The Phog Log

Inspiration

As a freshman at the University of Kansas, I quickly realized how hard it was to manage nutrition while relying on on-campus dining options. I wanted something that could help students like me track their intake, build meal plans based on what’s actually available on campus, and do it all in a personalized, accessible way. So, I built The Phog Log — a KU-specific nutrition assistant powered by AI.

What it does

The Phog Log helps KU students:

Create a profile with their physical info, dietary restrictions, and fitness goals
Generate custom meal plans based on available dining hall foods using Gemini AI
Log the foods they’ve eaten each day
Track calories, protein, carbs, and fat with visual progress stats
View past logs and edit their profile as they grow
Toggle dark mode and use the app comfortably on both desktop and mobile It’s a full nutrition assistant — tailored to Jayhawks.

How we built it

I built the backend in Python using Flask, and structured the app around reusable modules like users_info.py, food_logger.py, and build_prompt.py. For the AI integration, I used Google Gemini (via the google-generativeai API) to dynamically generate meal plans based on the user's profile, calorie needs, and what's actually being served in KU dining halls that day. The frontend is made with HTML, CSS, and Jinja templates, styled with Poppins, light/dark mode, flexbox/grid layout, and smooth transitions for a modern feel. It's fully responsive and optimized for use on both desktops and phones.

Challenges we ran into

API integration: Getting both OpenAI and Gemini working locally with API keys and request formatting took time, especially with rate limits and version mismatches.
KU dining data: Simulating real food availability and nutrition data required creating our own .json structure for realistic campus options.
Flask to Web: Moving from CLI to a full Flask web app introduced challenges with session management, routing, and persistent logging.
UI design: Making it visually pleasing in a limited time without frontend frameworks required creativity and trial-and-error.

Accomplishments that we're proud of

Full end-to-end experience: Profile creation, AI meal generation, logging, tracking, and review — all working in sync
Dynamic meal plans based on KU-specific food and user needs
Fully responsive UI with dark mode and animated stat cards
Cleanly modularized Python backend with reusable logic
Custom domain: thephoglog.us