Triad
In development / testing
AI-assisted referral decision support designed around source provenance, feedback and human verification.
Design example only. No deployed accuracy or client outcome claim.
Investigate the breakdown
Referrals get stuck. Knowledge walks out the door.
Nobody can see where the handoff failed. Investigate the breakdown with a scoped assessment, testable hypotheses and defined deliverables.
Recognition
Symptoms worth tracing
- A referral is sent but its next owner is unclear.
- Critical resource knowledge lives with a single experienced staff member.
Root-cause hypotheses to test
Routing, source freshness, eligibility rules or closure responsibility may be missing. A new directory alone may not fix the handoff.
Investigation & deliverable
Follow a referral from initial need to acceptance, rejection or closure; identify source and escalation gaps.
Example intervention
Define routing and closure responsibility; test decision support with human verification.
What you receive
Referral routing map, source-verification rules and an escalation protocol.
Related paid service: Decision-Support Prototype / Access MappingDocumented examples
Inspect the related work
Lantern
Publicly functioning project
No-account service navigation, weather safety and English / Spanish access.
Illustrative scenario using the real project. Listings do not establish service availability.
Before commissioning work
Questions to settle
Does this establish the cause?
No. These are hypotheses. The review traces actual records, handoffs and constraints before recommending changes.
What should we bring?
A description of the workflow, the decisions it supports, known constraints and redacted examples. Do not send confidential client or member information through the intake.
What happens after the assessment?
We agree a bounded scope and deliverables first. A prototype is not a clinical decision-maker or a guarantee of provider availability. There is no guaranteed outcome or pre-set turnaround.