Webinar Description
Key Takeaways
- Focuses on establishing deterministic trust in AI-generated code for safety-critical applications
- Demonstrates GNAT Foundry: Intersection, a new tool for verifying AI-produced artifacts
- Addresses aerospace, defense, automotive and critical infrastructure software development
- Explores how non-AI verification tools can reduce reliance on manual code review
- Examines the cost implications of AI adoption in high-integrity software environments
About the Webinar
AdaCore presents “Demonstrating Trustworthy AI Development in High-Integrity Environments,” a live webinar led by Tony Aiello, Product Manager for SPARK and AI. The session introduces GNAT Foundry: Intersection, a demonstration showcasing how deterministic verification tools can establish trust in AI-generated code without relying solely on human review.
The webinar addresses a fundamental challenge facing organizations that develop software for safety-critical systems: how to incorporate AI-assisted development while maintaining the rigorous assurance standards required for certification. In environments where software failures can have serious consequences, traditional approaches to validating AI-generated code through manual review do not scale effectively.
Deterministic Trust and AI-Generated Code
The core concept presented is deterministic trust, an approach that uses non-AI verification tools to validate AI-produced artifacts. Rather than treating AI-generated code as inherently untrustworthy and requiring exhaustive human inspection, this methodology applies static and dynamic analysis tools to provide measurable assurance. Human expertise can then be directed toward areas where judgment and domain knowledge add the most value.
AdaCore’s toolchain, including GNAT Foundry and SPARK, provides the verification capabilities central to this approach. SPARK in particular enables formal verification of code properties, offering mathematical proof of correctness rather than relying on testing alone.
Industry Applications
The techniques discussed apply to industries where software must meet stringent reliability and safety requirements. Aerospace and defense systems, automotive software subject to functional safety standards, and critical infrastructure all require documented assurance that code behaves as specified. Organizations in these sectors face pressure to adopt AI development tools while maintaining compliance with certification frameworks.
Who Should Attend
This webinar is relevant for software engineers, quality assurance professionals and technical leaders responsible for high-assurance systems. Compliance officers and product managers working on safety-critical applications will find value in understanding how tool-based verification can support certification efforts. The session suits those evaluating how AI development practices can be integrated into existing software assurance processes.