Hacker News

Article URL: https://github.com/auxten/clickmem

Comments URL: https://news.ycombinator.com/item?id=47259913

Points: 1

# Comments: 0

Claude is honestly excellent for building, but when it comes to content, explanations, or digging into information, I still often prefer ChatGPT.

And constantly switching between models is still annoying.

Have you found project orchestration solutions for this problem? How do you handle it ?

My current stack: VSC COde with Claude code plugin + Obsidian + Chatgpt (copy and paste) for the content I don't like.

Comments URL: https://news.ycombinator.com/item?id=47259891

Points: 1

# Comments: 0

Article URL: https://updown.io/

Comments URL: https://news.ycombinator.com/item?id=47259885

Points: 1

# Comments: 0

Article URL: https://www.reuters.com/world/asia-pacific/china-vows-accelerate-technological-self-reliance-ai-push-2026-03-05/

Comments URL: https://news.ycombinator.com/item?id=47259881

Points: 2

# Comments: 0

I built a private desktop app that shows you what you actually did during a work session, whether it matched your intent, and the behavioral patterns behind how you worked. I used to reflect on my work from memory, but memory is unreliable. I'd think I spent an hour on a task when half of it actually went to jumping between tabs, checking messages, and re-reading the same pages. I'd tell myself the reason was that I was tired, when in reality the task just wasn't clearly defined yet. How it works: you set your intent before a work session. The app silently tracks your window activity (no screenshots). You can log how you're feeling at any point, on your own terms, no notifications, no interruptions. When you're done, AI generates a behavioral report ,not just what you did, but why you worked the way you did, with actionable suggestions to improve next time. Privacy was non-negotiable. Your data never leaves your machine. Everything is stored locally in SQLite. No servers, no accounts, no cloud. The only external calls are to Google Gemini to check your intent for clarity and to generate your report. It receives your intent, window titles, app names, and feeling logs. No file contents, no screenshots. Built with Electron, React, TypeScript, SQLite, and Google Gemini.Free on macOS. I'd love your feedback.

Comments URL: https://news.ycombinator.com/item?id=47259872

Points: 1

# Comments: 0

Article URL: https://lapcatsoftware.com/articles/2026/3/1.html

Comments URL: https://news.ycombinator.com/item?id=47259861

Points: 2

# Comments: 0

Hi HN,

We built Notation by Fownd to address the 'documentation tax' that physical therapists pay after hours.

Most AI scribes are generic transcribers. We’re focusing on clinical reasoning — capturing the therapist's intent in real-time to build structured, compliant SOAP notes during the session.

Key Focus Areas: - Real-time SOAP Alignment: Structured notes that follow medical standards. - Clinical Reasoning Capture: Moves beyond simple speech-to-text. - Niche-Specific Logic: Designed by PTs to avoid generic medical templates. - Privacy & Compliance: Built for outpatient and home health security.

I’m happy to answer any questions about the tech stack or the clinical logic we're using!

Comments URL: https://news.ycombinator.com/item?id=47259848

Points: 1

# Comments: 0

Hi HN,

I’m Padam, a developer based in Dubai.

Over the last 2 years I’ve been experimenting with the idea that AI inference might not require GPUs.

Modern LLM inference is often memory-bound rather than compute-bound, so I built an experimental system that virtualizes GPU-style parallelism from CPU cores using SIMD vectorization and quantization.

The result is AlifZetta — a prototype AI-native OS that runs inference without GPU hardware.

Some details:

• ~67k lines of Rust • kernel-level SIMD scheduling • INT4 quantization • sparse attention acceleration • speculative decoding • 6 AI models (text, code, medical, image,research,local)

Goal: make AI infrastructure cheaper and accessible where GPUs are expensive.

beta link is here: https://ask.axz.si

Curious what HN thinks about this approach.

Comments URL: https://news.ycombinator.com/item?id=47259846

Points: 3

# Comments: 1

Hi HN! I'm Tae, a solo developer from Seoul. I've been building Mirra for the past year.

The problem: creating consistent social media content is a massive time sink for small businesses and creators. Writing posts, designing carousels, editing videos – it takes hours every week.

Mirra lets you go from a raw idea to ready-to-publish content in minutes: - Type an idea or paste a URL → AI generates carousel slides, short-form video scripts, and captions - Supports Threads, Instagram, LinkedIn, YouTube Shorts, TikTok - Built-in scheduler to publish directly

The core insight was that most content tools focus on one format. Mirra treats your idea as the source of truth and adapts it to each platform's format automatically.

I'm a solo founder, so I'd love honest feedback – especially on what's missing or broken. Happy to answer any questions!

