Process text data with AI

From procurement to cleansing and enrichment to in-depth analysis of large text datasets.

Now new: Our deepsight DataDesk
- DIY word processing in the cloud -

Sanity Check - Anonymization - Thematic analysis

Our promise

Too much text for manual processing of your collected comments or related texts? We use AI to analyze and evaluate text data sets of any size.

AI text services

With our data matching processes, we identify and link similar data records within large volumes of text data. This technique enables efficient data cleansing and enrichment by removing duplicates and intelligently merging related information. Our data matching plays a crucial role in market research and extensive customer surveys in particular. It helps to consolidate responses from different participants and enables a more in-depth analysis of opinion trends and customer preferences. We also use data matching to optimize customer records, eliminating redundant entries and precisely merging relevant customer data such as purchase history or interaction logs. This not only improves the quality of customer information, but also enables targeted communication and personalized marketing strategies. By applying advanced algorithms and machine learning, our data matching is continuously improved to achieve even more accurate and efficient results that directly contribute to increasing operational efficiency and supporting strategic decisions.
Our AI-supported sanity checks ensure that your text data records are free of inconsistencies, duplicate content and nonsensical comments. By automatically checking your data, we eliminate redundancies and ensure that all information is consistent and logical. This process is essential to ensure the quality and reliability of the data before further analysis. A precise sanity check not only helps to increase the efficiency of data usage, but also protects against incorrect conclusions and thus optimizes decision-making.
Our compliance check service offers a thorough review of your employee survey data for compliance-relevant information. Using AI technologies, we effectively filter out all sensitive data and forward it separately to the responsible parties. In this way, we guarantee that your data processing always complies with current data protection and compliance requirements, while you can gain valuable insights into your company. Our service is specifically designed to identify content such as sexism, racism and nepotism and ensure that such sensitive topics are immediately recognized and addressed. The comments and data concerned are automatically forwarded to the relevant departments, such as Human Resources or Compliance, to enable quick and appropriate responses. This process not only supports compliance, but also promotes a safe and healthy working environment. By accurately identifying and handling compliance issues, we help you minimize risk and protect the integrity of your business.
Our anonymization service uses advanced AI methods to identify and anonymize personal information in your text data. This process protects privacy and supports privacy compliance by hiding sensitive data while preserving the integrity and usability of the data for analytics. Additionally, we offer the option of pseudonymization, which allows you to safely use anonymized data for specific purposes such as training your own models. Our service is also designed to recognize and anonymize context-specific information. For example, a sentence like 'I was on maternity leave for the last 6 months last year' can reveal sensitive information in a small team. Our AI effectively recognizes and handles such nuances to ensure the anonymity of all stakeholders without compromising data quality.
Our sentiment analysis enables you to identify and quantify the moods and emotions in text data. This deep insight into customer opinions, employee feedback or market sentiment helps you to make informed decisions. Sentiment analysis is particularly valuable for identifying positive and negative trends in large data sets and adjusting strategies based on them. In addition to the basic sentiment analysis (positive, negative, neutral), our service offers a differentiated assessment of the emotional tone. This allows us not only to capture the general sentiment, but also to identify nuances such as 'rather positive' or 'very positive', which enables a finer assessment of the text content. Our offering also includes an aspect-based sentiment analysis, in which individual aspects of a text are evaluated separately. This makes it possible to identify specific strengths and weaknesses in customer feedback, such as the user-friendliness of a product or its price-performance ratio. Such detailed analyses are particularly valuable for companies that want to gain precise insights into the opinions and attitudes of their customers and make targeted improvements on this basis.
We offer three options for thematic analysis: Supervised Learning for exploring specific, pre-defined themes, Unsupervised Learning for automatically discovering new themes in your data, and a hybrid approach that combines both methods to provide flexibility and depth in analysis. These approaches are particularly valuable in a variety of application areas, such as employee surveys, market research, medical data evaluation and correspondence analysis. In practice, this means that we are able to identify and analyze both generic and highly specialized topics, depending on the specific requirements of your company or industry. For thematic analysis, we use both specially trained models tailored to a client's unique needs and data structures, as well as freely available models that cover broad use cases. The flexibility of our solutions allows us to generate thematic insights that are directly aligned with your strategic goals, be it to improve employee engagement, optimize market positioning or more efficiently organize and analyze clinical data. Our advanced algorithms can extract relevant themes from large volumes of data, providing in-depth insights that contribute to informed decision-making.
  • Supervised Learning Options: Better known as the "drawer system" - Our supervised learning options for thematic analysis allow you to precisely identify predefined themes in your text data. This method is ideal if you want to examine specific content or categories for which sample data is already available. The advantage lies in the accuracy and the ability to relate the results directly to your pre-defined questions.
  • Unsupervised learning options: Our Unsupervised Learning options offer the flexibility to automatically discover unknown patterns and themes in text data. This technique is ideal for gaining new insights without having to formulate specific hypotheses beforehand. The disadvantage can be less precision compared to Supervised Learning, but the advantage is the discovery of unexpected correlations.
  • Hybrid option: We recommend our hybrid approach, which combines the strengths of supervised and unsupervised learning. This approach makes it possible to analyze and expand repeated surveys or data sets over time to check whether new, previously undefined topics emerge. The hybrid approach thus provides a comprehensive solution that encompasses both the depth of predefined analysis and the breadth of exploratory discovery in your data.


We guarantee reliable results after just a few days, even for more complex analyses.


Your data is analyzed reliably and absolutely objectively - without any loss of concentration or subjective perception.


Semantic analyses provide you with reliable information about the meaning and correlations in your data.

What our customers say

dvj insights - Market research on end-user products & brands

"DVJ Insights focuses strongly on the topic "The Why behind the What". deepsight was able to help us and our clients decisively to make the free associations and customer stories collected by DVJ in large samples usable and to generate actionable strategies for brand growth."

Stephan Knaeble, Managing Partner dvj insights

If you are interested in our solutions or have any questions, please contact us at

Scroll up