Two safety nets keep AI agents reliable, and they work at different moments.
IBM’s Tejas Kumar and Werner Heijstek, SIG’s Chief Delivery Officer, show you how evals and harnesses fit together in the agentic SDLC.
Date | November 19, 2026
Time | 11:00 AM CEST
Already registered via our LinkedIn event?
We’ll send the dial-in link directly to your inbox.
AI agents have already moved from pilots into production. They talk to customers, handle data and take actions on your organization’s behalf. Each of those actions carries risk, and the people accountable for it need confidence that the agent will behave as intended.
Two safety nets keep agents reliable, and they work at different moments. Evals test the agent against anticipated scenarios before it ships. Harnesses watch what the agent receives and does once it is live.
In this webinar, IBM’s Tejas Kumar and Werner Heijstek, SIG’s Chief Delivery Officer, show you how the two fit together.
➡️ A clear mental model of evals and harnesses, and when each one earns its place.
➡️ A realistic view of how far to trust evals.
➡️ How to decide which scenarios to test, and why those human choices shape what you catch.
➡️ How to protect architecture and quality when agents write the code, using checks they cannot argue with.
➡️ How experienced engineers judge whether software is ready for production.
Engineering leaders and product owners putting AI agents into production, and the executives who sign off on the risk.
Meet your speakers
Software Improvement Group (SIG) empowers organizations to govern the software their business runs on. Through complete portfolio analysis and tailored strategic advice, SIG helps companies embrace AI with control, improve software quality and security by focusing strategic efforts across people, process, and technology.
Sigrid®—its software portfolio governance platform—analyzes over 400 billion lines of code across more than 40,000 systems and 300+ technologies, offering evidence-based insights to help organizations prioritize and manage their most critical IT initiatives.
Founded in 2000 and headquartered in Amsterdam, SIG has offices in New York, Copenhagen, Brussels, and Frankfurt. The company complies with leading ISO/IEC standards, including 27001 and 17025, and co-developed ISO/IEC 5338—the new global standard for AI lifecycle management.
Combining expert consulting with over 25 years of industry-leading research, SIG is the global authority on continuous software portfolio governance.
