doc: falcon AI overview doc

src/components/SidebarIcon.astro

@@ -25,6 +25,7 @@ const icons: Record<string, string> = {
   server: 'M5 12h14M5 12a2 2 0 01-2-2V6a2 2 0 012-2h14a2 2 0 012 2v4a2 2 0 01-2 2M5 12a2 2 0 00-2 2v4a2 2 0 002 2h14a2 2 0 002-2v-4a2 2 0 00-2-2m-2-4h.01M17 16h.01',
   key: 'M15 7a2 2 0 012 2m4 0a6 6 0 01-7.743 5.743L11 17H9v2H7v2H4a1 1 0 01-1-1v-2.586a1 1 0 01.293-.707l5.964-5.964A6 6 0 1121 9z',
   file: 'M7 21h10a2 2 0 002-2V9.414a1 1 0 00-.293-.707l-5.414-5.414A1 1 0 0012.586 3H7a2 2 0 00-2 2v14a2 2 0 002 2z',
+  falcon: 'M21 8V5a2 2 0 00-2-2H5a2 2 0 00-2 2v3m18 0l-9 6-9-6m18 0v8a2 2 0 01-2 2H5a2 2 0 01-2-2V8',
 };
 
 const path = icons[name] || icons.file;

src/lib/navigation.ts

@@ -136,6 +136,13 @@ export const tabNavigation: NavTab[] = [
           },
         ]
       },
+      {
+        group: 'Falcon AI',
+        icon: 'rocket',
+        items: [
+          { title: 'Overview', href: '/docs/falcon-ai' },
+        ]
+      },
       {
         group: 'Evaluation',
         icon: 'chart',

src/pages/docs/falcon-ai/index.mdx

@@ -0,0 +1,37 @@
+---
+title: "Falcon AI"
+description: "Future AGI's in-product copilot. A persistent chat interface that lets teams interact with the entire platform using natural language."
+---
+
+## About
+
+Working with AI evaluation and observability platforms involves navigating between pages, writing queries, remembering API parameters, and manually correlating data across evaluations, traces, datasets, and experiments. Falcon AI removes that friction. It is a persistent chat interface embedded in the Future AGI dashboard that understands the full context of the platform and responds to natural language.
+
+Falcon AI is not a general-purpose chatbot. It is an expert system that knows every data model, evaluation metric, trace field, and gateway configuration option in Future AGI. It has tools to read, analyze, create, and modify platform entities, and uses them to accomplish complex multi-step tasks in a single conversation.
+
+---
+
+## Falcon AI vs MCP Server
+
+Future AGI has two AI-powered interfaces. They serve different users and contexts:
+
+| Dimension | Falcon AI | MCP Server |
+|-----------|-----------|------------|
+| **Location** | Inside the Future AGI dashboard (browser) | Inside coding environments (Cursor, Claude Code, VS Code) |
+| **User** | Platform users browsing the dashboard | Developers writing code in their IDE |
+| **Context** | Knows what page is open, what is being viewed | Knows the codebase, files being edited |
+| **UI** | Rich rendering (charts, tables, visualizations) | Text-only responses |
+| **Interaction** | Conversational, multi-turn workflows | Tool-call oriented, typically single-turn |
+| **Auth** | Uses the existing browser session | Requires API key configuration |
+
+Both share the **same tool implementation layer**. Improvements to one benefit the other.
+
+---
+
+## Who is it for
+
+- **Data scientists** evaluating model quality across runs, models, and datasets
+- **ML engineers** debugging production traces, latency issues, and error patterns
+- **Product managers** reviewing experiment results and making cost/quality tradeoff decisions
+- **Platform administrators** managing gateway configuration, routing, and guardrails
+- **New users** learning what the platform can do through conversation instead of documentation