Add organization card

README.md

@@ -1,13 +1,40 @@
 ---
-title: README
-emoji: 🌍
-colorFrom: green
+title: protoLabsAI
+emoji: ⚡
+colorFrom: indigo
 colorTo: gray
-sdk: gradio
-sdk_version: 6.19.0
-python_version: '3.13'
-app_file: app.py
+sdk: static
 pinned: false
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+<p align="center">
+  <img src="https://raw.githubusercontent.com/protoLabsAI/.github/main/profile/assets/readme-banner.png" alt="protoLabs.studio" width="100%">
+</p>
+
+# protoLabsAI — open weights
+
+The model shelf of [**protoLabs.studio**](https://protolabs.studio): an indie studio running experiments in the open. One person and a fleet of agents building games, simulators, and specialized agents on infrastructure we run ourselves — from the hardware up.
+
+We run 35B-class open models on a pair of prosumer Blackwell GPUs, and the builds that make that possible go up here for anyone to pull: **FP8 quants** and **speculative-decode heads**, parity-verified against the source before we ship. The recipe and the verification are the product, not just the weights.
+
+The findings behind them get written up at [protolabs.studio](https://protolabs.studio) — CUDA graphs running 37–470% faster on Blackwell, INT4 holding quality on dense models but corrupting MoE routing, `NCCL_P2P_DISABLE=1` clearing TP=2 corruption on PCIe for a 9.3× jump on a 35B MoE.
+
+## The rest of the studio
+
+The weights below are one substrate. The patterns that build everything else are open too:
+
+- [**`rabbit-hole.io`**](https://github.com/protoLabsAI/rabbit-hole.io) — research agent that turns a topic + budget into a knowledge graph, not a pile of links
+- [**`protoAgent`**](https://github.com/protoLabsAI/protoAgent) — A2A-native agent on LangGraph that grows at runtime via plugins, skills, and delegate agents
+- [**`release-tools`**](https://github.com/protoLabsAI/release-tools) — release-notes + Discord embed as one GitHub Action, on every repo
+- [**`protoBanana`**](https://github.com/protoLabsAI/protoBanana) · [**`protoCLI`**](https://github.com/protoLabsAI/protoCLI) · [**`protoWorkstacean`**](https://github.com/protoLabsAI/protoWorkstacean)
+
+Built on top: [**ORBIS**](https://orbis.protolabs.studio) (voice-first AI companion), [**MythXEngine**](https://mythxengine.com) (TTRPG world generator on a Rust multi-agent sim), and PilotProtocol (mech agents on a hex grid).
+
+## Audience
+
+Practitioners already operating in the LLM-agent space. The code has no filter — fork it, hack it to fit. The breakdowns assume you've worked through the docs. If we have to explain tokens, this isn't the shelf.
+
+---
+
+**MIT licensed.** Findings ship as patterns to study and steal. Sibling rig (GGUF / llama.cpp / Ollama): **avaLab**.
+[protolabs.studio](https://protolabs.studio) · [GitHub](https://github.com/protoLabsAI) · [X](https://x.com/protoLabsAI) · [Substack](https://substack.com/@protolabsai)