Referral Coordination Agent
Tracks outbound referrals and drafts follow-ups so patients don't fall through the cracks.
Referrals are where patients most commonly fall through the cracks — a specialist's office never calls to schedule, a fax gets lost, and nobody notices until the patient shows up months later having never been seen. This agent keeps a structured log of every outbound referral (specialty, date sent, expected completion window) and, when a referral crosses its timeline without a completion note, drafts a follow-up: one to the specialist's office on scheduling status, one to the patient checking whether they've been seen. It works entirely from referral status and specialty category — never diagnosis — because that is all the tracking job requires. Staff reviews and sends every draft.
What this agent takes off your plate
- Every outbound referral tracked against an expected completion window automatically
- Overdue referrals surface with a drafted follow-up instead of being discovered by accident
- Patients get a gentle check-in instead of falling through the cracks for months
- A specialist's-office follow-up message ready to fax or call in under a minute
- A running referral completion rate you can review monthly
The stack
Real costs & ROI
- Claude or ChatGPT on a plan that includes a BAAEnterprise / API pricing — confirmAs of writing, neither vendor includes a BAA on Pro/Team/Business seats — a BAA is available on Enterprise plans or via an API-based build. Expect materially higher cost than the consumer tiers and get the BAA in writing before any real data is entered.
- Google Sheets or EHR-native referral tracker$0/momost EHRs already have a referral module; a shared sheet works fine if yours doesn't
- Zapier or Make (free or paid tier)$0–10/mofree tier usually covers a low-volume weekly overdue-check automation
The math
Count your own gap first: pull last quarter's outbound referrals and check how many have a specialist note back in the chart — most practices that do this find a meaningful share simply went nowhere because nobody followed up. If a practice sending 20 referrals a month closes even a third of its silent-referral gap, that means several more patients actually get seen by the specialist they were referred to, which matters clinically and reduces the liability exposure of a referral that silently went nowhere. On the time side, staff currently spend roughly 10–15 minutes per referral doing this tracking manually if they do it at all; automating the flagging and draft-follow-up step is meant to give most of that back. Log the hours for one month before and one month after and weigh the difference against the tooling cost plus what your BAA-covered AI plan actually costs.
Build steps
Build your referral timeline map
Different specialties have wildly different normal wait times — cardiology and dermatology are not the same clock. This step alone clarifies what "overdue" even means for your practice, useful even before automating anything.
Data you need
List the specialties you refer out to most often and, if you know it, your rough sense of typical scheduling lead time for each. No patient data needed — specialty categories only.
I run a [SPECIALTY] practice and want to build a referral timeline map for my most common outbound referrals. My most common referral specialties and rough typical wait times if known: [LIST — e.g. "Cardiology: 3–4 weeks", "Dermatology: 6–8 weeks", "Physical therapy: 1–2 weeks", "Orthopedics: 2–3 weeks"] My referral volume: about [NUMBER]/month Please build: 1. **An expected completion window per specialty** — a reasonable range for "referral sent → patient seen," based on typical wait times 2. **An "overdue" threshold per specialty** — how far past the expected window before it should trigger a follow-up (e.g. expected window + 50%) 3. **A priority tier** — which referral types are higher-stakes if delayed (e.g. cardiology, oncology) vs. lower-stakes (e.g. routine PT) 4. **A one-page reference table**: specialty, expected window, overdue threshold, priority tier This becomes the rule set for the tracking automation — keep it to specialty categories and timelines, no patient-specific clinical detail.
Confirm your BAA and set up the structural tracking log
Confirm your AI plan includes a signed BAA (Enterprise or API-based — not a Pro/Team/Business seat), and build a tracking log limited to referral status fields — never diagnosis or clinical reasoning for the referral.
Data you need
Loop in your compliance contact to confirm BAA coverage before entering real referral data. Your tracking log should include: patient initials/ID, specialty referred to, date sent, expected completion window, status (pending/scheduled/completed/no-show), last follow-up date. Do not include the clinical reason for referral in this log. Initials, dates, and IDs are still PHI under HIPAA — the minimum-necessary log reduces exposure, it does not remove the need for a BAA.
Agent Pack
Unlock the Medical Practice Agent Pack.
One payment unlocks all 3 blueprints in this pack — build steps, the complete system prompt, guardrails, a test checklist, and the maintenance routine — for $29.
one-time, unlocks all 3 blueprints in this pack on this browser. Secure checkout via Stripe.
More in this pack
More for Medical Practice
No-Show Prevention Agent
Reminder sequences and waitlist backfill drafts, built without touching PHI.
Patient Intake Agent
Structured pre-visit summaries built inside a BAA-covered tool — never a consumer chatbot.