Kodezi

Kodezi is an AI-powered Web IDE that analyzes your entire codebase — not just the current file — to detect bugs, suggest fixes, generate new code from natural language, translate between languages and frameworks, and produce documentation. The core problem it solves is context switching: developers constantly jump between their IDE, documentation, Stack Overflow, and ChatGPT. Kodezi collapses that into one environment that already understands your project.

V1: I built everything from scratch — the full website UI, Monaco-based Web IDE, backend APIs, and Stripe subscription integration. V2: I led KodeziChat, which meant building the entire chat UI and integrating the backend via Socket.io for token-by-token real-time streaming. I also built the VS Code extension UI using the Webview API. V3: Contributed to Kodezi Cloud as part of a larger team. V4: Rebuilt the user dashboard and implemented multi-subscription handling — multiple active Stripe subscriptions tied to a single customer ID, which is a non-trivial billing edge case.

Free users get 25 credits per day for standard features. Premium AI capabilities are gated behind an active paid subscription — and this is enforced at the API level, not just the frontend. The backend checks subscription status against the Stripe customer record on every request to a gated endpoint. The UI doesn't control access; the API does. This prevents the obvious bypass of removing the frontend guard.

KodeziChat is a codebase-aware AI assistant. I built the full frontend and integrated Socket.io for the streaming layer — the model response comes back token by token, not as a complete response. This matters for perceived latency: users see the answer forming rather than waiting for a spinner. The context includes the current file, open tabs, and relevant project files. It's not a generic chatbot with a code block — it knows what's in your project.

It's a separate system I built and deployed independently. It pings every Kodezi service API on a schedule, records the response status in a database, and renders a 90-day health history per service as color-coded daily bars: green for fully up, orange for down under 12 hours, red for down over 12 hours. Deployed on AWS EC2 with Nginx, SSL, and a GitHub Actions CI/CD pipeline. It replaced manual status updates entirely — no human needs to check anything or post a status page update.

A production-grade AI developer tool with a Web IDE, real-time streaming chat, subscription billing, a credits system, and a VS Code extension is a large scope. Budget $40,000 to $120,000 depending on AI integration depth, editor feature complexity, and the scale targets you're designing for. The Stripe and authentication work is a known quantity. The editor integration and streaming infrastructure are where scope varies most significantly.