ZenotiHelp

AI Employee Scheduler overview

ZTZenoti Docs Team Updated October 25, 2026 6 min readGuide

Availability: AI Employee Scheduler is in Beta and is available only to a closed cohort of centers. Zenoti enables it for your center β€” there is no in-product option to request access or enable it yourself.

AI Employee Scheduler analyzes historical demand patterns to generate suggested provider schedules for the week ahead. Rather than building a roster from scratch, managers review a pre-built suggestion on the Schedule Grid, make any edits, and copy the finalized week into Zenoti when ready. This article explains how the feature works, what the key surfaces do, and when to use the on-demand recompute.

Before you begin

  • Your center must be enrolled in the AI Employee Scheduler Beta. Zenoti sets up your center's data and enrolls it automatically β€” no in-product step is required from your side.
  • When your center is first enrolled, Zenoti sends an email confirming when the first forecast and schedule are ready (allow approximately 24 hours after enrollment).
  • There is no control inside AI Scheduling to turn the feature off or pause it. If you need it paused, contact Zenoti. Pausing leaves any schedules already in the Appointment Book unaffected. Resuming is also a Zenoti-side action.

How it works

AI Employee Scheduler uses historical demand patterns to forecast how busy the center will be across each day of the coming week, then generates a suggested provider schedule designed to match that demand against your target utilization setting.

The feature operates on a rolling 28-day horizon and produces prediction runs at regular intervals for the current week and the next. Managers do not receive a finished, published schedule β€” instead, they receive suggestions for review. Nothing reaches the live schedule until a manager explicitly selects Copy all.

Enrollment and first run

When Zenoti finishes setting up your center's data, opening AI Scheduling for the first time shows a "Training your Employee Scheduler" progress screen while the first forecast and suggested schedule are built. This process takes approximately 24 hours. Zenoti emails your center when the scheduler is ready to use.

The Schedule Grid

The Schedule Grid is the primary workspace in AI Scheduling (accessible via the AI Scheduling button in the schedule toolbar; the page opens titled Demand based Scheduling). It presents the suggested schedule for each provider, organized by day, and marks each provider's day with one of six recommended change types:

Change type What it means
No change The shift already matches forecasted demand
Add hours The shift should be lengthened to meet demand
Reduce hours The shift should be shortened to match demand
Remove The shift is not needed that day
New shift The provider is off but demand requires their presence
Move shift The shift length stays the same but the start and end times should shift

Note: The Help icon on the Schedule Grid opens the Demand based Scheduling - Definitions panel, which maps these grid labels to their Definitions-panel equivalents (for example, Add hours β†’ Extend, Reduce hours β†’ Trim, Remove β†’ Skip, New shift β†’ Schedule). Use the grid labels as the primary reference β€” the Definitions-panel names are a cross-reference only.

Day-level coverage summary

Each day column in the Schedule Grid shows a day-level coverage summary label indicating the overall staffing situation for that day:

Label Meaning
{h}h understaffed Scheduled supply is below the day's predicted demand
+{h}h overstaffed Scheduled supply is above the day's predicted demand
Optimised Balanced β€” no changes suggested
Reallocation Balanced overall, but changes are suggested within the day
No change needed The day requires no adjustment
Off day Nobody is rostered
No demand forecast Providers are rostered but no forecast exists
Center closed The center is closed that day

Each day carries exactly one of these labels. A Min coverage badge marks hours staffed by minimum-presence rules rather than by demand.

Target utilization

Target utilization is the key setting that tells the scheduler how much staffing buffer to maintain above predicted demand. It is a bounded value between 50% and 100% (0.50–1.0).

For example, setting target utilization to 80% means the scheduler plans for a 20% buffer above the forecasted demand β€” ensuring staff capacity exceeds the prediction rather than staffing to cover only the exact forecast. A higher target utilization means more buffer; a lower value means the suggested schedule aligns more tightly to forecasted demand.

Adjusting the target utilization setting is one of the most common ways managers tune the scheduler to their center's needs. After changing the setting, running an on-demand recompute applies the new value to the current week's suggestions immediately.

Calendar-week run selection

The Calendar-Week control lists every available prediction run for the current week and the next week. Selecting a run loads its suggestions into the Schedule Grid. Runs that have not yet generated appear as placeholder entries in the control β€” they become available once the system completes them.

Note: The Calendar-Week control lets you select which run to view. There is no run-comparison feature β€” only one run is displayed at a time.

On-demand recompute

On-demand recompute regenerates the suggested schedule from the demand forecast already on record, without re-running the underlying machine learning prediction pipeline. It typically completes in under 10 seconds.

Use on-demand recompute after any event that changes the scheduling inputs mid-week, including:

  • A leave request that changes which providers are eligible for the week
  • A new hire whose start date falls inside the active scheduling window
  • A change to the center's target utilization setting

On-demand recompute is the standard path for refreshing suggestions after these mid-week changes. It does not generate a new demand forecast β€” it generates a new schedule against the forecast already in place.

In-product feedback

The Definitions panel (opened via the Help icon on the Schedule Grid) includes a Give Feedback option. Selecting it opens a Share Your Feedback form that asks:

  • How useful are the schedule suggestions for your location? (Too many changes suggested / Not enough changes suggested / Suggestions seem about right / I'm not sure how to use this)
  • Tell us what's working or what could be better:

Submitting the form sends your response to the Zenoti product team with your center and the week being viewed attached automatically. The form confirms submission with "Thank you for your feedback!"

This form is the recommended channel for raising issues with schedule suggestions during Beta. Feedback submitted through this entry point includes the scheduling context the product team needs to triage effectively.

Manager review

After reviewing the Schedule Grid and making any edits to the suggested schedule, managers finalize the week by selecting Copy all. This copies the suggestions into the Appointment Book. Until Copy all is selected, no suggested changes reach the live schedule, and the system does not contact providers directly β€” acting on suggestions (such as scheduling a new shift for a provider who is currently off) is always a manager-initiated conversation.

Note: Split shifts are not supported by Copy all at this time.

Key features

  • Beta enrollment β€” Zenoti enrolls centers automatically when their data is set up. There is no customer enable, join, or turn-off control.
  • 28-day rolling horizon β€” Suggestions cover the current and next week, with prediction runs visible in the Calendar-Week control.
  • Six change types β€” The Schedule Grid displays No change, Add hours, Reduce hours, Remove, New shift, and Move shift as actionable recommendations.
  • Day-level coverage summary β€” Each day shows a single coverage label (such as Understaffed, Overstaffed, Optimised, or Reallocation) so managers can prioritize attention at a glance.
  • Target utilization β€” Set per job in the Utilization targets list (50%–100%, default 80%); it controls how much staffing buffer the scheduler maintains above predicted demand.
  • On-demand recompute β€” Refreshes the suggested schedule against the existing forecast in under 10 seconds, without re-running the machine learning pipeline.
  • In-product feedback β€” The Share Your Feedback form, accessible from the Definitions panel, is the sanctioned channel for raising issues with schedule suggestions during Beta.
  • Manager review β€” Suggestions do not publish automatically. Managers review, optionally edit, and then select Copy all to apply the week to the Appointment Book.
Was this article helpful?
Your feedback helps us improve our documentation.