TestGuild Automation Podcast Locust Performance Testing with AI and Observability with Lars Holmberg
Jan 13, 2026
Lars Holmberg, Founder and CTO of Locust, shares his expertise in performance testing and observability. He explains why Python is the preferred choice for AI-assisted load testing, highlighting Locust's flexibility and massive concurrency capabilities. They delve into the importance of observability for identifying performance bottlenecks and outline common mistakes teams make. Lars also discusses how integrating Locust with CI/CD and GitHub Actions streamlines the testing process, while offering insights into the future of testing with AsyncIO and GIL improvements.
AI Snips
Chapters
Transcript
Episode notes
Python Enables Flexible Test Modeling
- Locust's Python-first design makes complex, flexible performance scenarios easy to express.
- Lars Holmberg says Python's expressiveness reduces boilerplate compared to JavaScript and other tools.
From Battlefield Forums To Klarna
- Locust began as an internal tool used at Dice and later gained traction at Klarna where Lars contributed.
- Lars recounts taking over as core maintainer after using Locust in real performance testing roles.
Language Choice Enables Protocol Flexibility
- Locust's Python base makes it easy to integrate any Python SDK and test many protocols.
- Lars contrasts this with JMeter which often forces Java code and is less flexible for custom protocols.