Artifact 2 (Workshop 2.5): EveryParent AI Chat Application

Audience

This is for my instructor and classmates. It also serves as a portfolio piece for anyone who wants to understand how I approach building real AI systems that people can actually use.

Artifact

A real-world AI chat application built for EveryParent using Google Cloud Platform, including Firebase, Firestore, Cloud Functions, Gemini, and Vertex AI for retrieval-augmented generation (RAG).

Why it matters

I chose this artifact because it shows a very different side of me from Artifact 1. The first artifact was more reflective and focused on leadership, self-awareness, and how I think about AI adoption. This one is much more hands-on. It shows how I take AI from an idea and turn it into an actual system with architecture, constraints, tradeoffs, and users on the other side of it.

For me, that matters a lot. AI gets talked about in a very abstract way most of the time, but I am most interested in the part where it becomes real and useful.

Reflection

I chose EveryParent because it represents the kind of work I naturally gravitate toward: building things that are technically real, useful to people, and strong enough to live outside of a classroom prompt. The project centered on creating an AI chat experience that could give more grounded and context-aware responses instead of just sounding smart. I did not want a chatbot that simply generated polished text. I wanted a system that could retrieve relevant information, pass it through the right backend flow, and return something more reliable and practical.

That is where this artifact became meaningful to me. It was not just about calling a model API. It was about thinking through the whole system. I used Firebase and Firestore for the application and data layer, Cloud Functions for backend orchestration, Gemini for generation, and Vertex AI components for retrieval-based behavior. What I like about this stack is that it forced me to think in a more complete way: not just “what should the model say,” but “where does the answer come from, how is context retrieved, how does the application behave, and how do you make the whole thing hold together?”

This project also reflects how I tend to learn. I learn best by building. Once I have to make something work end-to-end, I start seeing the real challenges very quickly. In this case, those challenges included grounding responses well, structuring backend logic cleanly, making sure data flowed properly, and thinking more carefully about trust. A chatbot can look impressive very easily. Making one that should actually be trusted is a different level of responsibility.

That is also why this artifact fits well as a second portfolio piece. Artifact 1 showed how I reflect on leadership and growth. This artifact shows execution. It shows that my interest in AI is not only conceptual or managerial. I like getting into the system itself, understanding the moving parts, and making the pieces work together in a way that feels coherent and useful.

At the same time, this project reminded me that building AI systems is not just a technical exercise. It also raises questions about evaluation, safety, maintainability, and how much confidence a user should place in the system. I am comfortable building quickly, but one of the things I am still learning is how to formalize more of those guardrails in a repeatable way. That is part of why I wanted to include this project in my portfolio. It shows both what I can already do and what I am actively growing into.

Application View

You can interact with the EveryParent AI Chat Application directly below, or open it in a larger window.