This is my final post in a series about integrating AI into my .NET Web Development course (CSCD 379) at Eastern Washington University. (Earlier posts: If AI Writes the Code, What Should We Teach?, A Powerful Hope for the Future, Software Engineering Fundamentals Matter More Than Ever, and The Students Have Spoken)
The quarter is over. The final projects are deployed. Here’s what happened — and what I learned.
The Assignment
Students had roughly 2.5 weeks to build and present a full-stack web application. The requirements were ambitious: a vector database, AI/LLM features integrated into the app itself, a traditional database, full CI/CD pipelines, and a polished elevator pitch with technical depth. They needed to demonstrate everything they’d learned, not just using AI to build, but building AI into their applications.
What Surprised Me
I’ve run similar final projects for years (minus the AI). This time, a few things stood out:
🔹 Every student chose to work alone. AI makes collaboration harder, not easier. When one person with a good prompt can reshape the entire codebase in minutes, coordinating with teammates becomes a different kind of challenge.
🔹 Portfolio energy was real. I encouraged students to build in their own GitHub repos so their work was clearly theirs. The excitement about having a deployable, portfolio-ready project was palpable. Several students told me this was the most tangible thing they’d built in a university course.
🔹 The presentations were outstanding. Confident, funny, technically detailed. This class was more interactive and engaged than any I’ve taught in recent years.
🔹 Ambition was high. Students swung for the fences. Some features were more polished than others, but the apps looked good and generally worked. Most students spent 20–50 hours on their projects. Many spent extra time because it was genuinely “fun.” They were building things they cared about.
🔹 Tooling access mattered. Students who invested in GitHub Copilot Pro had a noticeably smoother experience. Others were copy-pasting between browser-based AI tools and their IDE. This is a real equity issue that needs addressing.
The Harder Question
Someone asked me: “Did they learn as much about web development as students in previous years?”
Honestly? No. And at first, that bothered me.
But here’s what they did learn: AI prompting for software development. Architecture-first thinking. Requirements engineering. Testing as a verification discipline, not busywork. CI/CD as a real deployment practice. And the confidence to build and ship a full-stack application — our survey’s highest-rated item at 4.875/5.
The web development knowledge gap was real, but it was offset by skills that feel more relevant to where the industry is heading. Students left this class feeling equipped and hopeful about their careers. In a time of enormous uncertainty, that matters.
AI becomes what you bring to it.
What’s Next
The next iteration of this course will front-load prompt engineering as a first-class skill, add design scaffolding to prevent “vibe coding” without killing creativity, and be more explicit about where AI falls short (Azure deployment, complex architecture, tests that validate requirements instead of just passing). The creative freedom, industry perspective, and real deployments stay.
Is this approach better or worse than what came before? At least it’s different. And in a key way, it feels more relevant.
Check out what the students built:
🌐 Journaling Site — https://journalsite.azurewebsites.net/
🌐 Warranty Tracking — https://agreeable-rock-06b51d41e.1.azurestaticapps.net
🌐 Party Planning — https://gentle-glacier-0780e230f.6.azurestaticapps.net/
🌐 Civil Pulse — https://green-flower-0c58be30f.2.azurestaticapps.net
🌐 Next Quest — https://icy-plant-0a2895310.6.azurestaticapps.net/
🌐 Movie Tree — https://white-desert-08300781e.6.azurestaticapps.net
🌐 Canon Guard — https://blue-dune-0adbddf0f.2.azurestaticapps.net
🌐 Backlogr (Video Game Tracker) — https://victorious-grass-0bb8bf10f.2.azurestaticapps.net/
🌐 Seasoned, a recipe site — Self-hosted on a container cluster she built at her house with automated local CI/CD pipelines. Yes, really. Very cool.
Course repo: https://github.com/IntelliTect-Samples/EWU-CSCD379-2026-Winter
Designing AI systems you can rely on
If you are exploring how to use large language models in real software, IntelliTect helps you turn clear intent into reliable, production-ready systems.