2025年7月14日 的 Show HN

I’m from Europe, and when I spent a summer at Stanford, I saw how different the startup ecosystem is in the US. Everything there feels connected. In Europe, it’s scattered. Hard to discover early-stage startups unless you’re in the right city or network.

So I built StartupList EU, a public directory where anyone can list or browse European startups.

The goals is to contribute to the EU startup ecosystem more accessible and transparent for founders, investors and operators.

What it does: - Founders can submit their startup for free - Each profile includes data like team size, category, funding, revenues, location, founders and more - You can search by country, industry, name, team size, country and business model - It works across the whole EU, not just big hubs like Berlin or Paris

Right now there are 34 startups listed. More are coming in daily. I’m working on better filters, API access, and a weekly newsletter.

Would love your feedback: - What data would be most useful to you? - What would make this genuinely helpful for founders, investors, or researchers? - If you are from US, what's your take about EU startup ecosystem?

Hi HN,

I built this after getting frustrated trying to map out things like “Ancient mayan ruins” or “James Bond movie locations” on Google "My" Maps. You basically have to drop pins manually, hunt down photos, copy links, it’s slow and never feels like an actual story.

With MapScroll, you just type something like “Marco Polo’s route” or “forgotten WWII sites in France”, and it geocodes, grabs images + articles, and plots everything into a shareable story map. Each marker comes with a little gallery and sources tied to your prompt.

You can drag stops around, add notes, or export it. I’m still tweaking things. Happy to hear edge cases or annoyances.

Give it a try: https://mapscroll.ai/

Hi HN,

I've built `goliteql`, a fast and lightweight GraphQL executor and code generator written in pure Go.

It parses GraphQL schema and queries, validates them, and executes resolvers without using reflection. It also supports introspection and generates Go code from a GraphQL schema.

Why I built it: - I wanted a fast, minimal alternative to gqlgen - Easy to embed in WASM or microservices - Focused on simplicity and performance (4x faster in benchmarks)

Hey HN,

We just recently launched Forge – an AI model API platform that lets you access and call AI models from multiple providers in one place. Find it on our website: https://tensorblock.co/forge

Some key features: Unified API Key – Store multiple provider API keys and access all with a single Forge API key.

OpenAI API Compatible – Drop-in replacement for any application that uses OpenAI’s API.

Advanced Security – Strong Encryption for API keys with JWT-based authentication.

Client Management – Easy key and user management via included command-line interface.

Open-source – https://github.com/TensorBlock/forge

Try out our product – it’s free! We’re open to all feedback and suggestions. Let us know in the comments.

Hey HN!

After seeing people struggle with the Lovable + n8n + Supabase stack, I built VibeFlow.

*The Problem:* Lovable/v0 and all the other prompt to apps builds great frontends, but no functional backends. Some people opt for using lovable + n8n public webhooks + supabase manual integration, but you become human middleware copying URLs, managing the communication between tools, and praying nothing breaks.

*The Solution:* Visual canvas for backend logic that generates actual deployable code.

*What makes it different:* It syncs with your GitHub repo (same codebase as Lovable) and generates real backend code

Try it: https://beta.vibeflow.ai/ Demo: https://youtu.be/3AlscfiAJmY

Built this after 100+ user interviews with Lovable builders. Would love your feedback!

We wanted to know how well LLMs can predict what you think and put them to the test in a game of codenmaes! Grab a friend and play in two teams, each consisting of one human and one LLM. Do you think LLMs can grok your clues?

I got frustrated with the experience of leconjugueur.figaro.fr, so I built a fast and light French verb conjugator over a day.

The development was surprisingly quick using Claude and Svelte. Started with a proof of concept using a JSON file containing all French verb conjugations, then iterated to optimize performance and UX. The whole process took maybe 2-3 hours of actual work.

Key features:

- Fast search and conjugation display

- Clean, minimal interface

- Works offline once loaded

- No ads or tracking

The conjugation data comes from open sources, and the app is client-side only for speed.

Try it: https://barim.us/conjugueur

Code: https://github.com/CodeCadim/conjugueur

Would love feedback, especially from other French speakers or language learners!

Hi HN, I organized this month's Who is Hiring using an AI agent. I have been building agent-rover.com to do repetitive tasks like data entry, finding candidates, or doing the social media outreach. It took around 11 hours to organize this into a spreadsheet with 1 minor hiccup. Parsing and extracting information was relatively easy. I had most trouble getting LLMs to enter data at correct location. LLMs are capable of doing these tasks if you carefully extract and feed the data. Here is the link to 11hr video of agent-rover doing the data entry- https://youtu.be/oIzGsAH2qrU.

I'm also building an agent that will go through "who wants to be hired", gather their resume, LinkedIn profile, and github profile if providided. Then it will check if the prompt and JD and the candidate's profile matches up. Then it will shortlist the candidate for you in a spreadsheet. If this sounds useful to you please fill form here, https://forms.gle/J2UFQYNSYfSNW9uJ9, so that I can reach out to you when that is ready.

Hey HN! I’ve been working on Dyan, a self-hosted visual builder for REST APIs. Just posted a short demo video that shows how it works:

Build endpoints visually Write logic in JavaScript Test live in-browser Serve APIs from your own backend – no codegen, no magic

Demo video: https://youtu.be/SBEPacMgpvk GitHub: https://github.com/dyan-dev/dyan Discord: https://discord.gg/ZQ4pKRA7

Would love feedback and contributions! The goal is to make backend API creation feel like frontend development.

I run small e-commerce business which means i deal with a lot of purchase invoices. I created ParsePoint.app to help me manage it. Simple drag and drop or bulk upload of PDF/PNG/JPG and AI powered data extraction. Same format parsing etc. easy to use and helps me save time

Sometimes you don't have 40 minutes to focus as someone slowly edges around a topic until coming to a point. Whether you're researching, screening educational content, or deciding if a conference talk is worth your time, tldw gives you an almost instant summary of what a YouTube video covers.

To use this just do:

  from tldw import tldw
  import os

  summary = tldw(os.environ.get('OPENAI_API_KEY'))
  summary.summarize('https://www.youtube.com/watch?v=LCEmiRjPEtQ')

I was originally inspired by a post I saw here: https://news.ycombinator.com/item?id=44376989, so I decided to make a more automated version of this.

I originally tried building this as a web application for better user experience, but retrieving YouTube subtitles from cloud provider IP ranges isn't trivial. It would require residential proxies, which was out of scope for a hobby project cost-wise.

Let me know what your thoughts are, feedback is a gift!