Why AI Scheduling Teams Struggle with eClinicalWorks

The Problem

Your AI scheduling model expects discrete time slots: “9:00–9:20 AM is available for a new patient visit.” But eClinicalWorks doesn’t think that way. It gives clinics a flexible system of schedule slot rules with time ranges and capacity counts — and every clinic uses them differently.

We’ve seen hundreds of eCW configurations across our customer base. They broadly fall into three patterns, and understanding which one you’re dealing with is the key to getting availability right.

Pattern 1: Single Appointment Slots

This is the simplest case. Each slot maps directly to one appointment. The time range IS the appointment duration. These clinics are essentially using eCW like a traditional calendar. You can treat the gaps between booked appointments as available time.

Pattern 2: Multi-Capacity Slots

Here the clinic is saying “between 10 and 11 AM, we can see up to 3 established patients.” The slot isn’t a single appointment — it’s a capacity block. You need to track how many appointments of the matching visit type have been booked against that slot to know remaining capacity. Importantly, a “new patient” appointment during the 10–11 AM block does NOT reduce the established patient capacity.

Pattern 3: Daily Capacity Blocks

The entire day is one block. The clinic is saying “we can handle 2 new patients and 4 follow-ups today, scheduled whenever.” Any appointment during the day reduces capacity, regardless of visit type overlap. This is the trickiest pattern because a naive subdivision would show dozens of open slots when only a handful of appointments can actually be booked.

The Real Challenge: You Don’t Know Which Pattern You’re Dealing With

The person integrating your scheduling API is usually a developer — not the clinic admin who configured eCW. They get a blob of schedule slot data and have to figure out which pattern the clinic is using, then write the right logic to interpret it.

And that’s before you account for clinics that mix patterns, have business rules on spreadsheets outside of eCW, or have eCW configured incorrectly.

How Cobalt Handles This

Our GET /availability endpoint has a calculation_method parameter that maps to each of these patterns:

You set the calculation method once per clinic based on how they’ve configured eCW, and the API handles the rest — capacity tracking, conflict detection, and all the edge cases. See the full availability endpoint documentation for request parameters and response shapes.