Large language models in findIQ

Where LLMs are used

LLMs take over the language part in findIQ: understanding requests, reading documents and phrasing answers. The troubleshooting itself is done by the findIQ diagnostic AI, not by an LLM.

LLMs are used in one function: Quinn, the findIQ AI assistant. Quinn is enabled per organisation and can take over knowledge in several ways:

• Documents, such as manuals, are turned into findIQ content

• Expert knowledge is recorded in the chat

• Videos are turned into routine drafts

• Existing symptoms and causes are structured, and questions about your data are answered

Model and hosting

• Quinn runs on Luna, a model from OpenAI, hosted in Microsoft Azure (Europe).

• The OpenAI models run in Microsoft's Azure environment. Your data is not available to OpenAI.

• findIQ selects and maintains the models and may change them over time. Regardless of the model, your data is never used for training.

• We inform you in advance if the model provider changes.

How your data is processed

• Quinn does not search a file store. It reads and writes the structured findIQ data model, so the source of each answer can be traced.

• Processing takes place at the model providers. Your data is stored exclusively on findIQ servers in the EU.

• Documents are analysed for one purpose only: to extract the relevant knowledge and make it usable in findIQ.

Is my data used for training?

No. Your data is not used to train, retrain or improve AI models, neither by the model providers nor across customers at findIQ. Technical processing such as embeddings only takes place to provide the service for you. There is no cross-customer retrieval.

What an AI is allowed to do

• The user signs in with their own findIQ credentials. The AI can never do more than that person.

• Deleting machines and templates is blocked, even if the signed-in person would be allowed to.

• Quinn is activated per organisation and can be disabled at any time by your findIQ contact.

Using your own AI system

Instead of Quinn, you can connect your own AI system, for example Microsoft Copilot, Claude or ChatGPT, via the findIQ MCP server. You then choose provider, model and cost model yourself, and data processing by that system is governed by your contract with the provider. The same permission rules apply: the AI acts with the rights of the signed-in user, and deleting is blocked.

More: Using findIQ in Microsoft Copilot and other AI systems and What Is an MCP Server?

Sub-processors

All providers involved are listed with their processing purpose in our data processing agreement (DPA), see Certifications and contracts.