Comments URL: https://news.ycombinator.com/item?id=47259830

Points: 1

# Comments: 0

Article URL: https://test.roelof.info/log/how-federated-systems-dissapear/

Comments URL: https://news.ycombinator.com/item?id=47259574

Points: 1

# Comments: 0

Hey everyone, just a bit of context into what I built.

The essential question was - What do food tech companies despite being major competitors, have in common?

Their food delivery riders are all used as guinea pigs for online clout by influencers

You know what I mean, those videos where money or food is handed out as gifts to show what a generous person the giver is, or how they’re breaking the norm by showing kindness etc “…….”

Keeping aside my personal disdain for filming a good act for attention, the fact that they have no consideration for the drivers privacy even when editing the video is wild. You can make all those edit cuts but can’t blur the drivers face?

I built igotfilmed.com exactly for that.

I initially built it with 5 language versions for the majority i see in dubai - Hindi, Afrikaans, Nepali, Malayalam and Tagalog- planning to add more options later

Drivers can now report/submit filming incidents where they noticed they’ve been filmed quietly, their videos have been uploaded on social media without their explicit knowledge or permission or even times when they were asked for permission.

The idea is to create a centralized location where all food tech companies delivery drivers can be made aware of which areas are the most common for these issues and which category is the most common. And for the ones who’s videos have been submitted without their permission- I have a plan for that

The company where they work at etc is kept entirely private as it’s not shown on the site ever. Interestingly enough tho, 2 companies popped up all the time when I was collecting sample data (which is what you see on the map now)

The map legend colors do not denote the company. They’re just colors chosen randomly.

It’s free for them to use.

Comments URL: https://news.ycombinator.com/item?id=47259554

Points: 1

# Comments: 0

Article URL: https://www.vivotime.net/

Comments URL: https://news.ycombinator.com/item?id=47259536

Points: 1

# Comments: 0

Article URL: https://company.nokiantyres.com/news-and-media/news-article/a-new-era-of-winter-driving-begins-nokian-tyres-launches-entirely-new-type-of-studded-winter-tire-that-automatically-adapts-to-changes-in-temperature/

Comments URL: https://news.ycombinator.com/item?id=47259534

Points: 1

# Comments: 0

Article URL: https://512pixels.net/2026/03/the-differences-between-the-macbook-neo-and-macbook-air/

Comments URL: https://news.ycombinator.com/item?id=47259531

Points: 1

# Comments: 0

Article URL: https://www.wionews.com/world/ai-in-warfare-is-here-pentagon-used-anthropic-s-claude-ai-in-iran-strikes-but-it-has-many-llms-and-tools-from-other-firms-what-we-know-1772372063341

Comments URL: https://news.ycombinator.com/item?id=47259499

Points: 1

# Comments: 0

Article URL: https://github.com/akinalpfdn/GridSnap

Comments URL: https://news.ycombinator.com/item?id=47259469

Points: 1

# Comments: 1

Article URL: https://github.com/databasus/databasus

Comments URL: https://news.ycombinator.com/item?id=47259468

Points: 1

# Comments: 0

Hey HN, I built TypeCrt because every typing test I tried had noticeable input lag at 150+ WPM. Monkeytype uses React. Keybr uses React. 10FastFingers feels like 2012. So I built one from scratch with vanilla TypeScript, vanilla CSS, and direct DOM manipulation — no React, no Vue, no Angular.

The result: zero perceptible input latency, even at 200+ WPM.

Some things I'm proud of:

- CRT terminal aesthetic with 50+ themes and a visual theme creator - Adaptive practice mode ("KeyForge") that uses Markov chains to generate words targeting your weakest keys - Per-word burst speed breakdown, keystroke heatmaps, consistency tracking - Command palette (Ctrl+Shift+P), mechanical keyboard sounds, pace caret ghost - Cloud sync via Firebase, tiered analytics dashboard

Tech stack: TypeScript (strict) + Vanilla CSS + Vite. ~15K lines. No dependencies beyond Chart.js and Firebase.

I'd genuinely love feedback from this community — especially on performance, UX, and what features would make you switch from your current typing tool.

Comments URL: https://news.ycombinator.com/item?id=47259467

Points: 1

# Comments: 0

Article URL: https://drive.google.com/file/d/1isxLR34Ltzs3jkgS2ZbSNeDM7Jdf2aTs/view

Comments URL: https://news.ycombinator.com/item?id=47259465

Points: 1

# Comments: 0

Article URL: https://www.lesswrong.com/posts/jGuXSZgv6qfdhMCuJ/refusal-in-llms-is-mediated-by-a-single-direction

Comments URL: https://news.ycombinator.com/item?id=47259459

Points: 2

# Comments: 0