Quinn – the findIQ LLM Assistant

Written By Henriette Hellweg

Last updated About 6 hours ago

Article goal: After reading this article, you will know what Quinn is, how it differs from a general-purpose LLM, which two routes connect an LLM-based chatbot with findIQ, and in which situations Quinn helps in everyday work.

Relevant for the Editor and Expert roles. Quinn is currently in beta.

What is Quinn?

Quinn is the findIQ AI assistant. Quinn operates the whole application the way a person would — chat-based, in natural language, and grounded in verified findIQ knowledge.

  • Knowledge base in minutes instead of weeks — Quinn evaluates existing sources and transfers them into findIQ in a structured way.

  • Available everywhere — in findIQ, in chat, or through the MCP server in your own tool.

  • Who is it for? — Editor and expert.

The technology difference: Quinn vs. LLM-only

A general-purpose LLM searches existing documents. Quinn closes the loop of knowledge transfer: knowledge is captured, structured, and made usable again.

LLMs search existing documents

Quinn captures knowledge and structures it

Existing documents — manuals, PDFs, and files: someone else's or outdated knowledge instead of relevant plant knowledge.

LLM — searches the file storage and formulates an answer: the more complex, the higher the risk of hallucinations.

Employee — receives an answer without a source: nothing is verified or approved.

Result: no feedback loop.

Experts and existing documents — relevant experience from people's heads is meaningfully combined with existing documents and data.

Quinn — classifies and structures the knowledge: gaps are detected and asked about, not guessed.

findIQ AI model — provides the structure (scalable and manageable even at high complexity), makes knowledge available deterministically, and learns with every case.

Employees — work with it on site: every answer with origin and approval.

Architecture: two access paths, one platform

Underneath findIQ sits an MCP server — the interface for AI systems. Both access paths use the same data:

  • Human → findIQ app — the interface as usual.

  • LLM-based chatbot → findIQ MCP server — Quinn, ChatGPT, Copilot, or Claude access documents, routines, and heatmaps through the interface.

🔒 Restrictions for LLM chatbot systems are built in: templates and machines cannot be deleted, for example.

Two ways to connect LLM-based chatbots with findIQ

Option A · Quinn – ready to go

Option B · Your own AI system via the MCP server

Setup

Ready to use immediately, no setup required

Your own AI system (Copilot, Claude, ChatGPT & Co.) is connected

Provider and model

findIQ selects provider and model for you

You choose provider, model, and costs yourself – licence or pay-as-you-go

Scope of functions

Additional capabilities beyond the MCP server

Functions of the MCP server

A good fit if

You want to start right away, have no AI system of your own, or need the Quinn extras

An AI strategy is already set or several systems need to work together

⚠️ Uploads of large files are currently technically limited.

Use cases

Quinn is currently optimised for processing documents and transferring them into the findIQ method. Further use cases are planned.
We look forward to your feedback!

1. Create a template from unstructured knowledge

  • Who is it for? Editor building a knowledge base from documents (PDF, Word, Excel).

  • Problem today: Structuring the knowledge is time-consuming.

  • How Quinn solves it: Analysis and structuring of all data according to the findIQ method.

  • Benefit: Quinn evaluates knowledge sources reliably and transfers them into findIQ.

2. Collect and evaluate expert knowledge

  • Who is it for? Editor or expert who wants to bring field knowledge into the system.

  • Problem today: Expert time is valuable and scarce, and experts are not familiar with the findIQ method.

  • How Quinn solves it: Knowledge is captured asynchronously, then structured and entered by Quinn.

  • Benefit: Quinn saves resources by capturing knowledge in a structured way in findIQ, regardless of when and where.

3. Structure knowledge

  • Who is it for? Editor structuring complex fault knowledge to optimise troubleshooting.

  • Problem today: Structuring is tedious in large knowledge bases. Without structure, troubleshooting takes long.

  • How Quinn solves it: Quinn analyses symptoms quickly and reliably and structures them by predefined or custom categories — for example as a labelled heatmap.

  • Benefit: Large knowledge bases are analysed and edited reliably according to your specifications.

4. General use

  • Who is it for? Editor or admin who wants an overview.

  • Problem today: Knowledge is nested hierarchically — getting an overview means searching everything.

  • How Quinn solves it: Quinn finds information instantly and summarises it as needed.

  • Benefit: Quinn has access to all information in the organisation and can read, edit, and provide it.

Prompt best practices

The use cases work better the more precisely the prompt says where Quinn should work — and the less it explains how findIQ is built. The following points come from practical work with Quinn.

1. Name the section, not the whole document. An anchor in the prompt — plant, assembly, section, table range — keeps the data volume small. The result comes faster, is shorter, and is easier to check.

  • Instead of: "Analyse this operating manual."

  • Better: "Use only the cover sheet and, from the troubleshooting table (T-01 to T-09), the faults of assembly AS-3."

2. Don't explain what findIQ already knows. The structure, setup, and logic of the heatmap are predefined. Describing them in the prompt makes it longer without making it better.

  • Instead of: "Build a heatmap with symptoms as rows, causes as columns, and probabilities based on frequency."

  • Better: "Add the tickets for component AS-6 to the heatmap."

3. Say which object to work in. Without a target, Quinn may create something new instead of extending what exists.

  • Example: "Work in the existing template Bottle filler and do not create a new one."

4. Work in steps instead of one large prompt. Add, label, check — each step on its own. The result stays traceable, and if one step doesn't fit, you don't have to repeat everything.

5. Specify or let Quinn suggest — both are useful, but different. Concrete specifications produce reproducible results; open questions make use of Quinn's view of the data.

  • Reproducible: "Create a label Pressure system and label all symptoms and causes belonging to the pressure system (CO2 pressure)."

  • Collaborative: "I want to label symptoms and causes by fault type. Which labels do you suggest?"

6. Use the terms of the application. Template, label, routine, symptom, cause, logbook — each of these terms means something specific in findIQ. Calling an Excel file "logbook" when the logbook is a feature invites misunderstandings.

7. Operate the application where the application is faster. Filtering, sorting, and clicking through is more direct in the interface than via a prompt. Quinn is strong at evaluating, structuring, and creating.

8. Decide in advance what the result should be. If you know the expected answer — for example "four faults, because four belong to AS-3" — you spot deviations immediately.

💡 No reason to worry if a prompt isn't perfect. Quinn knows the findIQ method and makes sure to follow it. If information is missing, Quinn asks instead of guessing. The more precise the prompt, the more efficient the processing — but it is not a prerequisite: Quinn helps you get the best possible data quality out of your documents in any case.

Quinn in the beta

  • Now: Activation for interested customers.

  • Mid-September: Active feedback round.

  • End of September: Beta ends, further information will follow.

💶 Quinn is free of charge during the beta. Pricing information will follow towards the end of the beta.

FAQ