Upload README.md with huggingface_hub

README.md

@@ -1,13 +1,45 @@
 ---
 title: Mobile Agent
-emoji: 🦀
-colorFrom: green
-colorTo: yellow
+emoji: 🤖
+colorFrom: blue
+colorTo: indigo
 sdk: gradio
-sdk_version: 6.19.0
-python_version: '3.13'
+sdk_version: "4.44.0"
 app_file: app.py
-pinned: false
+pinned: true
+license: apache-2.0
+tags:
+  - agent
+  - mobile
+  - on-device
+  - smolagents
+  - mcp
+  - dispatchai
+models:
+  - dispatchAI/Llama-3.2-1B-Instruct-Q4-mobile
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# 🤖 dispatchAI Mobile Agent
+
+**A real AI agent running on a 1B parameter model — small enough for your pocket.**
+
+This Space proves that useful AI agents can run on phone-sized models. The agent is powered by [Llama-3.2-1B-Instruct-Q4-mobile](https://huggingface.co/dispatchAI/Llama-3.2-1B-Instruct-Q4-mobile) — a 1B parameter model quantized to 4-bit, with a 700MB file size that runs at 18 tokens/sec on a Snapdragon 865.
+
+## Features
+
+- 💬 **Chat** — Conversational AI on a 1B model
+- 👨‍💻 **Code generation** — Python code from natural language
+- 🔧 **MCP server** — Add this agent as a tool in Claude Desktop / Cursor
+
+## The Point
+
+This isn't about matching GPT-4. It's about proving that a 1B model on a phone can be genuinely useful. For the tasks people actually do on phones — quick answers, code snippets, summaries, classifications — a 1B model is enough.
+
+## Run On Your Phone
+
+```bash
+hf download dispatchAI/Llama-3.2-1B-Instruct-Q4-mobile model.gguf
+llama-cli -m model.gguf -p "Hello!" -n 100 -t 4
+```
+
+🚀 [dispatchAI](https://huggingface.co/dispatchAI) — Small. Mobile. Free. UAE-built.