Services
Our services can be booked individually and are flexibly adaptable to the risk class, sector and maturity level of the test subject.
EU AI Act Compliance Check
Preparing for regulatory requirements
Alignment with the EU AI Act, ISO/IEC 42001 and other requirements. Classification of AI systems (risk level), transparency and bias analyses, support with documentation and audit preparation.
from CHF 220 per hour
Model Risk Assessment
Model Risk Assessment
Assessment of training data, model architecture and operational context; creation of a risk profile, including classification according to potential damage.
from CHF 12,000
LLM Audit
Evaluation of language models
Targeted testing of large language models for hallucinations, susceptibility to prompt injection, data leaks, bias and potential for misuse, including a structured audit report.
from CHF 6,000
RAG Reliability & Risk Audit
Analysis of retrieval-based AI systems
Review of the retrieval pipeline and analysis of response quality, source validity and performance, including risk assessment.
from CHF 15,000
Continuous monitoring
Continuous monitoring
Ongoing monitoring of model behaviour in production, including drift, anomaly and error detection.
from CHF 2,500 per month
Governance & Documentation
Governance & Documentation
Development of policies, roles, approval processes and technical documentation, including a governance handbook.
from CHF 26,000
-
Preparing for regulatory requirements
EU AI Act Compliance Check
Alignment with the EU AI Act, ISO/IEC 42001 and other requirements. Classification of AI systems (risk level), transparency and bias analyses, support with documentation and audit preparation.from CHF 220 per hour
-
Model Risk Assessment
Model Risk Assessment
Assessment of training data, model architecture and operational context; creation of a risk profile, including classification according to potential damage.from CHF 12,000
-
Evaluation of language models
LLM Audit
Targeted testing of large language models for hallucinations, susceptibility to prompt injection, data leaks, bias and potential for misuse, including a structured audit report.from CHF 6,000
-
Analysis of retrieval-based AI systems
RAG Reliability & Risk Audit
Review of the retrieval pipeline and analysis of response quality, source validity and performance, including risk assessment.from CHF 15,000
-
Continuous monitoring
Continuous monitoring
Ongoing monitoring of model behaviour in production, including drift, anomaly and error detection.from CHF 2,500 per month
-
Governance & Documentation
Governance & Documentation
Development of policies, roles, approval processes and technical documentation, including a governance handbook.from CHF 26,000
We offer you
-
Stable systems and long-term reliability
AI models and applications are regularly tested whilst in operation. Changes in behaviour, quality and performance are identified at an early stage, deviations are documented in a traceable manner and addressed in a targeted way. -
Traceability and reproducibility
Responses, decision-making logic and data flows are analysed in a structured manner. Risks such as hallucinations, bias or inconsistent results are identified, assessed and mitigated. -
Auditable AI systems
All testing processes, results and adjustments are documented transparently. Models remain traceable, decisions verifiable, and requirements arising from audits and regulatory bodies can be met at all times. -
Basis for decision-making
Realistic test datasets enable AI systems to be assessed in a targeted manner under defined conditions. Quality, robustness and limitations can be quantified.
-
Stable systems and long-term reliability
AI models and applications are regularly tested whilst in operation. Changes in behaviour, quality and performance are identified at an early stage, deviations are documented in a traceable manner and addressed in a targeted way.
-
Traceability and reproducibility
Responses, decision-making logic and data flows are analysed in a structured manner. Risks such as hallucinations, bias or inconsistent results are identified, assessed and mitigated.
-
Auditable AI systems
All testing processes, results and adjustments are documented transparently. Models remain traceable, decisions verifiable, and requirements arising from audits and regulatory bodies can be met at all times.
-
Basis for decision-making
Realistic test datasets enable AI systems to be assessed in a targeted manner under defined conditions. Quality, robustness and limitations can be quantified.
Using AI Safely
We begin every collaboration with a no-obligation initial consultation, during which we work with you to assess your specific use case, the regulatory context and the appropriate scope of testing.