Update README.md

README.md

@@ -9,15 +9,27 @@ pinned: false
 
 # Metaphi     
                                                                                                                              
-  We build RL environments for commercial sale to foundation model companies, and select enterprises.   
+  We build RL environments for foundation model companies.   
+
+  We introduce, LH-Agents, a suite of enterprise grade evaluation environments, modelled on real-world professional workflows.  
+
+## LH-Agents                                                                                                               
                                                                                                                              
+  | Agent | Domain | Scale | What It Tests |                                                                                 
+  |-------|--------|-------|---------------|                                                                                 
+  | **[FigmaAgent](link)** | Design → Production Code | 37 tasks, 147 expert preferences | Figma file → deployed React application |
+  | **[FinAgent](link)** | Credit Underwriting | 4,406 cases, 45K+ PDFs | PDF extraction → categorization → P&L → credit worthiness assessment |
+  | **[Enterprise Knowledge Agent](link)** | Content Synthesis | 1,220 pitch decks, 45 video tasks, 279 preferences | Source materials → structured knowledge artifacts |
+                                                                                                                             
+## Leaderboard
 
-  ## About
+  Results at [evals.metaphi.ai/leaderboard](https://evals.metaphi.ai/leaderboard)
+  
+## About
 
-  Metaphi is an applied AI research lab founded on the mission to bring real world intelligence to frontier models. We focus on out of distribution tasks, 
+  Metaphi is an applied AI research lab founded on the mission to bring real world intelligence to frontier models. We focus on out-of-distribution tasks, 
   and create high fidelity RL environments that serve as training gyms for models.
 
-
-  We partner with the world's leading domain experts in curating our environments, and creating step wise verifiers. 
+  We partner with the world's leading domain experts in curating our environments, and creating step-wise verifiers. 
 
   Website: [metaphi.ai](https://metaphi.ai)