AI Services
AI Development
We develop the methods that turn your data into reliable results: machine learning, AI pipelines, agents, multimodal analysis, and hybrid models.
AI Technologies
Agentic AI
Individual model calls become controllable systems based on a division of labour: specialised agents handle subtasks, use tools, and work together in verified workflows.
Methods and techniques
- Orchestration: break down the task, distribute the work, merge the results
- Analysis and extraction agents, also in parallel
- Knowledge and retrieval agents
- Normalisation agents for terms, classifications, and formats
- Test and validation agents
- Human-in-the-loop: approval and correction by people
Application examples
- Agents for requirements, tests, fault analysis, and verification evidence in engineering
- Research assistant that collects source references, evaluates them, and prepares them for a decision
- Agentic tools in our own software development
What matters
Productive agents are not uncontrolled autonomy. Success depends on tools with clear permissions, structured interfaces, machine-verifiable feedback, and human approval for critical steps.
For us, agentic AI is not a buzzword but the consistent systematisation of complex AI processes: a task is broken down into subtasks, and each subtask is assigned an agent with a clearly defined role.
Principles for productive agents
| Principle | What it means |
|---|---|
| Tools and permissions | The agent may read, search, modify, and test, but not everything without control. Critical operations are restricted or require approval. |
| Goal instead of single step | Not “change file X,” but: “Implement function Y, verify it with tests, deliver the changes with a rationale.” |
| Context management | Rules, architecture, standards, and prohibitions are laid down in permanent guidelines. Large tasks are broken down. |
| Verifiable feedback | Tests and checking tools deliver a machine-readable pass or fail. |
| Roles and specialisation | Exploring, implementing, reviewing, security, and documentation are separate roles. |
| Governance through rules | What must always apply is laid down in automatic checks and policies, not only in the instruction to the model. |
| The human decides | Human approval for architecture decisions, breaking interface changes, data migrations, and security exceptions. |
Agentic coding
We also use agentic tools in our own software development. Coding agents analyse code, propose changes, write tests, find bugs, and add documentation. This practice helps us assess realistically what is possible in productive use today and where human control remains necessary.
Further topics
AI Technologies
Document & Knowledge Intelligence
Making large, heterogeneous bodies of information accessible and turning unstructured information into knowledge models that are usable for the business domain.
AI Technologies
Multimodal AI
Every data source has its strengths and limits. The result becomes reliable when text, images, audio, sensor data, and geodata are interpreted together.
Solutions
AI Agents and Intelligent Workflows
Agents for tasks in your company: they read documents, query databases, apply rules, and prepare decisions.
Solutions
Engineering Intelligence
From existing knowledge to a usable engineering system: linking requirements, tests, failure analysis, and evidence with AI.
Solutions
Document and Knowledge Intelligence
Understanding documents instead of merely reading them: archives, files, and technical documentation become structured knowledge you can query.
Would you like to make AI usable in your company?
Whether a first AI adoption or a complex development task – we discuss the requirements and suitable next steps with you.