Healthcare · Medical Practice Owners & Administrators

AI Agent for Patient Referral Coordination

Tracks referrals out and in, chases the ones that stall, and flags the patients who have fallen between two providers.

Start from this template
Edit it — the agent is built from this briefBuild this agent
How it works
1 Step
Track from the moment it is sent
2 Step
Check each stage
3 Step
Chase and flag
Every referral, with the receiving provider and the date it went out.

Overview

The patient who was referred and then disappeared.

An AI agent for patient referral coordination closes a loop that is open in almost every practice. A referral goes out, and then one of several things happens: the specialist accepts and books, the patient never makes the appointment, the referral is rejected and nobody notices, or the appointment happens and no letter comes back. Only the first is fine, and the practice usually cannot tell which occurred without somebody checking. Patients fall through this gap routinely and the practice finds out when they return months later, still unseen. The agent tracks each referral from the moment it is sent, chases at the points where they typically stall, and flags the ones where the loop has not closed. None of the tracking requires any medical judgment, which is what makes the gap so persistent: it is work that is unmistakably somebody's responsibility in principle and nobody's in practice. Built on Agentplace: the agent runs on its own page, so a visitor finishes the whole request in the conversation.


Capabilities

What the Referral Agent does

Tracks the loop and flags where it broke.

01

Tracks every referral sent and its current state

02

Confirms whether the receiving provider accepted it

03

Checks whether the patient actually booked

04

Chases outstanding referrals at the usual stall points

05

Flags referrals where no report has come back

06

Routes clinical questions to the referring clinician

Why you should use the Referral Agent

The referring clinician has done their part when the referral goes out, the receiving provider's responsibility starts when they accept it, and the space in between belongs to nobody. That is where patients are lost — not through anybody's error, but because there is no point in the process at which somebody is supposed to notice that a referral sent six weeks ago has produced nothing. Tracking the loop is administrative work with a clinical consequence, which is an unusual and important combination: the task requires no medical judgment at all, and failing to do it means patients go unseen. The flag that matters most is the quiet one — a referral accepted, an appointment made, and no report received — because that is the case where everybody assumes it went fine.

Before
The middle of a referral belongs to nobody
A rejected referral is not noticed for weeks
A patient never books and nobody checks
An appointment happens and no report comes back
The practice finds out when the patient returns unseen
After
Every referral is tracked from the moment it is sent
Rejections are noticed within days rather than weeks
Patients who have not booked are identified and contacted
Missing reports are chased rather than assumed fine
The loop closes, or somebody is told it has not
Process

How it works

Track it out, check each stage, flag what stalls.

Step 01

Track from the moment it is sent

Every referral, with the receiving provider and the date it went out.

Step 02

Check each stage

Accepted or rejected, booked or not booked, seen or not seen, report received or outstanding.

Step 03

Chase and flag

Chase the stages that stall, and flag anything where the loop has not closed to the referring clinician.


Example

Example workflow

Three referrals, three different failures.

Scenario: a practice ran a referral audit and found patients who had been referred more than four months earlier with no record of any outcome. The agent begins tracking. Of three referrals sent in one week, one is accepted and booked and needs nothing. The second is rejected by the receiving service eight days later for an administrative reason, which under the old process would have sat unnoticed in an inbox — the agent flags it to the referring clinician, who resubmits it correctly the same day. The third is accepted, but four weeks on the agent finds no record that the patient ever booked. It contacts the patient, who says they never received the appointment letter. The appointment is arranged. None of the three required clinical judgment from the agent, and two of the three would have failed silently.

Medical Practice Front Office AirtableGmailTwilio SMSSlack AI Agent flow

Audience

Who can benefit

Anybody who cannot say what happened to a referral.

✍️ Practice owners and partners

Patients lost in referrals are a safety issue, not an admin one.

💼 Practice managers

Nobody is responsible for the middle of the process.

🧠 Referring clinicians

You do not find out that a referral was rejected.

Clinical governance leads

Unclosed referral loops are an auditable risk.

🎯 Practices referring at volume

The queue is too large to track by memory.

📋 Practices receiving referrals

The same gaps exist on the inbound side.

Integrations

Where referrals are tracked and who is told.

Airtable

Holds every referral, its stage, dates and the receiving provider.

Gmail

Chases receiving providers and records their responses.

Twilio SMS

Contacts patients who have not booked their appointment.

Slack

Flags rejections and unclosed loops to the referring clinician.

Google Drive

Stores reports as they come back against the referral.

Google Sheets

Reports loop closure rate and where referrals stall.

Applications

Best use cases

The referral failures worth catching.

A referral rejected for an administrative reason
A patient who never received the appointment letter
A patient who received it and never booked
An appointment attended with no report returned
A referral sent months ago with no recorded outcome
A receiving provider who has not responded at all


FAQ

FAQ

Questions about closing the referral loop.

An AI agent for patient referral coordination tracks every referral from the moment it is sent, checks whether it was accepted, whether the patient booked and whether a report came back, chases the stages that stall, and flags unclosed loops to the referring clinician.

The quiet one. A referral accepted, an appointment apparently made, and no report received looks like success to everybody involved, and it is the case where a patient can go unseen for months.

None. Every step is administrative — sent, accepted, booked, seen, reported. That is precisely why it is worth automating and why it is so consistently left undone.

For booking status, yes, with the practice's agreement. A patient who did not receive a letter is the most common single cause of a stalled referral and a text resolves it.

The same tracking applies inbound and it is often worse, because the receiving side has less visibility of what the referrer expected. Both directions are worth closing.

Unclosed referral loops are auditable in most systems and are a recognized patient safety concern. That framing usually matters more internally than the administrative time saved.

Loop closure rate and where referrals stall. Practices are usually surprised by both, and the stall points cluster in two or three places that can be addressed directly.

It runs on Agentplace. Agentplace is an AI agent platform where the agent gets its own page, talks to your visitors there, and carries the request through to the end instead of handing it to a form. You can open this template and change any step before you publish it.


AI Agent for Patient Referral Coordination

Tracks referrals out and in, chases the ones that stall, and flags the patients who have fallen between two providers. Open it in Agentplace and change any step before you publish.

Start from this template
Edit it — the agent is built from this briefBuild this agent