Qualitative Research with Karl AI

From efficient to in-depth.

Expert prompts, interactive guidance, and a variety of language models—some GDPR-compliant.
Open to diverse methods and flexible—explore with AI in your own way.

Watch introductory videos (YouTube)

(For English translation, please use subtitles or YouTube translation option)

This is Karl AI

Evaluating qualitative data with AI
Expert-crafted prompts
Create and save your own prompts
Access to various providers (latest models from OpenAI, Mistral, and Anthropic)
Data hosted in the EU with Mistral
Additional models (such as Llama) coming soon

What Does Karl AI Cost?

A monthly subscription to Karl AI costs 24 euros, including VAT.

There are no additional costs for tokens, credits, etc. (for standard usage).

By subscribing, you can use AI models from multiple providers—such as Mistral, OpenAI, and Claude—without any extra charges.

Do you need more information?

Contact us via info@dokumet.de

Frequently Asked Questions

It’s best to just get started. Or you can take a look at our tutorials 😉

Please send an email to info@dokumet.de outlining your requirements (number of licenses, desired license duration, and any additional services such as workshops, API integration, or Single Sign-On (SSO)). We will get back to you promptly with an offer.

We currently offer GPT models from OpenAI, Claude models from Anthropic, and models from the French provider Mistral. Additional models will be added in the future; for example, the integration of the Swiss model Apertus is planned.

The models differ, among other things, in the number of parameters, the structure of their neural networks, and the algorithms they use. As a result, they can produce different outputs. The direction these results take also depends heavily on your input data (materials and prompts).

We recommend simply trying it out with a transcript: upload the same transcript to all models and compare the interpretations. This will give you a sense of how the different models behave and help you determine which one is most suitable for your research interests.

Karl AI is a tool for AI-supported interpretation that evolved from a QDA software, but places its main focus on AI integration.

Karl AI includes an interface for managing data (prompts, interpretations, interviews, etc.).

No, Karl AI is web-based. You can use the software directly in your browser.

The level of data protection varies depending on which language model you select: The data processing of the French provider Mistral is subject to the legal framework of the GDPR. The processing of data originating from the EU by U.S. providers such as OpenAI or Anthropic takes place under the EU Standard Contractual Clauses (SCC) provided for by the GDPR, which apply when neither a certification under the EU–US Data Privacy Framework nor an adequacy decision by the European Commission exists.

Accordingly, processing takes place on the basis of contractual assurances, but within a different legal framework in which, in particular, the disclosure of data to government authorities—such as for law enforcement purposes—is regulated differently and more broadly than under European data protection law.

No, as a general rule we do not allow any of the AI providers to train models using data from DokuMet AI.

It usually takes about 5 minutes for the subscription or credit to be processed. Please wait these 5 minutes.

You may also need to refresh the page. Try using Ctrl+F5, Shift+F5, or Ctrl+Shift+R.

If it still doesn’t work after that, please contact us.

That depends on the prompts used and on researchers providing sufficient context to the language models. In many cases, large language models produce elaborate interpretations. In any case, AI-generated interpretation offers you a new perspective on your texts. It is therefore worth trying out Karl AI.

Yes, but please note: AI-generated texts are not protected by copyright. In addition, according to the EU AI Act, texts that have not been edited by humans will need to be labeled starting from August 2026 onward.

Lieder, F. R. & Schäffer, B. (2023). Lehren Und Lernen Rekonstruktiver Forschungsmethoden mit Generativen Sprachmodellen in hybriden Forschungswerkstätten? Theoretische und empirische Befunde. Journal für Psychologie 31 (2):131-54. https://doi.org/10.30820/0942-2285-2023-2-131
Schäffer, B. Lieder, F. (2023). Distributed Interpretation - Teaching reconstructive methods in the social sciences supported by artificial intelligence. In: Journal of Research on Technology in Education, Special Issue “Reciprocal Issues of Artificial and Human Intelligence in Education”, 55:1, 111-124, DOI: 10.1080/15391523.2022.2148786. https://www.tandfonline.com/doi/full/10.1080/15391523.2022.2148786

No, Karl AI is an independent product. The further development of DokuMet QDA has been discontinued since September 2024, as we are now fully focusing on the development of a browser-based product.

Please log in to your account using your user credentials and change the data there. If that does not work, please send us an email.