask HN: How to recover Google auth after phone stolen?
    2comments
  18. A Taxonomy of Toddler Jokes(lastwordonnothing.com ↗)
    discuss
  19. Jev's Architecture Unmasked(archerhume.com ↗)
    discuss
  20. God Eye View(spatialintelligence.ai ↗)
    discuss
  21. Modern Monetary Theory: Economics for the 21st Century (MOOC)(mmted.org ↗)
    discuss
  22. How we made one of our largest inference workloads 4.7× more GPU-efficient(decagon.ai ↗)
    discuss
  23. AI has transformed The Pentagon's aging networks into a national security risk(washingtonpost.com ↗)
    1comments
  24. Show HN: Stroq – trace your agent's commands back to the text that produced them
    2comments
  25. ZSA Backpack(zsa.io ↗)
    discuss
  26. Designing for Amnesia(kasrin.com ↗)
    discuss
  27. Sub-ms deterministic parsing vs. LLM-based policy for agent safety?(github.com/midhunweb ↗)
    discuss
  28. The suddenly explosive world of AI safety(theverge.com ↗)
    discuss
  29. Agentic Societies Need a Social Harness(social-harness.org ↗)
    1comments
  30. We Need a Science of the AI Mind(wsj.com ↗)
    1comments

The AI graveyard: a running list of projects and startups that didn't make it

1 pointsby 1h agotechcrunch.com
1 comments
1h agoHN ↗

I think alot of companies I trying to figure out where the actual business value is.

Most business spend a lot of money on making processes reliable not probabilistic.

I think most business process probably need something like:

text/image -> classifier -> events -> workflow -> template

                         -> exception -> human

then text/image -> llm/agent loop -> creative text.

with programming the model used to be

non-coder -> vague AI slop spec -> programmer/questions -> QA -> Prod

now it's:

AI prototype -> execs -> wow -> non-coder -> vague spec -> AI slop Code -> QA/Verify -> missed requirments/lacks business context, etc. -> programmer spends just as much time fixing.