$ day-1 --pm · demo 2
Coding AI Interfaces with Claude
Move from a rule-based system to an AI-powered one. Instead of fixed logic, your application will interpret what the user types and generate a personalized response—built and refined entirely by prompt.
$ demo --rebuild [+][-]
Start from the recommender concept from the previous demo—but this time, build it as an AI app. Begin your prompt by saying you want to build “an AI app,” which signals Claude to use its AI-powered Artifact features rather than fixed, rule-based logic.
$ demo --iterate [+][-]
Refine through at least five more prompts to make the app your own and distinct from any sample. Don’t just polish—you might try something totally different in approach or interface.
Ask for new designs and interface concepts, and describe the features you want as concretely as possible. Specificity is what separates a generic result from one that fits your idea.
$ demo --publish [+][-]
Publish the finished application using Claude’s Artifact sharing, and open the live link to confirm it works for someone other than you.
$ demo --critique [+][-]
Share your application link and reflect:
- How do AI-powered apps differ from traditional rule-based programming? What can the AI version do that fixed logic cannot—and what does it lose?
- What assumptions does the AI make about user preferences?
- How do training-data limitations shape the recommendations the app produces?
- What are the implications of embedding AI reasoning directly in a user-facing application?
$ source — adapted from Week Ten: Coding AI Interfaces, Humanities & AI. $ cd .. — back to the course packet