Forecast with AI gives you a system-generated demand starting point for eligible choices in New Assortment Plan. It is best used when a choice is already linked to historical selling data, has more than 1 year of sales history at the channel level, and you want a forecast that better learns sales behavior than the standard Forecast workflow.

Before you run Forecast with AI, confirm the following:

Choices with 1 year or less of channel-level sales history are not currently eligible for Forecast with AI. Choices that did not sell at least 1 unit at the channel level during the most recent week are also not currently forecastable. If the choice is too new, inactive, or still missing core setup, the standard Forecast workflow is usually the better starting point.

It is also important to plan the price and allowance inputs that the forecast should follow. If these values are missing, Forecast with AI often underestimates demand.

[Screenshot: Right-click menu in New Assortment Plan showing the Forecast with AI action on selected choices.]

[Screenshot: Task Manager showing a Forecast with AI task in progress or completed.]

New Assortment Plan includes both Forecast and Forecast with AI, but they are not the same workflow.

Use Forecast when you want direct control over the forecast logic through your planning attributes. Use Forecast with AI when you want Toolio to generate a forecast from the choice's linked selling history and setup, with a model that better learns sales behavior and is natively sensitive to price, discount, and markdown plan changes.

Forecast with AI is sensitive to price and allowance inputs, so these values should be planned before you rely on the result. In this context, allowances means your planned Discount and Markdown values.

If all of these inputs are left unplanned, Toolio currently treats Discount and Markdown as zero and falls back to last year's Average Unit Retail. In practice, that often leads to a materially underestimated forecast because the model is very sensitive to these values and the forecast magnitude is affected a lot.

When the task finishes successfully, review the Original AI Forecast Units metric for the choice.

If the AI result does not fit the business context for that choice, you can still use the standard Forecast workflow or adjust the plan manually.

[Screenshot: Choice metrics or details section showing Original AI Forecast Units after a successful run.]

Start with this checklist:

If the task still fails after setup is confirmed, contact Toolio Support.

This message means the feature is not enabled for your tenant. Reach out to Toolio Support if you believe the feature should already be available in your environment.

Forecast with AI only works when the choice offering is linked to an actual choice. Link the choice to its actual selling history, then run the task again.

Today, the linked actual choice must have more than 1 year of sales history at the channel level to be eligible for Forecast with AI. It must also have sold at least 1 unit at the channel level during the most recent week. If the choice does not meet those current requirements, use the standard Forecast workflow instead.

At the moment, having enough history is not enough by itself. The linked actual choice must also have sold at least 1 unit at the channel level during the most recent week. This is a current product limitation and is expected to evolve in the coming quarters.

Plan Discount and Markdown when you know them for the forecast period. If you are not planning them week by week, set a default discount through Unit Ticket and Unit Retail on the choice offering.

If all of those inputs are left unplanned, Toolio currently sets Discount and Markdown to zero and falls back to last year's Average Unit Retail. In practice, that often leads to a materially underestimated forecast.

The AI forecast needs an automated cluster on the choice offering so Toolio can place the result in the correct planning context. Assign the automated cluster, save the change, and retry the task.

If launch and end dates are missing or not set correctly, Toolio cannot build the forecast interval. Review those dates on the choice, save the update, and run Forecast with AI again.

This usually means there is a setup mismatch between the automated cluster channel and the channels assigned to the locations included in that cluster. For example, an automated cluster assigned to Channel A cannot include locations assigned to Channel B. Review the automated cluster assignment, the channel on that cluster, and the channels on the included locations, then retry.

This message can mean Toolio could not access the backend AI forecast resources needed for that run. If Forecast with AI is enabled for your tenant, the most common causes today are that the linked actual choice does not yet have more than 1 year of channel-level sales history or did not sell at least 1 unit at the channel level during the most recent week.

This error message is still being improved. For now, review whether the linked actual choice has more than 1 year of established sales history at the channel level and whether it sold at least 1 unit at the channel level during the most recent week. If the item is too new, inactive, or has limited history, use the standard Forecast workflow instead. If the choice should be eligible and the problem continues, contact Toolio Support.

This means Toolio could not finish distributing the AI forecast into the automated cluster tied to that choice offering. Retry the task after confirming setup. If it keeps failing, contact Toolio Support.

This means the task completed without producing usable forecast output for that choice offering and automated cluster. Recheck the choice setup, confirm the linked actual choice has more than 1 year of channel-level sales history, and then try again.

Include the choice offering you were forecasting, the automated cluster if one is shown in the error, the exact message from Task Manager, and a screenshot of the choice setup. That gives Support the fastest path to diagnose the issue.

Treat Forecast Confidence Score as a beta capability for now. It is not yet fully consistent and is expected to be reworked in the coming quarters, so use it as a directional input rather than a final decision rule.

Forecast in Assortment Plan

Choice Level Planning

Receipts Generation and Reviewing Receipts Plan

Forecast with AI is still under active development, and some parts of the experience are expected to change in the coming quarters.

Treat these behaviors as current product limitations rather than long-term final rules.