The agentic era in software has set off a chain reaction of novel, bespoke, and personalized technical systems. This is letting teams get more done, but navigating these new systems, evaluating whether they’re working, and integrating them with more familiar technologies can create chaos. If we’re not writing systems ourselves, or even reading them, how do we do good technical problem-solving?

Some teams are thriving, and others are flailing. The difference might come from which ones are building comprehension, transmitting new solutions, and forming reliable mental models even in systems they didn’t write or read.

Cat Hicks is an experimental psychologist, the author of The Psychology of Software Teams, and the former VP of Research at Pluralsight, where she built a research department from scratch to study how software teams learn and thrive. At Catharsis, she leads radical science projects built with developer communities to bring evidence to the questions engineering usually answers with anecdotes. Her open access, community-driven studies of thousands of developers point to critical psychological factors that separate teams that keep learning and creatively iterating through novel solutions in agentic work from teams that don’t.

In this talk, Cat will translate the science of expertise into practical strategies. We'll explore how metacognitive calibration (our “thinking about our thinking”) reshapes how teams create problem-solving scaffolding in complex technical systems. We’ll also examine how these practices help teams avoid agentic coding pitfalls like dampening collective creativity with “plausible first answer machines,” and how self-correcting expertise helps us puncture metacognitive distortions like the illusions of fluency and explanatory depth.

Along the way, we’ll challenge stereotypes about what drives technical breakthroughs, unpack viral claims about AI and cognition, and identify the features that help teams tell the productive, creativity-enhancing friction from the rote work they can leave behind.

Grounded in evidence across Cat’s in-the-field experiments, interviews, case studies, and empirical interventions, including a project motivating thousands of developers to adopt just-in-time learning practices in their everyday work, this talk will leave you with experiments to help you build your metacognitive toolkit, protect divergent thinking on your teams, and redefine what it means to do great technical problem-solving in the AI era.