1. Team: Localhoster

<aside> 👥

Team Members / Course :

• Radoslaw Aryananda / Artificial Intelligence and Society
• Michael Ryan Sutanto / Data Science and Artificial Intelligence
• Ignatius Jovan Liem / Data Science and Artificial Intelligence

School: Nanyang Technological University

</aside>

2. Idea Summary

<aside> 💡

Singapore faces a dire structural food security mismatch: 784,000 tonnes of food waste each year, and poor coordination means this surplus reaches only 10% of the 10% food-insecure residents. The visibility gap gives rise to economic losses of S$2 billion. Current manual models have not yet established the delivery of undignified aid or mobility issues.

This leads to waste of about two bowls of rice per person daily; Of it, only 18% is recycled, and 22% of insecure households received support, as existing information flows are weakly connected to bridge an effective supply-demand gap.

We achieve this by targeting a logistics response time of fewer than 5 minutes and a Jain Fairness Index of over 0.9, thereby enabling resources to be fairly allocated in a way that overcomes privacy obstacles and that supports the 78% of households that have been previously excluded.

</aside>

3. Proposed Solution

<aside> 🔧

Our product, OptiMeal OS, transforms Notion from a passive database into the active "Central Nervous System" of a nationwide logistics network. Using Notion AI and the Model Context Protocol (MCP), we elevate disintegrated logistical noise into an intelligible, self-governing operating system.

copy_2C7BFCF6-BD70-4370-92F4-2755324EF163 2.mov

A Bi-Directional Brain

Framed within the concept of a "Headless" architecture, it is Streamlit that bears the burden of physically executing interactions with Drivers and Donors. But it is solidly Notion 3.0 that brings in the strategic intelligence layer.

1. The Sensory Layer: MCP Ingestion

• Workflow: Upon a Donor submitting a manifest through the public-facing portal, such a system triggers an email to one NGO. At that point, a Custom Python MCP Server instantaneously injects the live data into the Notion "Command Center."
• Notion 3.0 Capability: This feature instantly automates the previous manual data entry process. The system itself listens in on real time and finetunes the absorbed GPS coordinates, perishability, and timestamped status from the external web application.

Figure 1: The Donor Portal (Input). Data submitted here is instantly synced to Notion via MCP agents.

2. Intelligence Layer: Notion AI & Fairness Engine

• Workflow: Data once in the grid gets analyzed. We employ Notion AI to keep watch over the Jain Fairness Index on the districts of Singapore, and most importantly, sorting food based on temperature and expiry date (persihability).
• Once it alerts on the fact that one community—let's say, Bedok—starts getting "oversupplied," while another one—say Jurong—gets to be underprovided, this triggers a "Fairness Alert."
• Notion 3.0 Capability: We use Notion AI's generative capability to auto-draft "Strategic Rerouting Orders." It reads into the system logs and recommends the best possible action to be taken by human operators.

Notion Page : https://fragrant-error-7d8.notion.site/OptiMeal-OS-2cd33f68bace8086816ae1b3b7a5882a

Figure 2: The Notion Command Center (The Brain). The graph tracks 'Community Saturation' in real-time.

3. The Action Layer: Automated Dispatch

• Workflow: Cycle closes once the NGO operator agrees with the AI's recommendation. From Notion, the decision is sent to the Driver App, updating the route on their behalf.
• Value: Visibility Gap compressed from days to milliseconds. Food waste is not recorded; it is actively intercepted.

Figure 3: Fleet Execution. The system calculates the nearest driver (Proximity Optimization) and dispatches them to the AI-verified location.

Figure 4: Sequence diagram

Github Repo : https://github.com/SLAWWWW/Notion-Ideathon---Optimeal/tree/main/OptimealHackathon

Why Notion?

A standard SQL database can store numbers but cannot do reasoning. By building on Notion, we gain:

1. Contextual Intelligence: Notion AI understands the nuance of "Urgent Perishable" vs. "Canned Goods" better than rigid code.
2. Human-in-the-Loop: While the agents run the logistics, the Notion interface provides a dignified, readable workspace for human volunteers overseeing the "Invisible Co-Founder." </aside>

4. Impact & Use Case

<aside> 📊

Qualitative Impact:

Nowadays, NGOs are burning out. They spend hours frantically trying to reach out to drivers and frantically updating spreadsheets. OptiMeal OS changes that. By letting Notion handle the complex logistics, these teams are freed up to do what they signed up for: help people.

