Inspiration

Healthcare decisions are high-stakes, yet finding the right care remains surprisingly difficult.

Patients often know they need help, but not where to go, how much it will cost, whether a facility can actually provide the required treatment, or how to plan the journey. Traditional healthcare directories provide lists of hospitals but rarely explain why one option is better than another.

We built Prism, an AI-powered Health Concierge Copilot that helps patients, caregivers, and care coordinators find suitable healthcare facilities, understand the evidence behind recommendations, and plan their healthcare journey with confidence.

What it does

Prism helps users move from a healthcare need to an actionable care plan.

A user can ask:

"Find dialysis near Jaipur."
"Help me locate the nearest trauma center."
"Which hospital fits my budget and travel preferences?"

Prism then:

Understands the healthcare need
Asks clarifying questions when needed
Finds and compares healthcare facilities
Evaluates supporting evidence
Explains recommendations and uncertainty
Supports both patient and doctor/coordinator workflows
Helps users plan their healthcare journey

Every recommendation includes:

Distance and travel time
Care-match explanation
Supporting and missing evidence
Pricing signals (when available)
Confidence score
Shortlist functionality

How we built it

Prism was built using:

Databricks Apps for application hosting and deployment
OpenAI GPT-5 for conversational AI and recommendation workflows
Python for backend logic and agent orchestration
SQL for querying and transforming healthcare data
Delta Lake for storing and managing structured healthcare datasets
Git and GitHub for version control and collaboration

The application combines healthcare facility data, AI-powered reasoning, and evidence-based ranking to help users discover suitable healthcare facilities and understand the reasoning behind recommendations.

The system uses specialized AI agents for:

Healthcare facility discovery
Evidence verification
Travel planning
Facility ranking and recommendation

Users interact through:

Data & Insights dashboards
AI Concierge Chat (text and voice)
Recommendation workspace

Challenges we ran into

Healthcare requires transparency and trust.

Our biggest challenge was ensuring the system could provide useful recommendations without making unsupported assumptions. We focused on surfacing uncertainty, identifying missing information, validating facility capabilities, and explaining why recommendations were made.

Accomplishments that we're proud of

Built an AI healthcare concierge instead of a simple hospital search tool.
Created evidence-based facility recommendations with transparent reasoning.
Combined healthcare discovery, travel planning, and care coordination into a single workflow.
Designed support for both patient and care coordinator use cases.
Leveraged Databricks Apps, Delta Lake, SQL, and OpenAI GPT-5 to build an evidence-based healthcare decision-support experience.

What we learned

We learned that healthcare navigation is fundamentally a decision-support problem.

Users trust recommendations more when they can see the evidence, understand the reasoning, and clearly identify what information is missing or uncertain.