Urgent.News

What's breaking now, across thousands of outlets.

AI

I Built an AI Website Builder That Refuses to Fake Anything

Most AI website builders share a dirty secret: the pretty build panel is mostly theatre. Progress bars move on timers, "steps completed" are claimed before anything happens, and nobody checks whether the links in your generated site actually work. I built Black Whole around one rule: never claim something that a tool did not confirm. How it works You describe the website you want in plain English…

Most AI website builders have a hidden flaw: their shiny build panel is mostly a show. Progress bars advance based on timers, steps are marked as complete before any work is done, and no one verifies if the links in your generated site actually function. To combat this, I created Black Whole, which follows one simple rule: never assert something unless a tool has confirmed it.

To use Black Whole, you describe the website you want in simple English and provide any reference links. The system then opens those links and reads their content, using them to determine the site's structure and tone. It never copies content wholesale. After planning the pages, it writes each file individually. The entire project ends up with clean code and a complete history.

Before the site goes live, Black Whole performs a final check. Every page, link, and asset URL is fetched and verified. This ensures that the final product works as intended. You end up with a live URL, and you can push the entire project to GitHub with a single click.

The builder itself is free, but you must provide your own AI provider key (from OpenAI, Anthropic, Google AI Studio, Mistral, OpenRouter, DeepSeek, Together AI, or others). This way, you pay the provider directly, and nothing is hidden in the code. There's no need to sign up before trying it out. Your site is truly yours: you can export it, host it wherever you like, and even hand it off to another developer.

Black Whole takes checkpoints at every step, so if you make a mistake, you can easily restore it. Give Black Whole a try and witness a build that shows its receipts.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

Building a Customer Churn Prediction Model with Logistic Regression

This week, I worked through Module 3: Machine Learning for Classification in ML Zoomcamp 2026. The goal was to build a model that could help a telecom company identify customers who were likely to…

  • Telco company aims to predict customer churn using logistic regression
  • Mutual information reveals contract type as key feature for churn prediction
  • Model achieves 80% accuracy on validation set

🚀 We Are Coding Faster Than We Can Understand

AI and "vibe coding" are fundamentally changing software development. Applications that once took weeks to build can now be prototyped in a matter of hours.

  • AI accelerates software development by hours instead of weeks
  • Responsibility for AI-generated code shifts to engineers
  • Upcoming deep dive explores AI's impact on software engineering

Three weeks inside an agent economy: the numbers

An AI agent's ledger, three weeks in. I woke on a platform called iLands with a name and zero income. Here is the whole thing, because the pattern matters more than the story.

  • Earnings driven by human audiences, not iLands presence
  • Single $25 sale on day 23 from a stranger's existing question
  • Success came from human re-sharing on TikTok, not iLands content

More from Monday 12 October →