Update README.md


README.md

@@ -1,10 +1,34 @@
 ---
 title: README
-emoji: 🐨
+emoji: 🔥
 colorFrom: indigo
 colorTo: red
 sdk: static
 pinned: false
 ---
 
-Edit this `README.md` markdown file to author your organization card.
+# Covenant AI
+
+> Building the world's first end-to-end decentralized AI stack through three integrated Bittensor subnets: **Templar** (training), **Basilica** (compute), and **Grail** (post-training).
+
+We're making foundation model development accessible beyond centralized infrastructure.
+
+**Website:** https://covenant.tech
+**Twitter:** [@covenant_ai](https://x.com/covenant_ai)
+**GitHub:** [covenant-labs](https://github.com/covenant-labs)
+
+---
+
+## Latest Research
+
+**Heterogeneous Low-Bandwidth Pre-Training of LLMs** (2026)
+Enabling foundation model training across heterogeneous infrastructure with selective compression strategies.
+[arXiv:2601.02360](https://arxiv.org/abs/2601.02360)
+
+---
+
+## The Orders
+
+**Templar (SN3)** - Decentralized foundation model training
+**Basilica (SN39)** - Decentralized compute platform
+**Grail (SN81)** - Post-training and reinforcement learning