AI-Powered Requirements-Based Testing: Vector Releases 'VectorCAST 2026'

Stuttgart, Germany, March 19, 2026 – Vector Informatik GmbH (Vector) has released the latest version of its test automation platform for embedded software, 'VectorCAST 2026.' This version introduces an AI-powered 'Requirements-Based Test Creator.' This new feature enables the direct generation of unit tests from software requirements, improving traceability in safety-critical systems, simplifying certification, and streamlining quality assurance processes.

The advancement of generative AI in software engineering offers new possibilities for embedded software development. However, risks must be considered. While AI tools can automate repetitive tasks and boost productivity, safety-critical fields require managed processes, transparent traceability, and proper human oversight. Without these safeguards, AI-generated artifacts can increase review efforts and potentially degrade software quality.

To address this, VectorCAST 2026 introduces a new AI-assisted toolchain called 'Reqs2x.' Reqs2x automatically generates unit tests directly from software requirements. By combining advanced code analysis with AI, Reqs2x bridges the gap between requirements and implementation, ensuring sufficient testing and traceability for all functions. Key features of Reqs2x include: - Automatic Mapping to Requirements: Reqs2x automatically maps requirements to the code implementing them, streamlining the test creation process. Requirements can be imported from common management tools like DOORS or Polarion, or via CSV files through the Requirements Gateway. - AI-Assisted Test Generation: VectorCAST uses program slicing and Large Language Models (LLMs) to generate executable test cases aligned with requirements, reducing manual effort while maintaining traceability. - Human-in-the-Loop Review: The generated mappings and test cases are designed for user review, ensuring compliance with safety and quality standards.

To meet enterprise requirements for governance and data protection, VectorCAST 2026 allows customers to choose their own AI models. The 'Bring-Your-Own-Model' (BYOM) approach supports various infrastructures, including on-premise, cloud, and hybrid, giving customers full control over data privacy, compliance, and costs. Vector provides simple configuration connectors, while customers operate and manage the AI infrastructure.

Vector's AI strategy is designed for precisely defined use cases within a controlled context, based on modular, task-specific micro-agents. This ensures predictable behavior, facilitates verification, and supports alignment with industry standards such as MISRA C:2025, ISO/IEC AI frameworks, and the EU AI Act. Consequently, AI-generated artifacts are subject to the same verification and quality control as human-written code.

As part of Vector's comprehensive software ecosystem, VectorCAST 2026 is integrated into automated quality monitoring pipelines, supporting test automation results with advanced analysis and rich reporting. This enables meaningful quality metrics that provide transparency across the entire lifecycle, fostering continuous improvement.

VectorCAST is adopted by leading companies in aerospace, automotive, medical devices, industrial control, and railway sectors. With the release of VectorCAST 2026, Vector strengthens its embedded software testing solutions, focusing on automation, AI assistance, and traceability for safety-critical systems. The Reqs2x feature is provided at no additional cost to existing VectorCAST 2026 license holders.