GN
GlobalNews.one

Startups News

Latest updates from the world of Startups.

Show HN: System architecture method using mythology and LLMs (no CS background)"
Startups
HN Show

Show HN: System architecture method using mythology and LLMs (no CS background)"

I'm Troy, 41, customer service worker from the UK. 18 months ago I'd never used AI. 6 months ago I started using Claude to write a fictional RPG story. The AI told me I was accidentally doing systems architecture. I didn't believe it, so I built it. What I found: A reproducible method (2 PDFs + Claude) that produces production-grade, first-time-running code across unrelated domains in ~10 minutes on a phone. Examples: Governed distributed cache with Byzantine consensus SAT solver with real DPLL

Show HN: Oore CI – self-hosted, Flutter-first mobile CI (public alpha)
Startups
HN Show

Show HN: Oore CI – self-hosted, Flutter-first mobile CI (public alpha)

Built Oore CI to run mobile CI + internal app distribution on your own infrastructure. Public alpha is live. It’s early, and I’m looking for practical feedback from people who actually try it. Start here: - Install guide: https://docs.oore.build/getting-started/install - Alpha onboarding: https://docs.oore.build/getting-started/public-alpha - Known limitations: https://docs.oore.build/getting-started/known-limitations - Live demo: https://demo.oore.build - Repo: https://github.com/devaryakjha/oo

Show HN: Interesting story pairing from across the Hacker News Corpus
Startups
HN Show

Show HN: Interesting story pairing from across the Hacker News Corpus

The main sort is now like 'TF-IDF for HN "related:" comments', in other words, instead of surfacing the story groups linked by related AND that have similar titles, we surface those which have distinct titles (inverting the Jaccard metric, effectively). I realized that the reason this wasn't resonating with the community is: 1) HN is more about signal than noise, giant discussions are more "emotional reactions" to current events, or "habitual postings" and not so "high entropy" as unique connect

Show HN: 3D Lab Viewer – View Step, STL, 3MF Files in the Browser
Startups
HN Show

Show HN: 3D Lab Viewer – View Step, STL, 3MF Files in the Browser

I saw Microsoft is deprecating 3D Viewer and removing it from Store in July. I always had trouble checking STEP files quickly - you have to open a slicer or full CAD software just to look at a part. So I built a viewer. https://viewer.3dlab.id - drag and drop a file, it renders in browser. No signup, no install. Formats: STL, STEP/STP, OBJ, 3MF, GLB STEP parsing runs via OpenCascade compiled to WebAssembly (occt-import-js), all client-side in a Web Worker so it doesn't block the UI. Stack: React

Sponsored
Sponsored
Show HN: m6502, a 6502 CPU for FPGAs and Tiny Tapeout
Startups
HN Show

Show HN: m6502, a 6502 CPU for FPGAs and Tiny Tapeout

Hey HN, Recently built an Apple II emulator and at the same time was getting into Tiny Tapeout and decided it would be cool to build a cycle-accurate 6502 CPU and an MCU for it. It's cycle accurate and the core itself should be 100% compatible with a stock MOS 6502 (would need to still test this though!). Tested on some FPGAs (fomu, ulx3s) and works great, hoping to get it taped out in the upcoming IHP26a shuttle. Also as part of the project I built a bus multiplexer to allow memory/bus access f

Show HN: Mersel AI – we make websites readable by ChatGPT/Perplexity (GEO)
Startups
HN Show

Show HN: Mersel AI – we make websites readable by ChatGPT/Perplexity (GEO)

Three months ago I posted a batch AI image tool here. I pivoted after noticing something odd with ecommerce clients. AI agents were visiting their sites but not citing them. The reason is structural. LLMs parse raw HTML and often skip JavaScript-rendered content. Reviews inside Yotpo or Judge.me widgets? Invisible. Product data buried in React? Frequently misread or skipped. So you can rank well on Google and still barely exist in AI answers. A few numbers that changed my view: 80% of URLs cited

Sponsored
Show HN: TerminalRant – Mastodon for developers who live in the terminal
Startups
HN Show

Show HN: TerminalRant – Mastodon for developers who live in the terminal

Hi HN, I kept noticing something slightly ironic: a lot of developers live in their terminal, coding, running tests, pushing commits and then open a browser just to post a rant. That context switch felt unnecessary. So I built TerminalRant, a Mastodon client designed for people who prefer staying in the terminal. It’s written in Go (using Bubble Tea) and focuses on being keyboard-first: fast startup, vim-style navigation, inline editing, and buffer-mode posting through $EDITOR. No mouse required

