Sonar CEO Tariq Shaukat draws an analogy between manufacturing quality systems and software development as AI agents increase the volume of generated codeCode generation uses AI or another automated system to create source code from instructions, examples, schemas, or higher-level specifications.. He argues that higher output does not automatically translate into reliable releases.
Shaukat discusses limits of human reviewCode review examines proposed source changes to assess their correctness, clarity and suitability before or around integration., difficulties navigating large codebases, errors that compound across long tasks and variability in model-based judgments. He presents these as reasons to build quality controls into a development workflow.
His proposed confidence framework combines clear intent, guidance for agents, independent verification and repeatable evidence. He describes using deterministic and model-based checksDeterministic software verification checks an output with a repeatable procedure that produces the same result for the same inputs and state. together, and presents examples from Sonar’s own products and customers.
Watch on YouTube
