All 12 modules of the deepsight Cloud at a glance: from topic analysis and sentiment through anonymization to dashboards and AI assistant. Learn how the modular platform works.
Text data is everywhere: in customer surveys, support tickets, product reviews, social media comments, and employee feedback. But to extract real insights from this data, you need more than a single tool. You need a platform that covers the entire analysis workflow — from data quality through analysis to visualization.
That is exactly what the deepsight Cloud is: a modular AI platform for professional text analysis, consisting of 12 specialized modules that can be used individually or in combination. In this article, we introduce all modules, explain their functions, and show how they work together.
The Principle: Modular Architecture for Maximum Flexibility
Why 12 modules instead of a monolithic solution? Because text analysis projects differ. A market research institute evaluating survey data has different requirements than a CX team analyzing NPS comments. Or a product team trying to understand app reviews.
The modular architecture of the deepsight Cloud offers three key advantages:
- Flexibility: Use only the modules you actually need
- Scalability: Add modules as your requirements grow
- Integration: Modules work seamlessly together and share results
The 12 modules are organized into three categories: Analyze (core analysis modules), Quality (data preparation and protection), and Understand (visualization and deep-dive).
Category 1: Analyze — The Core Modules
These five modules form the analytical heart of the platform. They extract topics, sentiments, rating patterns, and intents from your text data.
Module 1: Topic Analysis
Topic analysis is the foundation of every text evaluation. The module automatically identifies what people are talking about — without predefined category lists. It uses topic modeling algorithms combined with transformer models to cluster semantically related statements into topic groups.
An example: from 5,000 hotel reviews, the module extracts topics like "Room Cleanliness," "Breakfast Quality," "Location and Accessibility," "Staff and Service," or "Value for Money" — fully automatically. Analysts can then refine and merge the topics.
Module 2: Sentiment Analysis
Sentiment analysis determines the emotional tone of texts — aspect-based, with intensity gradations, and contextual understanding. The module recognizes mixed sentiments within a single sentence and maps sentiment to individual topics.
Special feature: the system is specifically optimized for German texts and masters the peculiarities of the German language — compound words, modal particles, irony, and regional expressions.
Module 3: NPS Analysis
The NPS module connects quantitative scores with qualitative comment analysis. It automatically segments comments by Promoters, Passives, and Detractors and displays topic-specific NPS values: which aspects drive the score up, which drive it down?
This transforms the NPS score into a detailed driver model that enables concrete action recommendations.
Module 4: Intent Detection
While sentiment tells you how someone feels, intent tells you what someone wants. The Intent module detects customer intentions such as purchase interest, churn signals, feature requests, complaints, or praise. Especially valuable for proactive action: when the AI detects a churn indicator, teams can respond before the customer leaves.
Module 5: Training
The Training module enables you to adapt AI models to your specific domain. Through targeted fine-tuning with your own data, you significantly improve detection accuracy — whether for insurance terminology, automotive feedback, or medical texts.
The process is intuitive: you label sample data via a graphical interface, start the training, and see accuracy improvements in real time.
Category 2: Quality — Prepare and Protect Data
Good analysis requires good data. The three Quality modules ensure that your text data is clean, complete, and legally compliant — before the analysis begins.
Module 6: Sanity Check
The Sanity Check examines your data for quality issues before it flows into analysis: duplicates, empty responses, non-relevant texts (e.g., "no comment," "asdf"), language mixtures, and outliers. This saves time and prevents skewed results.
In practice, a typical dataset contains 5–15% problematic entries. Without a Sanity Check, these entries distort your topic and sentiment distributions.
Module 7: Anonymization
Text data frequently contains personally identifiable information: names, email addresses, phone numbers, customer IDs. The Anonymization module detects and masks this data automatically — GDPR-compliant and reliable.
This is not only a legal necessity but also a prerequisite for internal teams to work with the data without concerns.
Module 8: Translation
Global companies receive feedback in many languages. The Translation module translates texts into a unified language before analysis begins — so you can compare across countries without needing separate models for each language.
The translation is AI-based and context-aware — not simple word-for-word transfer, but a translation that preserves meaning and tone.
Category 3: Understand — Visualize and Explore Results
Analysis results are only useful when they are understood and shared. The four Understand modules make results accessible, interactive, and shareable.
Module 9: Dashboards
The Dashboard module transforms analysis results into interactive visualizations: topic distributions, sentiment trends, NPS driver analyses, word clouds, and more. Filters allow you to zoom by time period, source, segment, or topic.
Dashboards can be shared and exported — ideal for reports, presentations, and stakeholder updates.
Module 10: Segment Analysis
Segment analysis compares results across different groups: customer types, regions, products, channels, or time periods. Where do segments differ? Where are there surprising similarities? These comparisons reveal patterns that get lost in overall evaluations.
Module 11: Refinement
The Refinement module enables manual post-processing of automatically generated results. Analysts can correct topic assignments, merge or split categories, and handle edge cases. These corrections flow back into the system as training data and improve future analyses.
Module 12: AI Assistant
The AI Assistant allows you to query analysis results in natural language: "What are the top 3 complaints from Detractors in Q3?" or "Which topics improved the most compared to last year?" The assistant accesses the analysis data and delivers precise, contextualized answers.
This makes complex analysis results accessible even to stakeholders who do not work with the tool directly.
How the Modules Work Together: A Typical Workflow
In practice, modules are rarely used in isolation. A typical end-to-end workflow looks like this:
- Import data: CSV, Excel, or API connection
- Sanity Check: Review and clean data quality
- Anonymization: Mask personally identifiable data
- Translation: For multilingual data, translate into one analysis language
- Topic Analysis: Automatic topic extraction
- Sentiment Analysis: Determine sentiment per topic
- NPS/Intent: Specialized analyses depending on the question
- Refinement: Review and optimize results
- Dashboards: Visualization and reporting
- AI Assistant: Answer ad-hoc questions
This workflow runs largely automatically. After initial configuration, a data import is all it takes, and the platform delivers complete analysis results within minutes.
Who Is the deepsight Cloud For?
The platform is designed for teams and organizations that regularly evaluate large volumes of text data:
- Market research institutes: Efficiently evaluate open survey questions and prepare results for clients
- CX teams in enterprises: Systematically analyze NPS comments, complaints, and customer feedback
- Product teams: Understand and prioritize app reviews and feature requests
- HR departments: Evaluate employee surveys and engagement data
- Consultancies and agencies: Offer text analysis as a service for their clients
Security and Data Protection
The deepsight Cloud is operated in German data centers and is fully GDPR-compliant. All data is encrypted in transit and at rest. Personal data can be automatically anonymized before analysis.
For companies with particularly high security requirements, we also offer on-premise installations and dedicated instances.
Conclusion: One Platform Instead of Ten Tools
The deepsight Cloud combines everything you need for professional text analysis — in a seamless, modular platform. Instead of combining different tools and manually transferring data, you work in one system where data quality, analysis, and visualization seamlessly interlock.
12 modules. Three categories. One goal: turning text data into real insights — fast, reliable, and scalable.
Curious? Discover all modules of the deepsight Cloud or
