BEACON

BEACON Logo
BEACON Summary Map

Inspiration

Millions of people face housing instability, food insecurity, unemployment, mental health challenges, and other hardships every year. Yet one of the biggest barriers is not the absence of support—it is the difficulty of understanding and navigating it.

A parent receiving an eviction notice may not know which deadlines matter, which resources apply to their situation, or what steps should be taken first. Someone struggling to afford food may face dozens of websites, forms, and eligibility requirements while already under significant stress.

Our team was inspired by a simple observation: people in crisis usually start with a story, not a service category.

Existing resource directories and search tools often assume users already know what they need. In reality, people describe situations:

"I lost my job and have two kids."

"I received this notice and don't understand it."

"I'm scared I won't be able to pay rent next month."

We built BEACON to bridge the gap between confusion and action by helping users understand their situation, identify what matters most, and connect with support they can realistically access.

What It Does

BEACON is an AI-powered support navigation platform that transforms stressful situations into clear explanations, prioritized action plans, and trusted local resources.

Users can either describe their situation in their own words or upload a document such as:

Housing notices
Benefits paperwork
School communications
Program eligibility letters
Assistance forms

BEACON then analyzes the information and generates:

Plain-Language Explanations

Complex documents and bureaucratic language are translated into simple, understandable summaries that explain:

What the document means
Why it matters
Important deadlines
Potential consequences of inaction

Situation Analysis

BEACON identifies important factors affecting the user's situation, including:

Housing instability
Food insecurity
Employment challenges
Medical concerns
Household dependents
Transportation limitations
Digital access barriers

Risk Prioritization

The system categorizes situations into four levels:

Level 0: Informational Support
Level 1: Emerging Need
Level 2: Significant Risk
Level 3: Immediate Intervention Recommended

This helps users understand urgency without making legal, medical, or safety decisions on their behalf.

Recovery Roadmap

Rather than presenting a long list of resources, BEACON generates a structured action plan divided into:

Immediate Actions (Days 1–3)
Short-Term Actions (Days 4–15)
Longer-Term Stabilization (Days 16–30)

This helps users focus on the most important next steps first.

Local Resource Navigation

Using location-aware resource matching, BEACON recommends relevant community services and displays them on an interactive map.

Recommendations are adjusted based on real-world barriers such as:

Transportation access
Presence of children or dependents
Internet connectivity limitations
Physical accessibility concerns

The goal is not simply to identify available support, but to identify support that is realistically reachable.

Why AI?

A traditional directory or keyword search cannot effectively solve this problem.

Most support systems require users to already know:

What type of assistance they need
Which program category applies
Which organization to contact

People in crisis often do not know any of these things.

Instead, they provide narratives:

"I got laid off last month and can't afford groceries."

"My landlord posted this notice on my door."

"My child was suspended and I don't understand this letter."

AI enables BEACON to:

Understand free-form narratives.
Interpret complex documents.
Extract relevant contextual factors.
Prioritize urgent needs.
Generate personalized action plans.
Match users to appropriate support resources.

Without AI, users would need to manually navigate multiple websites, forms, and service directories while already under stress.

AI allows BEACON to transform raw situations into understandable, actionable guidance.

How We Built It

BEACON was developed using a full-stack architecture designed for accessibility, reliability, and transparency.

Frontend

React
Vite
TypeScript
Tailwind CSS
Framer Motion

The interface guides users through a structured intake process and presents recommendations in a clear, low-stress format.

AI Processing

We integrated the Google Gemini API to:

Analyze uploaded documents
Process user narratives
Generate plain-language explanations
Extract key situational indicators
Produce structured recommendations

The AI outputs standardized JSON objects that drive downstream recommendation and planning workflows.

Resource Mapping

We integrated:

Leaflet.js
OpenStreetMap

to visualize nearby support resources and improve accessibility.

Offline Simulation Mode

We developed a deterministic rules-based simulation engine that mirrors the platform's decision structure.

This allows:

Demonstrations without API calls
Offline testing
Consistent validation
Reduced operating costs

The simulation mode produces identical data structures to the AI-powered workflow, ensuring reliability during demonstrations and development.

Report Generation

We implemented automated PDF generation using jsPDF so users and caseworkers can export recovery plans and summaries for offline use.

Challenges We Faced

The biggest challenge was balancing helpful AI assistance with responsible system design.

In high-stakes situations, inaccurate information can cause real harm.

We addressed this challenge by:

Providing source transparency
Using confidence-based recommendations
Maintaining human oversight boundaries
Building an offline fallback system
Separating informational guidance from decision-making

Another challenge involved designing recommendations that account for real-world barriers.

A theoretically available resource may not be useful if a user lacks transportation, internet access, childcare, or physical accessibility accommodations.

We spent significant time designing BEACON to consider these constraints so recommendations remain practical rather than purely informational.

Accomplishments We're Proud Of

We are most proud of BEACON's ability to move beyond simple resource lookup.

Instead of showing users a list of programs, BEACON helps answer:

What is happening?
What matters most?
What should I do first?
Which resources are realistic for me?

We are also proud of the platform's barrier-aware recommendation system, which adapts guidance based on household circumstances and accessibility constraints.

Finally, we are proud of the integrated dashboard experience that combines document understanding, risk assessment, action planning, and local resource navigation into a single workflow.

What We Learned

Throughout this project, we learned that one of the most valuable roles AI can play in social support systems is reducing confusion.

People experiencing hardship are often overwhelmed by information, paperwork, deadlines, and uncertainty.

The goal is not to replace human judgment or professional support.

Instead, AI can help people better understand their situation before they interact with social workers, nonprofit organizations, legal aid providers, educators, healthcare professionals, or community support services.

We also learned that reliability is an ethical requirement.

Systems designed to support vulnerable populations must continue functioning even when external services fail, which reinforced the importance of our fallback architecture and transparent recommendation process.

Responsible AI

Realistic Risk

A user could receive incomplete, outdated, or incorrect resource recommendations and rely on them during an urgent situation.

Mitigation

BEACON prioritizes verified resources, displays confidence indicators, provides source transparency, and continuously presents emergency support options when relevant.

The system is designed to provide guidance rather than definitive answers.

Human-in-the-Loop

BEACON does not make legal, medical, housing, or safety decisions.

If the system detects indicators of immediate danger, self-harm, violence, or other severe risks, it recommends escalation to qualified professionals and emergency services rather than attempting to resolve the situation autonomously.

Humans remain responsible for all final decisions and interventions.