Beyond Prompting: Build reusable AI workflows for UX research.

"AI infrastructure for UX research" refers to the reusable systems and practices that make AI-supported research work in practice: context files and knowledge bases, research repositories and source indexes, reusable workflows and automations, review and approval steps, evidence and usage rules, and clear ownership for keeping the whole system reliable over time. Most researchers are prompting ad hoc, getting useful output but starting over each time. This course teaches you to build the layer underneath the prompting: the structure that makes AI reliable at scale, not just occasionally useful. Each week adds a piece to your AI Infrastructure Kit, and in the final session you assemble those pieces into a function-level AI strategy you can take back to your own organization.

The AI for UX Research Course teaches how to use AI to accelerate coding, synthesis, and insight generation on active research projects. It's the right starting point for researchers who want to get faster and deeper on project-level analysis. This course picks up at the layer above: once you know how to use AI productively, how do you build a practice around it? Context systems, workflow libraries, research memory, and function-level strategy. The two courses are designed to work together, not compete.

No. The course covers version control and Git at a conceptual level, enough to know when those tools are useful for a research practice, but you won't write code or commit files. The goal is judgment about when these tools belong in a research workflow, not technical fluency with them. No prior programming experience is needed.

This course is designed for researchers who can already run studies independently. If you're new to UX research, start with research fundamentals first. If you're a mid-career to senior researcher, research manager, or ResearchOps practitioner who is already using AI but hasn't systematized it, this is built for you. It assumes research competence and builds AI infrastructure on top of it.

This is one of the course's central questions, and there's no single formula. As a general principle: AI handles volume, drafting, and pattern-matching well; researcher judgment is essential for framing, interpretation, reading nuance, and making calls that affect real people. The course is designed to help you build decision frameworks that answer this question for your specific context, not hand you a rule that doesn't survive contact with a real research project. Module 1 asks what an AI-supported output has earned the right to mean and what review it needs before it travels; the remaining modules teach you where to put it in a workflow and where to keep it out.

Both. Individual researchers will leave with context systems, workflow libraries, and a research repository they can use immediately. Research managers and ResearchOps practitioners will also be equipped to build team norms, evaluate tools for their function, and write the kind of AI strategy a senior leader can act on. Module 7 is specifically designed for the function-level challenge: what does a sustainable AI strategy for a research team actually include?

The course is tool-agnostic by design. Concepts like context files, research memory, and workflow libraries apply regardless of whether you use Claude, ChatGPT, Gemini, or another AI system. Where specific tools appear in exercises, they're used to illustrate a principle, not to lock you into a particular vendor. You'll come out with infrastructure you can apply in whatever AI environment your organization uses.

By the end of the 7 weeks, you'll be able to decide which research tasks are good candidates for AI, build reusable context and research-memory systems, create repeatable AI-supported workflows for planning, reporting, and reuse, evaluate synthetic research material and AI tools against real criteria, and turn your work into a practical research-function strategy. Each session adds a piece to your AI Infrastructure Kit, and the final session assembles it into a function-level AI strategy for your own organization.

Not the premise of this course, and not what the evidence actually suggests. AI does not remove the need for research judgment; it raises the premium on it. Organizations that adopt AI in research still need someone who can frame the right question, interpret behavior in context, separate signal from noise, and help the organization make better decisions. What AI changes is the ceiling on what a researcher can produce and the baseline speed expectation. This course is about positioning yourself on the right side of that shift, as the person who builds the infrastructure, not the one who gets outpaced by it.

The live sessions are the core of the course. To get the most out of it, students can complete the optional homework between sessions, which may add 1–2 hours on top of the session time. The exercises are designed to be applied directly to work you're already doing, so you're building real artifacts for your own practice rather than toy examples.

Yes. Every session is recorded and posted the same day. The live sessions are where most of the building and discussion happens, so attending live is worth it, but you won't fall behind if you need to catch a recording.

Prompt engineering courses teach you to write better instructions for an AI in the moment. This course teaches the layer above: how to build systems that carry your expertise across sessions, projects, and team members without requiring you to re-prompt everything from scratch each time. Free resources (blog posts, YouTube tutorials, LinkedIn posts) tend to cover individual tactics. This course builds the strategy that connects those tactics into a practice. The difference shows up months later: either you're still prompting each project from scratch, or you have infrastructure that compounds.

This is one of the most important concerns practitioners are raising right now, and it's exactly what the course is designed to address. CHI 2025 research on generative AI in UX research found that "people are cognitive misers: following up an LLM feels like the work is already done," which means hallucinations and thin synthesis pass through unchallenged. The course doesn't treat AI as a synthesis replacement. It treats AI as infrastructure for the parts of the workflow where it genuinely helps (drafting, structuring, surfacing gaps) and builds in the judgment practices that keep the researcher in charge of interpretation, evidence fidelity, and claim strength. Module 1 specifically covers checking claims against sources and catching unsupported inference and false confidence before an output reaches a stakeholder deck.

This is a real concern, and it deserves a direct answer. A UXR strategist surveyed by UXtweak in 2025 put it plainly: "They don't need us to sit around and prompt AI. They'll ask someone cheaper to do that." She's right: if AI strategy in your organization gets defined as "researchers now operate AI tools," that's a race to the bottom. This course is built on the opposite premise: the researchers who are hardest to replace are the ones who build the infrastructure layer, set quality standards, advise the organization on where AI does and doesn't belong in research, and preserve the interpretive judgment that makes findings credible and defensible. Prompting is a commodity. Building the systems, and exercising the judgment that makes those systems produce reliable research — that is not.