MCP and open AI models: keeping knowledge automatically up to date

The knowledge base is the foundation of every Wingmate. But knowledge dates quickly: products change, processes get updated, new insights emerge. Anyone maintaining their knowledge base only manually is fighting time. Today we introduce two building blocks that make the foundation more robust — and more maintainable in the long run.

🔌 Model Context Protocol: the bridge to your existing systems

The Model Context Protocol (MCP) is an open standard that lets AI systems access external tools and data sources in a structured way. For violet dragonfly, this means: existing systems — CRM, CMS, project management, document management — can populate, update and reorganise the knowledge base directly and automatically.

In practice: a new product data sheet is captured in the document system. Via the MCP connection, the content automatically lands as a curated knowledge entry in the Wingmate. No manual transfer, no forgotten updates. Knowledge stays current because it is created where it is created — and flows from there directly.

MCP integration suits both remote maintenance by internal teams who need no direct platform access, and fully automated synchronisation processes. Integration rather than a parallel world — your existing enterprise systems and the Wingmate work together instead of side by side.

🤖 Open AI models: control over the model stack

Violet dragonfly is not tied to a single AI model. An open-source LLM can already be selected in the standard offering — and this building block is growing in importance.

Open models like Llama (Meta), Mistral or Qwen have caught up substantially in recent months. For many use cases they produce results that hold their own against proprietary solutions. At the same time they offer something no commercial provider can guarantee: full transparency over the model stack, no external data flows, no vendor lock-in.

In practice: for organisations with specific compliance requirements, or simply for those who want to keep control over their AI infrastructure, an open model is no longer a compromise solution — it is a fully equivalent alternative. No content leaves your own infrastructure. No training on your data. Complete traceability.

Both building blocks — MCP integration and open models — pursue the same goal: the platform should fit into existing ways of working rather than replace them. Knowledge flows to where it is needed. The model that answers stays under control. And the curated core that makes the Wingmate reliable remains in place — now with easier maintenance and broader system support.