AI Inventory Manager overview
AI Inventory Manager is a beta AI-driven replenishment module built into Zenoti that combines a machine-learning demand-forecasting layer with transparent, auditable reorder math to surface daily restocking recommendations for your centers. It never places an order automatically β a human always reviews and confirms before any purchase order is sent.
Availability: AI Inventory Manager is a beta feature that requires enablement for your organization before the module is accessible. Contact your Zenoti representative or customer success manager to activate it.
Before you begin
- AI Inventory Manager must be enabled for your organization before the module appears in the navigation.
- On first-time access, you must complete a 4-step setup wizard before any AI Inventory Manager features are usable.
- To view vendor performance metrics in the Vendor Scorecard, you need Add or Edit permission on the Inventory security object.
How it works
AI Inventory Manager operates as two cooperating layers stacked on top of your existing inventory data.
Layer 1 β AI demand forecasting
The first layer is a machine-learning model that analyzes your center's historical consumption to produce a Predicted daily usage figure for each product. This is the AI's estimate of how many units a product consumes per day, adjusted for day-of-week patterns, growth trends, seasonality, holidays, and promotions.
Alongside the predicted daily usage, the model assigns a confidence tier β a stability and reliability score (T1 β High, T2 β Medium, T3 β Low, or T4 β Very Low) that reflects how consistent the product's demand history is. The tier appears as a badge on the action grid and inside the explainability panel. For the score range behind each tier and guidance on how closely to scrutinize a recommendation, see Check confidence scores and tiers.
The AI model refreshes nightly. Recommendations shown during the day reflect the most recent nightly run.
Layer 2 β Deterministic reorder math
The second layer takes the AI's predicted daily usage and applies straightforward, auditable arithmetic to answer three questions:
- When should I reorder? β The Reorder level (also called the Dynamic Reorder Point) is the on-hand quantity at which a replenishment order should be raised. It equals the units needed during the delivery window plus a buffer for demand spikes or late deliveries.
- How much stock do I want after reordering? β The Target stock is the on-hand quantity you want to reach, based on your Reorder coverage setting (the number of days of stock you want on hand after an order arrives).
- How many units should I order? β The Suggested order quantity is the gap between target stock and current on-hand stock, rounded up to the vendor's pack size.
This layer is entirely deterministic β the same inputs always produce the same outputs. You can inspect every number in the Explainability Panel for any product.
What the system tracks as daily usage
AI Inventory Manager tracks demand separately for retail and consumable products because each type is consumed through a different mechanism.
Retail products β Daily usage counts Point of Sale (POS) sales to guests only. The following do NOT count toward daily usage: returns, transfers, write-offs, positive adjustments, and voided sales. Negative audit corrections DO count as untracked consumption.
Consumable/professional products β Daily usage counts employee checkouts and bill-of-materials (BOM) auto or manual consumption recorded when a service completes, plus negative adjustments and negative audit corrections. The following do NOT count: transfers out, write-offs, positive adjustments, and vendor returns.
The base daily usage figure is the total units consumed over the trailing 90 days divided by 90. The AI model then adjusts this base figure to produce the predicted daily usage shown in the action grid.
Note: For consumable and professional products, the AI-predicted daily usage also incorporates forward demand from scheduled service appointments. When appointments are booked, the system reads the booked services and their associated bill-of-materials quantities to anticipate upcoming consumption. Retail products do not carry this forward-demand signal.
Stock status
Every product in the action grid carries one of four status indicators. Three are derived from days of stock left; a fourth identifies products holding more stock than the target coverage:
| Status | Meaning |
|---|---|
| Critical | Days of stock left is less than the lead time β you may run out before an order can arrive |
| Low | Days of stock left is less than the lead time plus safety days β you are within the buffer window |
| Normal | Days of stock left exceeds the lead time plus safety days |
| Overstock | On-hand stock exceeds the target coverage β you are holding more than the replenishment target |
Note: If a product's daily usage is zero, the reorder level is zero and no reorder alert fires. The product is flagged as slow-moving instead.
What is gated by AI Inventory Manager and what is not
Some Zenoti inventory settings are available only when the AI Inventory Manager add-on is enabled for your organization. Others are exclusive to the AI Inventory Manager module's own screens. Understanding this distinction helps administrators know which settings are active and when.
