Artificial Intelligence

Our team accompanies client projects from proof of concept to go-live, offering a wide range of services including data preparation, model development, testing, and deployment. Our focus areas are integration, knowledge transfer, hosting, and benchmarking.

At the same time, we actively drive several projects in AI research with a focus on AI models and robotics.

Generative AI and Large Language Models (LLMs)

Generative AI

• TNG AI Assistant TAIA: Our secure and sovereign AI assistant enables access to various LLMs. TAIA combines diverse tools for context enrichment and stores user inputs exclusively locally in the browser. Queries, documents, and responses are never logged.
• AI-generated presentations (slide decks)
• AI-generated children’s books
• AI assistant for dataset search via Text-to-SQL
• AI copilot for real estate exposés
• Atlassian Jira app: Rock your Sprint Review
• Creation and editing of portraits
• Chatbots with Retrieval-Augmented Generation (RAG) and agent-based functions
• Text-to-3D diffusion and training of text-to-image models

AI Productivity Tools

• Automated Unit Test Generation: Automated creation of unit tests for Java, Kotlin, C#, and TypeScript projects. The tool works autonomously and validates test quality through advanced methods while minimizing developer intervention.
• Automated End-to-End Test Generation: Creation of end-to-end tests based on instructions in simple natural language and generation of maintainable Playwright code. The tool runs locally with optional support for on-premise models and adapts tests to your team’s conventions.
• Automated code migration for Java as well as language conversions
• Analysis of code and technical architecture

Methodological Expertise

• Evaluation and benchmarking of models and pipelines with advanced methods, e.g. LLM-based evaluation
• Domain-specific fine-tuning of models, e.g. for Legal Tech, source code, or the German language

Computer Vision

• Defect detection and quality control in industries such as agriculture, high-tech manufacturing, and pharmaceutical wholesale
• Autonomous driving and flying, including Simultaneous Localization and Mapping (SLAM) and control
• Motion analysis for competitive athletes
• Biometric identification (face, iris, fingerprint)
• Security applications with edge computing in public spaces, e.g. video anonymization and touchless gesture control
• Research in the field of real-time deepfakes

Statistical Machine Learning

• Fraud and risk detection in fintech and telecommunications
• Predictive cost calculation in global supply chains
• Size recommendation for clothing in web shops

Reinforcement Learning

• Optimization of maintenance decisions using Monte Carlo Tree Search (MCTS)
• Solving satisfiability problems (SAT)
• Automatic adjustment of control parameters for industrial machines

Platforms & Hosting

Local Hosting

• Hosting of LLMs and AI services on GPU clusters, including embedding, transcription, translation, and re-ranking
• Infrastructure management

Cloud Services

• AWS ML Services
• Azure AI Services and Azure OpenAI
• Google Vertex AI
• Lambda Cloud and Hugging Face hosting

Edge Devices

• Deployment on compact computers like Raspberry Pi or NVIDIA Jetson Nano
• AI on microcontrollers such as ESP32
• Neural Compute Sticks