For families receiving assistance, the regularity of this service restores dignity. They know that in such times, they can rely on regular support instead of stressing over "feast or famine" cycles brought about by poor distribution logistics.

Quantitative Success Metrics:

• Speed: We cut response times from 4–6 hours to under 5 minutes.
• Fairness: We track the Jain Fairness Index (>0.9). This way, a family in the West (Jurong) will have the same shot at a meal as a family in the East (Bedok), and the unequal distribution will break the cycle.
• Waste: Our goal is zero spoilage for captured items. The AI automatically prioritizes "High Risk" perishables; therefore, good food does not rot while waiting for a driver.

Helping the Community:

• For the Controllers (OptiMeal/Gov): All-in-one simple dashboard. Notion is a one-stop command center where a single oversight team can monitor food security nationally, impose fairness algorithms, and optimize logistics at a nationwide scale, all from a single dashboard.
• For the Donors: We make charity frictionless. Donors get a simple, one-click interface to clear their inventory. More importantly, we give them the peace of mind that their contribution will be rescued instantly and consumed while fresh rather than rotting in a warehouse.
• For the NGOs: We ensure they are never forgotten. Instead of competing for donations, NGOs receive a regulated, steady stream of supplies managed by our central fairness engine. This guarantees that a center in a remote area is just as well-stocked as one in the city without them needing to manage complex logistics.
• For the People (Beneficiaries): We ensure equitable reliability. By balancing the flow of donations across the island, we guarantee that a center in a remote area is just as well-stocked as one in the city. Families can rely on their local NGO having food on the shelves, regardless of their zip code. </aside>

5. Implementation Feasibility

<aside> 💻

Technology Stack

Primary Language: Python

• Framework: Streamlit
• Database / Backend: Notion API (acting as a lightweight backend CMS)
• Deployment: Streamlit Community Cloud (GitHub-connected deployment)

This project is implemented entirely in Python using Streamlit, enabling rapid prototyping of an interactive web application without a traditional frontend-backend split.

System Architecture

OptiMeal is designed as a single-page Streamlit application with two operational modes:

1. Public Portal (Individual / Donor Facing)
2. NGO Operations (Command Center Dashboard)

The app communicates with Notion databases via the official Notion API, allowing structured storage of donation records and logistics data without setting up a custom backend server.

Core Modules & Implementation

1. Public Donation & Public Listing Portal

Implemented under the "PUBLIC PORTAL" view in Streamlit.

How it works:

• Individuals submit surplus food through a Streamlit form
• Each submission creates a new entry in the Notion Donations database.
• Required metadata includes:
  • Item description
  • Pickup location (mapped to real GPS coordinates in Singapore)
  • Expiry date
  • Optional temperature data
• All public listings are initially marked as "Pending" safety status.

This public listing feature enables individual-to-individual food sharing, aligned with the Good Samaritan Food Donation Act, ensuring donors acting in good faith are protected and listings are transparently documented.

2. NGO Command Center

Implemented under the "NGO OPERATIONS" mode.

Features include:

• Live metrics dashboard (verified inventory, idle drivers, active missions)
• Marketplace view of verified donations
• AI driver brief display (stored in Notion fields)
• Manual dispatch system assigning idle drivers to pickup tasks

Dispatching a donation:

• Randomly selects an idle driver
• Updates driver status to "En Route"
• Archives the donation record to prevent duplication

3. Logistics & Fleet Control

A real-time fleet visualization is implemented using [st.map](<http://st.map>)().

Map includes:

• 🟢 Active drivers
• ⚫ Idle drivers
• 🔴 Pending pickup locations

Driver states are updated dynamically via Notion API calls, and operators can:

• Confirm arrivals
• Mark deliveries complete
• Reset drivers to idle

Deployment Feasibility

The application can be deployed via Streamlit Community Cloud, which:

• Connects directly to the GitHub repository
• Requires no containerization or server setup
• Supports environment variables for secure Notion API keys </aside>