Joining the Google x Kaggle Agentic AI Hackathon was my first real step into building a complete Agentic AI application from scratch. I was excited, but also unsure where to begin. Between work responsibilities and new concepts I had never fully explored before, this challenge felt like juggling multiple things at once.
Yet that is exactly what made the experience so meaningful.
I wanted to build something simple, relatable, and helpful. A daily question came to mind:
What should I eat for lunch?
From that small question, the idea grew into Healthy Lunch Planner AI, a project that taught me far more than I expected.
How the Agents Work Together
I kept the technical design simple enough for a beginner in Agentic AI, but structured enough to reflect real collaboration between agents.
Planner Agent interprets the user’s goal and sets the direction.
Researcher Agent gathers options from the meal dataset.
Coder Agent calculates calories and compares details using Python.
Critic Agent checks whether the final options truly meet the user’s constraints.
Orchestrator Agent ties everything together and guides the workflow from beginning to end.
Working on these agents helped me understand the beauty of Agentic AI. Each model has a purpose, but the real magic happens when they work together and review each other’s steps. It reminded me of teamwork within an organization: different strengths contributing to one outcome.
Deploying It on Google Cloud Run
I challenged myself to deploy the system, even though it felt intimidating as a first timer. Using Cloud Run, I turned the orchestrator and agents into a real application with an API endpoint.
Seeing the project run outside the notebook was a proud moment for me. It made the entire experience feel complete and grounded.
Deliverables That Made Me Grow
To complete the Hackathon, I had to produce three major components:
- A clear write up
Writing this helped me organize my thinking, explain the workflow, and reflect on the reasoning behind every step.
- A working notebook
https://www.kaggle.com/code/irwanprabowo/healthy-lunch-planner-ai
This taught me patience, debugging discipline, and the importance of breaking a system into clear parts.
- A meaningful dataset
Building and integrating the dataset gave the agents real context, making the app feel purposeful and practical.
These deliverables pushed me out of my comfort zone, but they also boosted my confidence as someone who leads teams while still enjoying being hands-on.
Key Takeaways From the Experience
This Hackathon was more than a technical challenge. It became a reminder of why continuous learning matters. Here are the lessons that stayed with me:
- Starting small still leads to real impact
A simple lunch question grew into a functional AI application because I allowed myself to begin with what I had.
- Collaboration is powerful, even between models
Agentic AI reinforced a leadership lesson I value deeply: clarity of roles leads to clarity of outcomes.
- Being a first timer is not a weakness
I faced uncertainty, errors, and time pressure, but I grew a lot because of them. Stepping into something new is how we stretch our abilities.
- Tools matter, but mindset matters more
Cloud Run, Gemini Flash, datasets, and orchestrators were important, but the biggest driver was curiosity and the willingness to try.
- Finishing is its own achievement
Completing the project felt genuinely rewarding. It reminded me that learning by doing is still one of the most powerful ways to grow.
Final Thoughts
Healthy Lunch Planner AI started as a small idea, built by someone joining his first Agentic AI Hackathon while balancing work and daily routines. Yet it turned into a meaningful experience filled with learning, challenges, and personal growth.
This journey showed me that innovation does not always start with complexity. Sometimes it begins with a simple question, a willingness to learn, and the courage to build something new.
If this story inspires even one person to explore Agentic AI or step into something unfamiliar, then the journey becomes even more worthwhile.
Top comments (0)