The latest articles on Forem by Paolo (@paolomulas). https://forem.com/paolomulas https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F674127%2F1e2eabe5-9d9b-41c0-a906-5dd8dc8f8af7.png https://forem.com/paolomulas en Paolo Sun, 02 Nov 2025 08:36:35 +0000 https://forem.com/paolomulas/built-datapizza-ai-in-php-on-2011-raspberry-pi-edge-ai-without-gpu-22ba https://forem.com/paolomulas/built-datapizza-ai-in-php-on-2011-raspberry-pi-edge-ai-without-gpu-22ba <h1> Built Datapizza-AI in PHP on 2011 Raspberry Pi (256MB RAM) — Zero GPU </h1> <p>I ported <strong>Datapizza-AI from Python to PHP</strong> to explore edge AI on ultra-constrained hardware. This runs on a 2011 Raspberry Pi Model B with only 256MB of RAM.</p> <h2> The Challenge </h2> <ul> <li>Ultra-constrained hardware (256MB RAM, 700MHz CPU)</li> <li>Zero heavy dependencies — just curl + OpenAI API </li> <li>No local ML models, no GPU</li> </ul> <h2> The Result </h2> <p><strong>Fully functional AI agents</strong> using ReAct pattern with:</p> <ul> <li>Tool-calling and reasoning</li> <li>Multi-model support (OpenAI, DeepSeek, Mistral, Anthropic, Kimi)</li> <li>Vector search with JSON-based storage</li> <li>Full streaming support</li> </ul> <h2> Why This Matters </h2> <p>Proves that intelligent <strong>API architecture beats raw compute</strong>. You don't need expensive hardware when you design systems API-first.</p> <h2> The Project </h2> <p>📦 <strong>GitHub:</strong> <a href="https://github.com/paolomulas/datapizza-ai-php" rel="noopener noreferrer">https://github.com/paolomulas/datapizza-ai-php</a></p> <p>📝 <strong>Full Technical Breakdown:</strong> <a href="https://medium.com/@paolomulas_29555/the-smallest-ai-lab-in-the-world-built-with-php-and-a-raspberry-pi-6decc45639e6" rel="noopener noreferrer">https://medium.com/@paolomulas_29555/the-smallest-ai-lab-in-the-world-built-with-php-and-a-raspberry-pi-6decc45639e6</a></p> <h2> Key Features </h2> <ul> <li>ReAct agents with tool-calling</li> <li>Vector search with JSON persistence</li> <li>Streaming responses</li> <li>Modern PHP 7.x+ code</li> <li>Framework-agnostic</li> <li>~150 lines to get started</li> </ul> <p>Curious if anyone else is doing edge AI experiments on legacy hardware! Happy to discuss implementation details.</p> php ai raspberrypi openai