Food.ily

Food.ily is a Gemini-powered AI dining concierge that learns your taste, provide unique suggestions catered to each user , eliminating decision fatigue with accuracy that improves as you use it.

Comment

Inspiration

Living in metropolitan Kuala Lumpur, deciding what to eat is a daily struggle. I'd spend hours scrolling through Google and Pinterest, suffering from decision fatigue on what to eat. Whether it's organizing a team lunch for diverse colleagues or finding authentic local spots while traveling, the question "What should we eat?" is a universal hassle. That frustration gave me the inspiration for Food.ily.

What it does

Food.ily is a community based AI-powered food companion, designed to solve the universal dilemma of "What should we eat?". Unlike traditional directory apps that overwhelm users with static lists, Food.ily leverages Gemini 3.0's advanced reasoning and multimodal capabilities to act as a proactive Concierge and specialized Food Hunter. Instead of suggesting random restaurants, Food.ily is engineered to learn and understand each user's taste and preference, suggesting cuisine for user based on their habits and provide them with accurate anwsers to mitigate decision fatigue and time wasted. The application features three core modes: Food Hunter, which uses targeted grounding to find restaurants based on specific cravings or vague vibes; The Dining Concierge, an agentic planner that orchestrates complex group dining experiences by considering occasion, budget, and guests; and Food Gatcha, a gamified discovery engine for the indecisive. Besides that, Food.ily recreates the Review feature into a "Post", promoting community engagement, credit and trust of restaurants for user to select their best place to dine in.

How we built it

Built on a modern React/Vite stack, Food.ily features a sophisticated multi-model architecture. We utilize Gemini 2.5 Flash for high-speed, interactive chats and Gemini 3.0 Pro for deep analysis of weekly user habits.

  • Context-Aware Engine: We built a custom pipeline that feeds real-time context—Time (Lunch vs. Late Night), Location, and Taste Profile data—into every prompt, ensuring 100% personalized results.
  • Hallucination-Free: By strictly integrating Google Maps Grounding, every recommendation is verified against real-world data, ensuring users never walk to a closed restaurant.
  • Antigravity & Firebase: We used Google Antigravity to accelerate complex agentic workflows and Firebase for seamless real-time data sync across our community features.

Challenges we ran into

Balancing "Vibe" with "Accuracy". Prompting Gemini to understand abstract concepts like "romantic but affordable" while keeping data strict was a challenge. We overcame this by implementing Chain-of-Thought prompting and JSON schema enforcement to ensure the AI "thinks" like a foodie but outputs structured data for our UI.

Accomplishments that we're proud of

As an Architecture student with zero prior coding knowledge, I am incredibly proud to have built a fully functional, AI-powered application. We didn't just build a wrapper; we designed a solution that solves a real human pain point. Just as Brian Chesky said, the world needs more designers, we proved that design-thinking + AI can solve complex engineering problems.

What we learned

This hackathon taught us about the posibilities of AI and how it creates massive opportunities of human, developing solution and eases our daily life, just as Food.ily. It also teaches us sometimes all we need are obvious simple fixes in life, but its how we approach it and how can we create a creative solution that would appeal for everyone.

What's next for Food.ily

As the aims of Food.ily is to create a interactive and "gamified" based app, the next steps would be adding UI design as well as animations to enhance the app experience. Focusing on community as well, Food.ily would continue to innovate and create community-based functions such as future planning, or a "shake" feature to connect with friends to "plan" on group settings and "decide" , wheel-spinning to enhance community engagements. A MVP would soon be deployed for Food.ily and a waitlist to find our product-market-fit. We're hopeful of Food.ily and appreciates the Google Team for giving us this opportunity.

Built With

Share this project:

Updates