Hacker News

Top stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. OpenAI GPT–6 Astra breaks Enigma message that has resisted solution since 2005(cryptocellar.org)
    260comments
  2. Claude Opus 5.5(anthropic.com)
    89comments
  3. 16-bit Intel 8088 chip(allpoetry.com)
    2comments
  4. OpenAI is well positioned to fast-follow Jev(arcturus-labs.com)
    88comments
  5. Writing Rust code that's fast by asking agents to make the code faster(minimaxir.com)
    9comments
  6. Apple has added persistent 'ads' to iOS, and it's driving users crazy(techradar.com)
    219comments
  7. Show HN: Drop – a rootless Linux sandbox with gVisor support(droprun.sh)
    25comments
  8. Show HN: AI·rete·RAG – a Rete rule engine decides, RAG explains why(ai-rete-rag.com)
    discuss
  9. Solitaire Alone Together(solitairealonetogether.com)
    9comments
  10. Can gzip be a language model?(nathan.rs)
    120comments
  11. MiMo v2.6(xiaomi.com)
    460comments
  12. I asked Meta’s Muse for its filesystem and it sent me 6.8GB(mouse.dev)
    75comments
  13. Spymarks, Not Watermarks(brand.io)
    154comments
  14. Show HN: InstinctFlash – Run 5B world-action models in real time on Jetson Thor(github.com/general-instinct)
    discuss
  15. Xbox continues its "reset" with dramatic restructuring(arstechnica.com)
    8comments
  16. Muse, Meta's extraordinarily privileged AI assistant, has a serious 0-day(arstechnica.com)
    26comments
  17. Vacate a Drone Restriction That Criminalized Recording Immigration Agents(eff.org)
    7comments
  18. Jev – A curation of Jev demos on X, tools, skills, and integrations(github.com/amal-david)
    22comments
  19. Relativistic Raytracing(publish.obsidian.md)
    discuss
  20. Line Scan Photos from MUNI Heritage Weekend in San Francisco(lawrence.lu)
    17comments
  21. Attention is all you have(alicegg.tech)
    299comments
  22. Teleoperated Humans(jefftk.com)
    22comments
  23. The Economics of Open-Weight Inference(ornn.com)
    6comments
  24. Transformers Explained Visually(poloclub.github.io)
    84comments
  25. What Sun got wrong(dtrace.org)
    379comments
  26. I said no and Apple said yes(dbushell.com)
    510comments
  27. Americans' Drinking Remains at Record Low: Gallup Poll(gallup.com)
    discuss
  28. I don't want to read what you didn't write(colinbreck.com)
    387comments
  29. AMD's random number generator can't generate a 0?(flatassembler.net)
    146comments
  30. A font that reads what you wrote(rohanadwankar.github.io)
    38comments

Show HN: A writing coach that's not allowed to write your sentences

2 pointsby 6d agoapp.oghmere.com
0 comments
Hi HN,

I'm the technical advisor and one of the co-founders for Oghmere (https://oghmere.com), a writing tool for people who care about the integrity and authenticity of their writing, including journalists, essayists, newsletter writers. Along with inkandops and notgrrly, we are bootstrapping this project to make sure we can focus on building AI tools that benefit authors without replacing them. It is very easy today for anyone to fire up a chatbot or other AI tool to write for them, and it’s a really useful tool that I use every day at my $dayjob, or to summarize and review topics I am researching or working on, but those generated texts aren’t something I care about the authenticity because they are steps to getting a job done. When I write my own opinions, fiction, or D&D content, I don’t want to use generated text, I want to exercise my mind and creativity.

That is the constraint we care most about in Oghmere: the AI is not allowed to generate, autocomplete, or rewrite your prose. It can ask questions and point at things in a draft you already wrote, it can even suggest changes you can make, but that's it. This isn't a limitation of a capability - we use Anthropic models, and are experimenting with other providers as well; they are fully capable of writing the sentence or generating text for you, but building that authenticity and working to retain it is the core tenet of the product and platform that we enforce.

If you want to know how it actually holds together: there's an authoring pipeline: Voice (built from writing that you provide, or an interview to define a target Voice) → Brief → Outline → Editor Each stage is optional, but they all yield structured data from author interactions that steers the guidance that the coaches provide.

We enforce the constraints on AI authorship very simply; the AI tools that can provide coaching and feedback don’t have a mechanism to update the document directly, they can only provide review comments and feedback. There is a tooling surface for deterministic reviews like spellcheck and grammar check to recommend and apply changes, but if an author wants to leverage AI feedback, they have to rewrite their prose themselves.

In addition to the coaching and feedback, we are also building in research tooling, and project and document knowledge bases, and AI powered author feedback that assess how you are writing and where you can improve. We also have a weekly writing challenge that allows authors to stretch and grow!

Writing, editing, version history, and export are free and unmetered, with some AI powered wizards that can help customize and tailor the experience. There are no direct charges for basic editor features, but all AI and some other capabilities consume credits, all of which are indicated in the UX, examples include coaching passes, a voice-calibration pass, a research query. We hope to change the dependency on credits in the future, but we are still working out the longer term billing/pricing model.

You can try it without signing up using our test page (metered to 5 coaching passes): https://app.oghmere.com/try/HACKERNEWS2026 After you try it out, or if you want to skip trying it first, you can sign up for a free account (no card required) that includes a bundle of free credits at https://app.oghmere.com/campaign/HACKERNEWS2026

We would really love your feedback as we speed towards a general release, and we are happy to answer questions about how it was built and what we are planning for the future!

Yvan (ygjb), Monique (inkandops), and Sandhya (notgrrly)

A quiet thread, for now.Start the conversation on HN ↗