Exclusive to AI Inventory Manager β The following surfaces and fields are only accessible when AI Inventory Manager is enabled for your organization:
- AI Inventory dashboard (accessible via Inventory β AI Inventory in the left navigation)
- Action grid with AI-prioritized product recommendations
- Demand analytics tab (Demand Forecast, Top Performers, Profitability & Turnover, AI Insights views)
- Vendor Scorecard screen
- 4-step setup wizard (Welcome, Reorder coverage, Lead time & safety, Review & Activate)
- Pack size fields on product and vendor records β these fields are surfaced within the AI Inventory Manager module and controlled by the EnablePackSizeForProducts organization setting. This setting is distinct from the minimum order quantity concept; minimum order quantity is not stored or enforced anywhere in the AI Inventory Manager data model.
Administrator settings available only with the AI Inventory Manager add-on β The Configuration > Inventory settings for Quantity rules, Minimum order value, and Inventory budget appear only for organizations that have the AI Inventory Manager add-on enabled. Disabling the add-on turns off these settings, along with inventory budget, minimum order value, and pack size enforcement. The checks behind these three settings run on both the AI Inventory dashboard and the standard purchase order pages.
Key features
AI-predicted daily usage β The ML model adapts the base 90-day average for day-of-week patterns, growth trends, seasonality, holidays, and promotions, producing a more accurate forward-looking demand estimate than a simple rolling average.
Confidence tiers β Every forecast carries a T1 β T4 confidence tier so you can see at a glance how reliable the AI's estimate is for each product before acting on it.
Explainability Panel β For any product, you can open a panel showing four parts: Summary, Inputs, Calculations, and Confidence. The Calculations part shows the full reorder arithmetic β predicted daily usage, lead time, safety days, reorder level, target stock, and suggested order quantity β so you can verify every number.
Bulk purchase order creation β Select AI-recommended items (or a filtered subset) and raise purchase orders to all relevant vendors in a single action. The flow splits selected items into one purchase order per vendor, pre-fills quantities, and shows retail and consumable order amounts separately. Every line is editable before you send.
Vendor selection strategies β When building a bulk purchase order, you can choose how the system assigns a vendor to each line:
Strategy Description Last ordered vendor Default. Uses the same vendor as the previous order for that SKU. Least cost Uses the vendor with the lowest unit price for that line. Most frequently ordered Uses the vendor with the most past purchase orders for that SKU. Least lead time Uses the vendor with the shortest average delivery time β useful for urgent items. You can manually override the vendor on any individual line regardless of the selected strategy. All four strategies are computed from your organization's own purchase order history, not from any external rating service.
Pack size rounding β Suggested quantities are always rounded up to the vendor's pack size, never down, to avoid under-target orders. The system does not apply a separate vendor minimum order quantity (MOQ). If you enter an off-pack-size quantity manually, the system displays a warning but does not block you from sending.
Demand analytics β The Demand analytics tab provides forward-looking planning views covering a 30β90 day horizon, distinct from the immediate-action focus of the Inventory overview tab. It includes four views:
- Demand Forecast β Historical usage and an AI-predicted usage curve per product and center, with a 7/15/30-day horizon toggle and per-product insights such as trend direction, volatility, and surge or opportunity flags.
- Top Performers β A leaderboard of products ranked by revenue, units sold, or transaction count, with trend indicators.
- Profitability & Turnover β Margin against idle-capital exposure, drillable from the organization level down to a single product.
- AI Insights β Auto-detected pattern cards covering surges, slowdowns, growth, trapped capital, and below-cost discounting.
Vendor Scorecard β A dedicated screen showing vendor performance metrics derived from your organization's own purchase order history, including actual delivery times, order frequency, and pricing. Available as both an embeddable single-vendor view and a full grid view listing all vendors.
Org / Zone / Center views β AI Inventory Manager supports three view levels: Center (a single location, the primary beta experience with all action grid columns and actions available), Zone (a regional roll-up), and Organization (rolled up across every center, with a center picker so a central team can raise purchase orders on behalf of individual centers from one place).
Note: The Reorder alerts KPI tile is available in center view only. At organization or zone view, this tile is hidden and the remaining KPI tiles roll up across all centers.
Settings cascade β Reorder coverage, safety days, and lead time settings can be configured at the organization level and overridden at the center or individual product level. The most specific setting always wins: SKU overrides category, category overrides center, center overrides organization. Setting changes take effect at the next nightly refresh β they do not recompute existing recommendations immediately.
Human-in-the-loop design β The system never sends a purchase order without an explicit user action to confirm. Every suggested order is a recommendation you review, adjust, and approve before it is raised.