The Unreasonable Effectiveness of LLMs at Creating Broken Systems
by Grant Udstrand | at Minnebar20
Large Language Models are remarkably effective at generating code, tests, and even entire system designs. They’re also remarkably effective at creating systems that no one fully understands.
As engineering velocity increases, so does intellectual entropy. Add in the variance introduced by AI-assisted development, and teams fall into volatility traps: systems that appear to improve rapidly while quietly increasing risk.
Drawing from real-world edge and hardware systems, we’ll explore why correctness is not a property of code, but of the system as a whole. We’ll also examine how AI can either amplify entropy or reduce it, depending on whether it’s used to generate decisions or enforce constraints.
Who should attend: Engineers, architects, and technical leaders building real systems. Beyond that, anyone else living in a world where things are moving faster, but getting harder to reason about, debug, or trust.
Grant Udstrand
Grant Udstrand is a Software Architect and Edge Computing Specialist with a passion for designing low-latency, high-performance systems that bridge the gap between the cloud and the physical world. With deep expertise in IoT, real-time data processing, and distributed systems, Grant has spent his career building and optimizing edge-first architectures that power industrial automation, smart devices, and mission-critical applications.
Currently working in embedded and edge computing, Grant helps organizations deploy scalable, secure, and intelligent solutions that process data where it matters most—at the edge. He is an advocate for containerized deployments, real-time telemetry, and AI-driven insights in Edge ecosystems.
Beyond his technical expertise, Grant is passionate about mentoring engineers, fostering innovation, and demystifying Edge Computing for developers, architects, and decision-makers. When he’s not architecting next-gen computing systems, you’ll find him exploring emerging technologies or engaging in hands-on projects with hardware and IoT devices.
💡 Talk to me about: Edge Computing, IoT, AI at the Edge, and real-time systems.
Are you interested in this session?
This will add your name to the list of interested participants. It will help us gauge interest for scheduling purposes.
Interested Participants
Similar Sessions
Help us find similar sessions by signing up for them!