Update README.md

README.md

@@ -1,11 +1,38 @@
 ---
 title: Smol Private Spreadsheet
-emoji: 🏃
-colorFrom: indigo
-colorTo: red
+emoji: 💻
+colorFrom: blue
+colorTo: yellow
 sdk: docker
 pinned: false
 short_description: A spreadsheet powered by smolLM2 using WebGPU
+tags:
+  - custom-components-for-agents-track   # 🧩 Track 2
+  - track2
+  - gradio
+  - smolllm
+  - spreadsheet
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Smol-Private-Spreadsheet ✨
+
+A tiny proof-of-concept that embeds **SmolLM 2** directly inside a sheet.  
+Type a formula, get an answer—no extra UI required.
+
+## Why it matters
+
+| Feature | What it does |
+|---------|--------------|
+| **`=SMOLLM("prompt")`** | Asks the model and returns the reply inside the cell. |
+| **Context support** | `=SMOLLM("prompt", A1)` or a range `A1:C3` to pass extra context. |
+| **GPU / WebGPU ready** | Runs locally inside the Space; zero external API calls. |
+| **First Step Towards Creating a WebGPU custom Gradio Component** |I'm hoping that this space can showcase the power of UI abstraction when combined with SmolLM2 and allow me to reach mentors that are willing to teach me how to create a custom gradio component that allows the use of small models using WebGPU.  |
+| **Next Steps** | I'm still trying to create one Gradio app that allows me to do both, run a model locally using WebGPU and use MCP tools and other models using inference to bring the best of both worlds. It's a very challenging journey for me, especially given the limitation that all my ideas are competing for the same port at the same time: 7860. |
+
+## Quick start
+
+1. **Launch** the Space.  
+2. In any empty cell, enter:
+
+   ```text
+   =SMOLLM("Summarise A1:C3 in one sentence", A1:C3)