As the adoption of Generative AI (GenAI) accelerates, so do the concerns around its trustworthiness in real-world business applications. On March 19, nearly 100 business and product leaders, startup founders, and risk/compliance professionals gathered at AWS Singapore for an eye-opening session: Testing Your GenAI Application — A Beginner’s Guide for Non-Technical Users.
Co-hosted by AI Verify Foundation, AWS Responsible GenAI Community, and The Generative Beings, this event was designed for non-technical stakeholders who are increasingly held accountable for AI outcomes — but often rely on technical teams to assess the risks.
Why does GenAI testing matter? Because in the real world, it’s not just about whether the technology works in theory — it’s about whether it works reliably, accurately, and safely in high-stakes environments.
The Basics of Building Trustworthy AI
The session broke down complex testing concepts into digestible, real-world applications.
Shameek Kundu from AI Verify Foundation kickstarted the sharing by joining the dots between the various testing practices and actual business applications and impact.
- 🔍 Benchmarking for Safety – Harry Zhao from AI Verify Foundation introduced the Moonshot toolkit, showing how off-the-shelf safety benchmarks can quickly evaluate if a GenAI model might produce violent, criminal, or harmful outputs.
- 🧠 RAG Accuracy Testing – Shahul ES from Ragas AI explained how Retrieval-Augmented Generation (RAG) testing ensures AI retrieves the right information from your data—moving beyond gut instinct to structured evaluation.
- 🥊 Red Teaming – Alex Leung and Sue Yen Leow of Vulcan demonstrated how simulating attacks (whether by humans or another AI) can expose weaknesses before your app goes live.
- 🛡️ Guardrails in Production – Safeer Mohiuddin from Guardrails AI walked us through lightweight yet powerful mechanisms to enforce rules even after deployment.
- ⚙️ Bringing it Together with AWS – Dr. Alessandro Cerè from AWS shared upcoming capabilities within AWS Bedrock to embed RAG and guardrails into your workflows natively.
Why Non-Technical Leaders Should Care
AI testing is not just a technical task—it’s a strategic responsibility. As one participant noted, “We’re not here to test for the sake of testing. We’re here to make sure what we build actually works, and works safely.”
Through live demos and practical walkthroughs, the event empowered non-technical leaders to ask the right questions, collaborate better with tech teams, and take ownership of trustworthy AI adoption.
Recap of our community moments
