Tricentis on Why AI-Generated Code Needs Agentic Testing
Techstrong TV Interviews
•
17m
Coding agents now write much of today's software, but they can't be trusted to test their own work. Alan Shimel talks with David Colwell, VP of AI and machine learning at Tricentis, about why AI-generated code needs agentic testing.
Building AI into Tricentis since 2018
Colwell spent years in QA before joining Tricentis, where he started the company's AI department in 2018. His team built computer vision and OCR models first. Later, language models turned into assistants, then copilots and now agents. The goal stays the same: help testers keep up as more work flows their way.
Why coding agents can't test themselves
Colwell says the industry is speed running 40 years of software history in about two. A model bakes its assumptions into both its code and its tests, so those tests never challenge them. However, a separate testing agent works from the user story and business context instead. As a result, teams catch wrong assumptions early, deploy with fewer failures and spend fewer tokens.
Solving the last mile with agentic testing
Review and validation become the next bottlenecks once code arrives at AI scale. Without agentic testing, teams see only 10% to 15% gains from AI coding. With validation built into the pipeline, including clear evidence that code meets the requirement, productivity can rise many times over.
Explore more DevOps and testing videos and the latest Techstrong TV interviews.
For more information please visit the company website
Up Next in Techstrong TV Interviews
-
JoyStream.ai Builds an Operating Syst...
Most AI agents still live on someone's laptop, driven by prompts that can change at any time. Alan Shimel talks with Harpreet Singh, CEO of JoyStream.ai, and Vivek Pandey, its co-founder and CTO, about building an operating system for AI agents.
From Sun Microsystems to JoyStream
Singh an...
-
Sauce Labs Brings App Testing to the ...
Apple's foldable iPhone Duo changes how apps look and behave, and most apps were not built for it. Alan Shimel talks with Anoop Tripathi, CTO of Sauce Labs, about the company's new support for iPhone Duo app testing.
From firmware engineer to Sauce Labs CTO
Tripathi has spent more than 32...
-
Novee Brings AI Pentesting to the Ext...
Attackers now use AI to scan and probe around the clock, at a scale defenders have never faced. Alan Shimel talks with Netta Rager Dan, vice president of product at Novee, about the company's AI pentesting platform and its new External Surface Testing capability.
From weather forecasting to ...