Shorten organization card to core positioning and developer links

README.md

@@ -8,78 +8,24 @@ pinned: false
 ---
 
 # LibraOS™
-
 **The AI Core for Every App.**
 
-## Turn your application into an AI application.
+### Turn your application into an AI application.
 
 Add agent execution, knowledge, memory, policy, and audit to the software you already build. Run LibraOS on your infrastructure.
 
-[Start building free](https://libraos.com/developers/) · [Explore the runtime](https://libraos.com/engine/) · [Documentation](https://libraos.com/docs/)
-
-### Your app + LibraOS = AI app
-
-One governed runtime between your product and AI systems.
-
-| Your app | LibraOS | AI app |
-| --- | --- | --- |
-| Your user experience and business logic | Runtime and governance | Useful, controlled, inspectable AI capabilities |
-
-- **Execution:** run agents through supported APIs and SDKs.
-- **Knowledge:** connect approved sources and configure retrieval access.
-- **Memory:** maintain context for your application's workflows.
-- **Policy:** configure permissions and human approval requirements.
-- **Audit:** inspect actions and outcomes.
+**Your app + LibraOS = AI app**
 
-Your application owns its domain logic and business records. LibraOS provides the AI Runtime. **Libra Desk is one application built on LibraOS**, not the definition of the platform.
-
-[See how the parts connect](https://libraos.com/docs/architecture/) · [Explore Libra Desk](https://libraos.com/functions/)
-
-### Start building
-
-Install the Python SDK to connect to a running LibraOS server:
+[Start building free](https://libraos.com/developers/) · [Explore the runtime](https://libraos.com/engine/) · [Docs](https://libraos.com/docs/) · [GitHub](https://github.com/libraos/sdk)
 
 ```bash
 pip install libraos-sdk
 ```
 
-The SDK is a client; it does not install the Runtime. [Download a supported Runtime release](https://libraos.com/download/), then follow the [first-agent quickstart](https://libraos.com/docs/creating-an-agent/).
-
-[SDK and examples on GitHub](https://github.com/libraos/sdk) · [Build on LibraOS](https://libraos.com/docs/build-on-libraos/)
-
-### Deployment and capability boundaries
-
-- **Self-hosted:** run LibraOS on supported infrastructure you control. External models and tools receive the data sent to them; self-hosting does not automatically keep every request local.
-- **Air-gapped:** requires local models, dependencies and connectors, with external connections disabled.
-- **Managed Runtime:** LibraOS Cloud is a **private beta**, not instant self-service provisioning.
-- **Workflow-dependent capabilities:** check the maintained [recovery limits](https://libraos.com/docs/durable-runs/) and [knowledge-backend capabilities](https://libraos.com/docs/knowledge-retrieval/) before designing your application.
-- **Licensing:** personal use and eligible evaluation follow the Runtime release's license terms. Production commercial rights are agreed separately. SDK and model licenses are distinct.
-
-[Security and deployment responsibilities](https://libraos.com/security/) · [Runtime downloads and licensing](https://libraos.com/download/#license)
-
-### Models on Hugging Face
-
-This organization hosts model repositories, including quantized releases and mirrors. Read each model card for upstream attribution, licensing, available files and usage limitations. Listing a model here does not imply that LibraOS created its base weights or quantization.
-
-| Repository | Notes |
-| --- | --- |
-| [Qwen3.8-27B-GGUF](https://huggingface.co/libraos/Qwen3.8-27B-GGUF) | Mirror of [Unsloth's quantization](https://huggingface.co/unsloth/Qwen3.8-27B-GGUF); see the model card for attribution |
-| [Qwen3.6-35B-A3B-AWQ-4bit](https://huggingface.co/libraos/Qwen3.6-35B-A3B-AWQ-4bit) | Model card and release files |
-| [DeepSeek-V3.2-Exp-AWQ](https://huggingface.co/libraos/DeepSeek-V3.2-Exp-AWQ) | Model card and release files |
-| [Gemma-4-31B-IT-NVFP4](https://huggingface.co/libraos/Gemma-4-31B-IT-NVFP4) | Model card and release files |
-| [gemma-4-31B-it-int4-AutoRound](https://huggingface.co/libraos/gemma-4-31B-it-int4-AutoRound) | Model card and release files |
-| [gemma-4-12B-it-qat-w4a16-ct](https://huggingface.co/libraos/gemma-4-12B-it-qat-w4a16-ct) | Model card and release files |
-| [gemma-4-26B-A4B-it](https://huggingface.co/libraos/gemma-4-26B-A4B-it) | Model card and release files |
-| [L3.3-70B-Euryale-v2.3-AWQ](https://huggingface.co/libraos/L3.3-70B-Euryale-v2.3-AWQ) | Model card and release files |
-| [Cydonia-24B-v4.3-AWQ](https://huggingface.co/libraos/Cydonia-24B-v4.3-AWQ) | Model card and release files |
-| [MN-Violet-Lotus-12B-AWQ](https://huggingface.co/libraos/MN-Violet-Lotus-12B-AWQ) | Model card and release files |
-
-### Inference and assisted integration
-
-Use a compatible model provider or local inference. **Inference is separate from Runtime licensing and implementation services.** For MegaNova inference, manage top-ups, invoices and detailed usage at [MegaNova Console](https://console.meganova.ai/).
+Connect the SDK to a running LibraOS server. Your app keeps its user experience and business logic; LibraOS provides the Runtime.
 
-Need help with an integration? [Assisted pilots start at US$3,000](https://libraos.com/pricing/), one-time, for a defined scope under a written statement of work. Scope, schedule, acceptance criteria, dependencies and exclusions are agreed before work begins. Production commercial licensing is separate unless expressly included in your agreement.
+Self-hosting does not automatically keep external model or tool requests local. See [deployment boundaries](https://libraos.com/security/) and [license terms](https://libraos.com/download/#license).
 
-[Discuss a pilot](https://libraos.com/#demo) · [Support](https://libraos.com/support/) · [GitHub](https://github.com/libraos)
+Explore our model repositories below. Each model card documents upstream attribution, licensing and usage.
 
 Built by **Nebula Nova Inc.**