Deepflow Forecast Update: Explore New Features That Make S&OP Easier

Demand forecasting does not end with accurate numbers. To put forecasts to work, teams need to determine which items require attention first, whether current inventory is sufficient, and whether order quantities need to be adjusted.

With these everyday workflows in mind, Impactive AI has revamped Deepflow Forecast.

This update goes beyond simply displaying demand forecasts. We have comprehensively redesigned key features and screens to help users identify demand changes and inventory risks at a glance, review and adjust forecasts, and turn those insights into actionable decisions. Here is a closer look at what is new.

‍

An Integrated Demand and Inventory View That Highlights SKUs Requiring Action

For teams managing hundreds or thousands of SKUs, reviewing every item's forecast chart individually is impractical.

The redesigned Deepflow Forecast puts the SKUs that need attention and their key status indicators front and center as soon as users open the dashboard.

Items are categorized by inventory status, including normal, shortage, excess, and stockout. Products experiencing sharp changes in demand are also highlighted. Instead of reviewing every SKU, users can first answer, “Which products are at risk right now?” and then explore the items that need attention.

‍

Inventory Coverage Management That Accounts for Lead Times

Knowing that you have 10,000 units in stock does not tell you whether that inventory is sufficient.

What matters is how long current inventory can meet expected future demand. Deepflow now combines current inventory with demand forecasts to help users assess inventory coverage.

Users can also set shortage and excess inventory thresholds based on each item's lead time.

For example, if it takes 30 days from placing an order until a product is ready for sale, an item can be flagged as having insufficient inventory when current stock cannot cover expected demand over those 30 days. This allows teams to assess inventory based on both future demand and the time required to replenish stock, rather than stock quantities alone.

‍

Quickly Detect Sudden Spikes and Drops in Demand

Unexpected changes in demand can quickly translate into inventory risks. If a sudden increase in sales goes unnoticed, stockouts can lead to lost sales opportunities. Conversely, a sharp decline in sales can increase the risk of excess inventory.

The updated Deepflow Forecast automatically identifies significant demand changes using comparisons with periods such as the previous month and the same period last year.

Users can quickly identify SKUs experiencing demand spikes or drops and assess whether the change is temporary or requires further action.

‍

Assess AI Forecast Reliability Through Bias Analysis and Baseline Comparisons

Deepflow Forecast now includes metrics that reveal forecast bias and compare AI forecasting performance with conventional methods.

By comparing historical AI forecasts with actual results over a given period, users can identify whether the AI tends to consistently overforecast or underforecast a particular SKU. For example, if forecasts have consistently exceeded actual demand over the past 12 months, teams can account for that bias in their next ordering or inventory decision.

Users can also compare AI forecast errors with those of familiar methods, such as three-month and six-month averages, to see at a glance which approach performs better for each item.

This enables more informed decisions by considering forecast patterns and performance against conventional methods, rather than simply accepting AI forecasts at face value.

‍

View Shipments, Forecasts, and Inventory Levels in One Dashboard

The SKU detail view has also been redesigned. Users can view historical shipments and the corresponding forecasts alongside current inventory and projected inventory coverage in one continuous view. Demand fluctuations, actual sales, and inventory changes are presented together, giving users context beyond the forecast itself.

Users can also compare previous sales plans with actual shipments to assess whether current forecasts are reasonable. This makes it easier to understand how forecast results affect day-to-day inventory operations.

‍

Record the Causes Behind Unusual Sales Results

Actual sales data often reflects circumstances that AI cannot understand from numbers alone. A sharp decline in sales may result from a stockout rather than a drop in market demand. Conversely, influencer exposure or a one-time promotion may temporarily drive demand higher.

Deepflow Forecast now includes a feature for identifying unusual sales results and recording the events that caused them.

For example, users can document sales declines caused by stockouts, temporary increases driven by promotions, factory relocations or production issues, and one-time external events.

These records help teams understand historical conditions while building a foundation of higher-quality training data to improve forecasting models over time.

‍

Forecast Scenarios That Reflect Your Business Priorities

Inventory strategies vary from business to business. Some companies prioritize minimizing inventory costs, while others place greater importance on preventing stockouts, even if that means carrying some additional stock.

Rather than presenting only a single AI forecast, Deepflow Forecast provides multiple scenarios based on historical forecast errors and bias to support decision-making.

Users can compare scenarios that follow the baseline AI forecast, take a more conservative approach to quantities to reduce excess inventory, or increase quantities to prioritize stockout prevention.

The goal is not for AI to determine a single correct answer, but to give teams a better starting point and a practical basis for making decisions.

‍

A Connected Workflow from Demand Forecasting to S&OP

Deepflow Forecast connects AI-generated demand forecasts with the sales and operations planning (S&OP) process, enabling teams across sales, supply chain management, and other functions to review forecasts together and agree on a final demand plan.

Teams can start by comparing sales plans with AI forecasts to determine production and order quantities that better align with actual demand. This can help reduce excess inventory and stockout risks while supporting improved profit margins.

Key discussion points and decisions that were previously scattered across departmental spreadsheets and emails can now be managed in one integrated S&OP system. This makes it easier to consolidate data across teams and systematically preserve both data and decision histories as company assets.

For example, AI may forecast demand of 13,000 units for a product, while the sales team expects to sell 20,000 units based on an upcoming TV shopping promotion. Teams can record both figures and their supporting rationale, compare and review them during the S&OP process, and document the final decision and its reasoning in the system.

The decisions and data accumulated through Deepflow S&OP can then be used by demand forecasting models and AI agents to further improve models and AI workflows.

‍

An AI Assistant That Makes Use of Accumulated Business Data

Deepflow continues to evolve to capture not only forecast results, but also the causes of unusual sales results, past issues, and decision histories generated through everyday work.

The AI Assistant makes this data easier to access and analyze. Instead of manually searching historical records, users can ask AI what happened during a particular period or what caused a change in demand and quickly find the information they need.

Deepflow is expanding beyond displaying forecast values to help businesses use AI to work with their accumulated demand, inventory, and operational data.

‍

Bringing AI Forecasts into Everyday Supply Chain Operations

Deepflow Forecast is designed to do more than provide additional charts and metrics. Its purpose is to help teams continuously improve their decisions through a connected workflow—from forecasting and risk assessment to S&OP meetings and final decisions.

Guided by real-world workflows and customer feedback, we will continue to enhance Deepflow so that AI forecasts translate naturally into practical business decisions.