Describe any AI app, Get a live, MCP-powered app in minutes - NewAgentsHub

I'd been watching the AI tooling space and kept running into the same frustration: every "AI-powered" no-code tool was essentially a chat widget bolted onto a database. Beautiful UI. Real capability? Zero. The moment you needed the app to actually connect to a tool, call an API, or do something across multiple steps — you needed engineers. The "no-code" promise existed only until the AI part started. At the same time, MCP (Model Context Protocol) was quietly becoming the most important protocol in AI — a universal standard for giving AI apps real access to the world: web browsers, databases, code execution, external services. But using MCP required running servers, writing transport configs, understanding Docker networking. It was DevOps-heavy by default. Completely inaccessible to anyone without a technical background. So I started building. Day one: September 29, 2025. Using Claude Code as my development partner — essentially an AI helping a non-engineer build an AI platform. The first version was a multi-provider chat interface. Then I added import, create and generate Skills, Plugins, AI Agents, and Agent Teams with different workflow types, then 45+ MCP server presets that anyone could connect with just an API key. Also Smithery and Docker Gateways. Each layer made the platform more powerful. But I kept coming back to the same question: what does all of this unlock for someone who can't write code? The answer was App Projects. The idea: you describe any AI app in plain English. The platform generates a fully interactive functional web app, runs it live, and wires it to MCP tools, Skills, and AI Agents and Teams automatically. The user refines it through conversation. The Error System catches anything that breaks and fixes it. When it's ready, it embeds anywhere with one iframe. The hardest part wasn't the technical implementation — it was getting the generation loop right. Early versions produced apps that looked good but couldn't actually do anything useful. The breakthrough was treating the generated app not as static output but as a living system: connected to the same MCP infrastructure, Skills, and Agent and Teams that power the rest of the platform. The app isn't just code that was generated — it's a real participant in the platform's AI ecosystem. I built this because I genuinely believe the next wave of software won't be written by developers — it'll be generated by AI and deployed by domain experts who know what they need better than any engineer could. App Projects is the first step toward that. I'd love to hear what you'd build with it. → Try it at newagentshub.dev