Patients on the queue before they walk in

The problem

The EHR's data connection does not include an appointment's provider or status; those only appear in the scheduling screens. So a patient showed up on the front desk queue only once they arrived and were added at the desk or checked in at the kiosk, and the desk then looked up insurance and copay inside the EHR.

What was built

An RPA bot on a clinic computer signs into the EHR every 15 minutes during business hours, exports the scheduling report, and loads the next two weeks of provider appointments onto the front desk queue, routed to the right desk. It skips cancelled and rescheduled visits and adopts patients already on the queue instead of adding them twice.

For each pre-registered patient it can match, the system then looks up insurance, phone and copay and fills them in on the queue, without overwriting anything staff typed. Visit-type rules set the copay for special visits, staff can override any copay, and the kiosk tells the patient their copay at check-in.

What was hard

Walk-ins added by hand and booked in the EHR minutes later could have become duplicates, so the sync matches them the same tolerant way the kiosk does. The EHR also renames patient records and moves appointments during the day, so rows left over from earlier exports are pruned on every run.

The EHR's older interface takes about two seconds per patient and rejects parallel requests, so lookups run one at a time in the background with a cache, and the server answers right away while the work continues.

Result

The first clinic day had 137 patients pre-registered with zero duplicates, and the queue held 1,192 pre-registered visits for 1–14 October with none left unrouted. By 2 PM on the autofill's first day, 812 of 1,212 upcoming visits already had insurance filled in, and visit-type rules had set the copay on 412.

Not measured: how long the desk used to spend on this per patient.

Where the numbers come from

The bot's run logs and build notes from September and October 2026.