How should/shouldn't I use agentic coding to develop a portfolio?

Why Read This from an HCI Perspective

This article addresses a key question in HCI/UX practice: a portfolio is not just a collection of outputs, but a way to demonstrate what skills you have and how you prove them. In particular, the use of AI tools such as agentic coding to increase production speed and expressiveness creates a gray area for practitioners—whether it counts as efficiency or as a shortcut. Through this piece, HCI researchers and designers can examine how tool use affects perceptions of competence and how portfolio evaluation criteria may need to be reshaped.

CIT's Commentary

From a CIT perspective, this discussion is best read as a shift from ‘making’ to ‘showing.’ Agentic coding is not merely a tool for automating development; it can also be a means for individuals to quickly externalize their problem-solving process and design capabilities. However, in HCI, what matters is not the flashiness of the final artifact, but the usage context, the rationale behind decisions, and the quality of interaction design. So while strategies that use AI to speed up portfolio production are certainly valid, they do not, by themselves, prove your competence. In fact, more clearly structured evidence—such as case studies, research insights, and your role in collaboration—better communicates HCI identity. Ultimately, the key is not whether you used AI, but what you can explain more convincingly through AI.

Questions to Consider While Reading

• Q.In a portfolio that uses agentic coding, what components most convincingly demonstrate HCI competence?
• Q.When you use AI to increase production speed, what contextual information should you provide to prevent evaluators from reading it as a ‘shortcut’?
• Q.For an HCI aspirant with a software engineering background, how should they balance technical skills and user research skills in their portfolio?