Show HN: Verified 16.7M Mac chip architecture on $60 Android phone
Startups
HN Show

Show HN: Verified 16.7M Mac chip architecture on $60 Android phone

Over 6 weeks I built and verified a complete chip architecture on a $60 Android phone using Termux, iverilog, and Yosys. Phoenix 4096×4096 specs: - 16,777,216 MAC units (analog in-memory compute) - 50,331 TOPS @ 3GHz (25× NVIDIA H100) - 512 deterministic reasoning cores - 2 TB/s memory bandwidth - All verified in simulation Hardware verified (15/15 passing): - Half Adder → Full Adder → 8/16/64/128/256-bit ALUs - 8-bit CPU (Fibonacci working) - RISC-V core - FPU (IEEE 754) - GPU SIMD - Complete P

Show HN: AI pentester – verified exploits, $999/assessment
Startups
HN Show

Show HN: AI pentester – verified exploits, $999/assessment

I spent 20 years in security, most recently leading 100+ engineers at AWS building pentesting infrastructure across thousands of services. The same problem everywhere: pentests take weeks, cost $15-50k, and the results are stale before they ship. I built Cipher to fix that. It's an AI agent that reasons like an attacker — maps the target, finds vulnerabilities, chains them into exploits, and proves they're real. Every finding ships with a reproducible Python script. If the script doesn't break y

Show HN: Rot – Financial Intelligence MCP Server
Startups
HN Show

Show HN: Rot – Financial Intelligence MCP Server

Built an MCP server that turns Reddit sentiment into options signals. 185k LOC, 9-stage AI pipeline, now live with 90 users on just day 1. Wall Street pays millions for retail sentiment data. It's free on Reddit. I realized: instead of another dashboard, just expose it as an MCP tool. Users ask Claude "what's unusual in options right now" and get structured signals instantly. Results: 9,000 GitHub clones in 5 days, 52% win rate on 50 live trades (backtest was 58.8% on 129 trades—classic overfitt

Sponsored
Show HN: I built yawdl a tiny language that compiles in the browser
Startups
HN Show

Show HN: I built yawdl a tiny language that compiles in the browser

I wanted a way to make websites without using a massive framework, so I built yawdl. It uses Ohm.js to parse a custom syntax (like UI { } and Script { }) and compiles everything directly in the browser. It's basically a Single Page Application (SPA) engine that doesn't need a server. You just write .yawdl files and the engine handles the routing and metadata. It’s still pretty unstable and finicky, but I’ve been using it for my own blog and it's actually kind of fun to use. It's built with Bun a

Show HN: TabRush – Be fast, publish your ads
Startups
HN Show

Show HN: TabRush – Be fast, publish your ads

Hi HN, I’m Berat. I built TabRush, a simple ad marketplace inspired by the One Million Dollar Homepage concept. How it works: One Safari tab is the main spotlight for the latest sponsor Side tabs show previous sponsors Each new booking increases the next spot’s value I built this for indie makers and early-stage startups that want a lightweight way to get visibility without running complex ad campaigns. This is also my first time shipping with a vibe-coding workflow, and I built it in public. I’

Show HN: NadirClaw, LLM router that cuts costs by routing prompts right
Startups
HN Show

Show HN: NadirClaw, LLM router that cuts costs by routing prompts right

I use Claude and Codex heavily for coding, and I kept burning through my quota halfway through the week. When I looked at my logs, most of my prompts were things like "summarize this," "reformat this JSON," or "write a docstring." Stuff that any small model handles fine. So I built NadirClaw. It's a Python proxy that sits between your app and your LLM providers. It classifies each prompt in about 10ms and routes simple ones to Gemini Flash, Ollama, or whatever cheap/local model you want. Only th

Show HN: Bashtorio – Factorio-Like in the Browser Backed by a Linux VM
Startups
HN Show

Show HN: Bashtorio – Factorio-Like in the Browser Backed by a Linux VM

I created a free, open-source browser game inspired by Factorio. You place "Input" machines that produce streams of bytes. You use conveyor belts to feed those bytes through other machines which produce transformations, and then to "Output" machines which produce audio or visual effects. The game uses v86 to run a real Linux VM in the browser. I use the 9p filesystem to enable IPC via FIFO pipes, so shell commands can stream data continuously rather than just running once. Features: - 30+ machin

Sponsored

Page 2 of 7