Granted

Inspiration

Our project, Granted, was inspired by the critical challenges faced by social care organizations like the Tsao Foundation. They struggle to track and match their specific program needs with over 200 constantly changing funding opportunities from agencies like AIC and NCSS. The core pain points were:

Information Overload: Small organizations lack the resources to continuously monitor all funding sources, leading to missed deadlines.

Timing Mismatch: They often search for funding after the need arises, not proactively before.

Lack of Context: Existing solutions (simple chatbots) only provide surface-level information and cannot perform the deep, contextual matching required to align a project's goals with the funder's true intent and Key Performance Indicators (KPIs).

What it does

Granted is an intelligent, AI-powered discovery and matching platform for social care grants. It goes beyond simple search to act as a knowledgeable consultant for non-profit funding.

1. Intelligent Tracking System Scans and monitors funding landscape continuously Alerts proactively when relevant grants appear

2. Context-Aware Matching Specific service context Funder's intent behind each grant Nuanced requirements buried in grant documentation

3. Three Key Matching Criteria No. Criteria What to Match 1 Eligibility "Type of care service provider (elderly care, home care, etc.) 2 Quantum/Scope "Project size, duration (tactical vs. strategic programs) 3 KPIs "What outcomes the funder wants vs. what they can deliver

How we built it

We built Granted using a modern, data-centric approach, focusing heavily on the semantic understanding capabilities of Gemini to solve the contextual matching problem.

Backend: Django + PostgreSQL (Cloud SQL), containerized and served via Cloud Run.

Infra: Docker, Artifact Registry, gunicorn, Whitenoise for static assets.

Data: Playwright, BeautifulSoup, and lxml for scraping; Cloud SQL connector for secure DB access.

Ops: Cloud Run Jobs for migrations; environment-based configuration with CSRF and HTTPS-ready settings.

Challenges we ran into

Defining Semantic Match: The biggest challenge was moving past basic keyword search. We had to train the AI to correctly interpret the nuances of grant documents, such as differentiating between a generic "IT project" fund and one specifically encouraging "adoption of AI & data."

Generating Trustworthy AI Output: A high matching score alone is insufficient. We introduced the "Why It May Not Match" section to force a balanced gap analysis, ensuring the AI output is objective, actionable, and trustworthy, especially when dealing with high-stakes funding decisions.

Data Integration: Conceptualizing the flow for continuously monitoring and normalizing unstructured data from various "SG Portals" into a unified internal database of "Grant" entities.

Accomplishments that we're proud of

Successful Contextual Matching Logic: We achieved the "Magic" the client wanted—the ability to tell the system a program vision (e.g., "expand dementia care capacity") and receive grants perfectly aligned with both logistical criteria and strategic KPIs. Trustworthy AI Feature: Implementing the "Why It May Not Match" section, which adds significant value by acting as a final layer of due diligence for the user. Comprehensive MVP Design: We delivered a high-fidelity UI/UX design, a full feature rundown, and a clear entity-relationship model, forming a solid blueprint for the entire platform.

What we learned

We learned that in the social service sector, the true value of AI lies not just in information retrieval, but in contextual interpretation and strategic consulting. A simple chatbot falls short because it cannot capture the intricate alignment required for grant applications. This reinforced the need for an application built on semantic understanding rather than just keyword filtering.

What's next for Granted

Recommended Proposal Generation: Implement the feature where clicking "Recommended proposal" provides a template proposal pre-filled and tailored to the specific grant, to reduce the writing time for applicants. Advanced Analytics: Fully integrate the planned analytics dashboard features to provide strategic insights to the organization. Continuous Improvement on Matching Rules: Continuously refine matching algorithms based on user feedback and application outcomes

Built With

• beautifulsoup4
• cloudrun
• csrf/https
• django
• docker
• gemini
• gunicorn
• lxml
• lxml.-auth/config:-env-based-settings
• playwright
• postgresql
• python
• whitenoise.-postgresql-on-cloud-sql-(via-cloud-sql-python-connector).-containerized-with-docker;-served-on-cloud-run;-images-in-artifact-registry.-frontend-templates-+-static-assets-(html/css/js)-within-django.-scraping/data-tools:-playwright