#332 How to Build AI Your Users Can Trust with David Colwell, VP of AI & ML at Tricentis

57 snips

Nov 17, 2025

David Colwell, VP of AI & ML at Tricentis, brings over 15 years of experience in AI and software testing. He dives into the critical relationship between data governance and AI quality, emphasizing the risks of unregulated AI that can lead to serious errors in legal contexts. David discusses the importance of meaningful metrics over mere quantity in AI outputs and highlights innovative strategies like using critic agents for reviewing AI content. He also stresses the evolution of data governance and the need for a human touch in ensuring AI systems are both accurate and compliant.

Ask episode

AI Snips

Chapters

Transcript

Episode notes

INSIGHT

Volume ≠ Value With AI

AI increases output volume but can make organizations mistake quantity for real value.
Measure meaningful outcomes like velocity and impact rather than raw pages or generated content.

ANECDOTE

Hallucinated Legal Brief Sank A Lawyer

David shares a legal disaster where a lawyer submitted a ChatGPT-generated brief full of fabricated citations and was publicly sanctioned.
The lawyer defended themselves by claiming they treated the model like a search engine and trusted the output without verification.

ADVICE

Validate AI Outputs With Tests

Put validation tests and benchmarks around AI-generated outputs so you can detect regressions quickly.
Feed test results back to the AI to enable self-correction and tighter integration into the workflow.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

The relationship between data governance and AI quality is more critical than ever. As organizations rush to implement AI solutions, many are discovering that without proper data hygiene and testing protocols, they're building on shaky foundations. How do you ensure your AI systems are making decisions based on accurate, appropriate information? What benchmarking strategies can help you measure real improvement rather than just increased output? With AI now touching everything from code generation to legal documents, the consequences of poor quality control extend far beyond simple errors—they can damage reputation, violate regulations, or even put licenses at risk.

David Colwell is the Vice President of Artificial Intelligence and Machine Learning at Tricentis, a global leader in continuous testing and quality engineering. He founded the company’s AI division in 2018 with a mission to make quality assurance more effective and engaging through applied AI innovation. With over 15 years of experience in AI, software testing, and automation, David has played a key role in shaping Tricentis’ intelligent testing strategy. His team developed Vision AI, a patented computer vision–based automation capability within Tosca, and continues to pioneer work in large language model agents and AI-driven quality engineering. Before joining Tricentis, David led testing and innovation initiatives at DX Solutions and OnePath, building automation frameworks and leading teams to deliver scalable, AI-enabled testing solutions. Based in Sydney, he remains focused on advancing practical, trustworthy applications of AI in enterprise software development.

In the episode, Richie and David explore AI disasters in legal settings, the balance between AI productivity and quality, the evolving role of data scientists, and the importance of benchmarks and data governance in AI development, and much more.

Links Mentioned in the Show:

New to DataCamp?

Learn on the go using the DataCamp mobile app
Empower your business with world-class data and AI skills with DataCamp for business