# Feasibility Study & Architecture Plan: AI-Powered APK Builder This document summarizes the technical feasibility, infrastructure requirements, financial planning, and architectural options for launching the **Internal Cursor AI-to-APK Builder** as a public SaaS product. --- ## 1. Product Vision & Market Niche ### The Problem Setting up a local mobile development environment (Android Studio, Java JDK, Android SDK, and Flutter) is highly complex, takes hours, and is intimidating for beginners, designers, and non-technical entrepreneurs. ### The Solution A web or mobile application where users can: 1. **Chat with an AI** to describe their application idea. 2. **Edit the generated code** directly in a lightweight web/mobile editor. 3. **Compile to APK with one click** and instantly download the compiled binary to their phone. --- ## 2. Technical Architecture Options We analyzed three potential architectural paths for this product: ### Option A: Hybrid Cloud Model (Web/Mobile UI + Cloud Compiler) — 🌟 Recommended The frontend (a Web IDE or a Flutter mobile app) serves as the interface for AI chatting and code editing. The heavy compilation is outsourced to a cloud server. * **How it works:** When a user clicks "Build APK", the code is packaged and sent to a secure cloud build server. The server compiles the APK and returns a download link. * **Pros:** Users can build real Android apps from any device (including smartphones/tablets) without installing any SDKs. * **Cons:** Requires maintaining and securing cloud build servers. ### Option B: Local Development Setup (Zero Server Cost) The IDE runs locally on the user's PC (via a local PHP server like `php -S localhost:8000`), using their local CPU, RAM, and SDKs. * **Pros:** $0 hosting cost for the developer. No security sandboxing required. * **Cons:** Users must manually install Flutter, Java, and the Android SDK on their computers. ### Option C: Standalone Mobile Compilation (Not Feasible) Compiling the Flutter app directly on the user's smartphone without a server. * **Pros:** None, as it is technically impractical. * **Cons:** Android devices lack the official tooling to run the full Gradle and Android compiler chains efficiently. iOS completely forbids compiling and executing arbitrary binaries in their sandbox. --- ## 3. Hosting & Infrastructure Analysis ### Why Managed/Shared Hosting (e.g., Hostinger Cloud) is Insufficient Even though premium Cloud Hosting plans offer up to 3 GB of RAM and 2 CPU Cores, they cannot be used for this project because: 1. **No Root Access:** Installing the Java JDK and Android SDK requires system-level administrator (`sudo`) access. 2. **Disabled Functions:** Managed hosts disable PHP execution functions (`shell_exec`, `exec`, `proc_open`) for security. 3. **Process Limits:** Spawning Gradle and Java compilers quickly exceeds the "Max Processes" limit (usually 120), causing the server to kill the build. ### Recommended VPS Providers (For the Cloud Compiler) To run the compiler in the cloud, you need a **Virtual Private Server (VPS)** running Ubuntu Linux. | Provider | Plan | Monthly Cost | Specs | Verdict | | :--- | :--- | :--- | :--- | :--- | | **Hetzner** | CX22 | **~$3.60** (€3.29) | 2 vCPU, 4 GB RAM, 40 GB SSD | **Cheapest high-performance option.** | | **Hostinger** | KVM 1 | **~$4.99 – $5.99** | 1 vCPU, 4 GB RAM, 50 GB NVMe | Great to keep under your existing Hostinger account. | | **Contabo** | Cloud VPS S | **~$6.00** | 4 vCPU, 8 GB RAM, 50 GB NVMe | **Best value.** Highly recommended for fast builds. | | **AWS** | Lightsail | **$10.00** | 1 vCPU, 2 GB RAM, 60 GB SSD | Good if you want to stay in the AWS ecosystem. | | **AWS** | Free Tier (`t2.micro`) | **$0.00** (12 months) | 1 vCPU, 1 GB RAM, 30 GB SSD | **Requires a 3GB Swap File** or it will crash. | --- ## 4. Disk Space & Resource Management A major concern with Android/Gradle compilation is disk bloat. A standard build server can be kept clean using these strategies: 1. **Initial Footprint (~8–12 GB):** * Ubuntu OS: ~3 GB * Flutter SDK: ~1.5 GB * Android SDK (API 34 only): ~2.5 GB * Java JDK 17: ~0.3 GB * This leaves **18 to 38 GB free** on a standard 30–50 GB VPS disk. 2. **Automatic Cleanups:** * **`flutter clean`:** Run this automatically before or after every build. It deletes the temporary `build/` directory, saving up to 1 GB per project. * **Gradle Cache Pruning:** Schedule a monthly cron job to clear the Gradle cache folder (`rm -rf ~/.gradle/caches/`). Gradle will only re-download active dependencies on the next build. --- ## 5. Financial Model & Pricing Strategy ### Phase 1: Development & Private Beta (1–50 Users) * **Goal:** Build the prototype, refine the AI prompts, and test compilation. * **Monthly Budget:** **~$5.00 – $7.00** (Cheap VPS + Domain + Free Gemini API tier). ### Phase 2: Public Launch (50–500+ Users) * **Goal:** Secure the server using Docker sandboxing, set up a build queue, and charge users. * **Monthly Budget:** **~$30.00 – $65.00** (Web server, build server, paid Gemini API, email service). ### Revenue & Profit Projections (At $9.99/month subscription) * **5 subscribers:** **$50/mo** — Breaks even. * **20 subscribers:** **$200/mo** — Covers all costs with profit. * **100 subscribers:** **$1,000/mo** — **~$930+ net profit.** --- ## 6. Implementation Roadmap ### Step 1: Run and test locally on Windows Run the PHP server locally on your Windows PC using your local Flutter and Android SDKs: ```powershell cd c:\Users\Bai Jue\Desktop\cursor\backend php -S localhost:8000 ``` Open `http://localhost:8000` to test the AI editor and APK generation. ### Step 2: Get a VPS and run the Automated Setup Script When ready to move to the cloud, buy a VPS (Ubuntu 22.04/24.04 LTS), upload the automated setup script (`setup_flutter_server.sh`), and run: ```bash sudo bash setup_flutter_server.sh ``` This will automatically install Java, Android SDK, Flutter, accept all licenses, and configure permissions. ### Step 3: Secure and Scale (For Public Release) 1. **Containerization:** Wrap the compilation process inside temporary **Docker** containers. 2. **Queues:** Implement a queue system so the server compiles one APK at a time to prevent CPU overload.