# At A Glance (/docs/coreai/at-a-glance)



CoreAI is the layer of BullSequana AI that turns the Runtime foundation into usable GenAI product capabilities.

What CoreAI Provides [#what-coreai-provides]

* User-facing entry points for GenAI applications and agents
* Stable backend APIs to orchestrate application logic and AI workflows
* Model lifecycle services for importing, registering, installing, and exposing models
* LLM proxying and MCP integration for governed access to models and tools
* Vector and document services for retrieval, parsing, and knowledge workflows
* LLM observability and telemetry for operating AI features in production

Core Capabilities [#core-capabilities]

* LLM serving and model access for production GenAI workloads
* Model management across registration, versioning, and deployment flows
* RAG enablement through document parsing and vector retrieval services

Who It Serves [#who-it-serves]

| Audience       | What CoreAI provides                                                           |
| -------------- | ------------------------------------------------------------------------------ |
| End users      | A web portal and APIs for interacting with AI capabilities                     |
| Product teams  | Reusable services for GenAI applications, document workflows, and agents       |
| Platform teams | A structured AI application layer built on Runtime services                    |
| Architects     | A clear separation between infrastructure concerns and AI application concerns |

CoreAI In One Sentence [#coreai-in-one-sentence]

Runtime runs the platform. CoreAI turns that platform into usable GenAI building blocks.

Practical Entry Points [#practical-entry-points]

* End users typically work through the CoreAI Web Portal
* Developers should integrate with the CoreAI API as the stable application surface
* Internal CoreAI services can evolve behind that API without forcing client-side rewrites