Agent deployment jumped from 11% of organizations to 53% in two years, but only 23% see real ROI from AI agents. The gap isn’t the model. It’s what happens around it.
Listen to discover what successful organizations do differently, and how the right context and guardrails can help your AI agents deliver real business value.
This podcast goes live on August 12 at 9:00 AM CEST.
Follow our podcast channels:
Werner Heijstek talks with Pepijn van de Kamp, SIG’s Director of Research & Innovation, about the agentic AI development lifecycle and how the Model Context Protocol (MCP) has become critical for different stages.
They unpack the needs and different types of control per development stage harness engineering, the discipline of designing the full environment an agent works in, and the importance of letting developer agents check their output against deterministic, real-time signals instead of relying on confident guesswork.
You’ll learn how to:
Werner presents proof for one of the stages. SIG conducted a controlled experiment putting Claude Sonnet 4.6 to the test, building the same system with and without real-time quality and security guardrails. High-risk security findings dropped about 97%, maintainability rose roughly 15%, and every guided build beat every unguided one.
For technology leaders weighing how to scale agents responsibly, this episode makes the case for what needs to be in place before you do, grounded in SIG’s own research and the State of Software 2026 report.
Meet your speakers
The SIGNAL podcast, powered by Software Improvement Group, is a monthly show where we turn complex IT topics into business clarity. Hosted by Werner Heijstek, Chief Delivery Officer at Software Improvement Group.
Follow us on LinkedIn for updates and episode highlights, and subscribe on Spotify or YouTube to catch every new SIGNAL episode as soon as it drops.
Govern the software your business runs on.
Software Improvement Group (SIG) enables organizations to embrace AI with strategic control and to become more agile & resilient through secure, high-quality software.
Sigrid® – its software governance platform – leverages the world’s largest software metric database containing over 400 billion lines of code across more than 20,000 systems and 300+ technologies, and intelligently recommends the most crucial initiatives for organizations.
SIG complies with multiple ISO/IEC standards, including ISO/IEC 27001 and 17025, and has co-developed ISO/IEC 5338, the new global standard for AI lifecycle management.
SIG was founded in 2000 and has offices in New York, Copenhagen, Brussels, and Frankfurt, and is headquartered in Amsterdam. Sigrid®, together with expert software engineering consultants, and over 25 years of industry-leading research, positions SIG as the foremost authority on software portfolio governance.
