Sebastian Strong Foundation: Automated Resource Research for Pediatric Cancer Families
A Navigator-built resource research system that automates the discovery, vetting, and delivery of family support resources — housing, food, transportation, financial aid — to families facing childhood cancer diagnoses.
The Challenge
The Sebastian Strong Foundation runs a Navigator Program that connects families facing pediatric cancer diagnoses with the resources they need to survive treatment: emergency housing near hospitals, meal support, transportation, financial assistance, and dozens of other categories. Every family's situation is different, and every Navigator on the team was spending hours per case manually searching IRS nonprofit records, foundation databases, and regional support directories to find programs that fit. The work was meaningful, but it was also painfully slow.
The foundation's case management system needed to receive vetted, structured resource data — but the data lived across half a dozen unstructured sources, none of which talked to each other, and a human had to approve every entry before it reached a family. The Foundation needed a system that could do the heavy research without taking the human approval step away, and without putting unvetted information in front of families at the most vulnerable moment of their lives.
Our Approach
We built the Navigator Resource Research System: an automated research pipeline that scrapes and structures resource data from multiple authoritative sources, hands every record to a Navigator for human approval inside a familiar Google Sheets interface, and then syncs approved records into the foundation's case management platform.
Implementation Phases
Mapped every authoritative resource source the Foundation already used — IRS nonprofit records, regional support directories, foundation portals — and built scrapers tuned for each. Designed for graceful failure: if a source changes or a record is incomplete, the row is flagged rather than dropped.
Defined a single resource schema that every source maps into — name, category, eligibility, geography, contact, intake process, last verified date — so Navigators can compare, filter, and approve resources side by side regardless of where they came from.
Built the human-in-the-loop interface where Navigators actually live: a Google Sheets approval queue with conditional formatting, status flags, and one-click approve/reject. No new tool to learn, no training overhead, no behavior change required.
Once a resource is approved, it syncs automatically into the foundation's case management system, where Navigators can attach it to a family's case record without retyping. Phase 2 also adds API connections to the next-generation case management platform the Foundation is migrating to.
System Architecture
IRS nonprofit records, regional support directories, foundation databases, manual Navigator submissions
- Multi-source scraping with per-source extraction logic
- Normalization into a single resource schema
- Confidence scoring and incomplete-row flagging
- Google Sheets approval queue for Navigator review
- One-click approval pushes records to the case management system
Vetted, structured resource records ready to be matched to family case records
Every record passes through a human Navigator before it ever reaches a family. AI does the research; people do the judgment.
Results & Impact
Navigators spend their time on families, not on database searches
Every resource passes Navigator approval before it reaches a family
One schema across every research source, synced into the case management system