How to Use AI Features in Vertec with Full Sovereignty

Combining Vertec with AI features can offer enormous benefits. But not every company can work with all AI services. Regulatory requirements or concerns about data sovereignty require special setups. The goal is to maximize AI benefits with minimal risks.

The good news is that Vertec has designed the AI features in such a way that confident setups are possible.

1 Vertec MCP Server

The Vertec MCP server is very convenient to use with Claude from Anthropic. But what can you do if the big US provider is out of the question? Two components that Anthropic combines in Claude must be replaced. An AI client, i.e. the software that provides the chat. And the LLM, which processes the textual requests.

AI clients for sovereign MCP usage

Let’s first look at the AI client. Instead of Claude or ChatGPT, we look for options that offer more sovereignty in our scenario. Two solutions meet these needs particularly well: Librechat and Open webui.

Both solutions can run on servers of your choice. This is the significant difference between Claude and ChatGPT. Once you have installed one of the clients, configuration for the Vertec MCP server can begin. The steps are documented for both tools in the vendor’s Knowledge Base.

Important: These two solutions support OAuth and can therefore perform the authorization to Vertec.

Host LLM

With LibreChat or OPEN WebUI, sovereign solutions are possible for the connection AI-Client <> Vertec MCP Server <> Vertec. But now there is still a need for an LLM that understands the user requests, calls the appropriate MCP tools and returns the results in a comprehensible way.

Theoretically, you can run many so-called OPEN WEIGHT models yourself, but hardware and hours are expensive. This gap is filled by vendors such as Infomaniak and Exoscale. They make it possible to run models such as Mistral or Qwen on their hardware, in Switzerland, Germany and Austria. As a customer, you receive API keys for the selected models and can access them from the AI clients.

2 RAG projects with Squirro

The explanations on hosting Llm are also relevant for RAG systems. Instead of having their own AI know-how management module, Vertec customers can order a RAG project directly from Squirro, including the integration of their Vertec data. For structured business data, Vertec relies on MCP. You can find out why we chose this route in the blog The vertec mcp server has outgone the rank of the rag system.

a confident AI setup for Vertec is realistic, but requires a bit more hours than the established solutions of the leading AI services. The benefits of such a setup are:

• No lock-in at the LLM providers, easy change is possible at any time
• ISO27001 certified hosting providers for the LLMs ensure a high level of technical protection and data protection
• Self-hosted AI clients provide more sovereignty than Claude or Chatgpt

Want to learn more about the AI features in Vertec? Watch the Video Overview for insights and tutorials.