Update app_gradio.py

app_gradio.py

@@ -535,4 +535,30 @@ Upload a CSV with columns: **`source`**, **`log_message`** Output includes: `pre
             )
 
             gr.Markdown("""
-**Sample CSV format:**
+**Sample CSV format:**
+""")
+
+        # ── Tab 3: Architecture ───────────────────────────────────────────
+        with gr.Tab("🏗️ ARCHITECTURE"):
+            gr.Markdown("""
+## 3-Tier Hybrid Pipeline
+
+| Tier | Method | Coverage | Latency | Trigger |
+|------|--------|----------|---------|---------|
+| 🟢 **Regex** | Python `re` patterns | ~21% | < 1ms | Fixed patterns |
+| 🔵 **BERT** | `all-MiniLM-L6-v2` + LogReg | ~79% | 20–80ms | High-volume categories |
+| 🟡 **LLM** | HuggingFace Inference API | ~0.3% | 500–2000ms | LegacyCRM + rare patterns |
+
+## Model Performance
+- **Training data**: 2,410 synthetic enterprise logs
+- **Confidence threshold**: 0.5 (below → escalate to LLM)
+- **Source-aware routing**: `LegacyCRM` → LLM directly
+
+## Environment Variables
+| Secret | Purpose |
+|--------|---------|
+| `HF_TOKEN` | LLM inference for LegacyCRM logs |
+""")
+
+if __name__ == "__main__":
+    demo.launch(server_name="0.0.0.0", server_port=7860, theme=THEME, css=CUSTOM_CSS)