Skip to content

What AI changes about demand planning, forecasting, and replenishment

Learn how spreadsheet and formula-based planning compares with what AI adds to demand planning, forecasting, and replenishment, and where Katana’s AI tools fit in.

November 17, 2026
12 min read
Andreia Mendes

Andreia Mendes

In the first half of 2026, about 17% to 20% of US businesses used AI in some part of their operations. Companies with 20 or more employees increased their use during this time, while for smaller ones, it stayed the same. 

Still, most growing brands run day-to-day planning on spreadsheets or fixed reorder-point formulas. Even before AI, planning software improved on those methods, with reorder alerts, seasonality settings, stockout correction, and purchase orders built from the same screen. 

Those features made the routine faster, but the basics of demand planning stayed the same. When you sell across multiple channels, such as Shopify, Amazon, and wholesale, you need all your channels in one system and reorder points that match your lead times, just as we explained in our multi-channel demand planning guide. AI keeps those basics too. What’s different is the software itself and how much more it can do with the same sales history that spreadsheets or static formulas use. 

Now, let’s look at how AI improves standard planning, forecasting, and replenishment, and see how Katana’s AI tools handle each of these tasks. 

Demand planning: skipping the manual review

Using spreadsheets for planning depends on a regular manual review: you need to export stock and sales data, check what’s running low, and decide what to reorder. The more SKUs, channels, or locations you add, the longer the process takes, and if anything changes between reviews, nothing automatically flags it. 

In most planning tools, you set a reorder threshold for each item, and the software tells you when stock falls below it, using live stock and open orders from every channel. That saves the export, but you still have to keep the thresholds up to date. 

Katana’s AI Forecasting uses what’s already in your account, so there’s no export at all. It reads your sales orders, open purchase and manufacturing orders, and current stock levels. You can run it on a single SKU or up to 50 at once, filtered by supplier or ABC class. Larger runs process in the background, and results show up item by item. 

In practice, this means you don’t have to manually check each item in a supplier’s catalog before placing an order. You can select the SKUs you buy from that supplier, and Katana gives you a recommended order quantity for each with an explanation of how it was calculated. You still make the final decision on what to order. You still make the final decision on what to order. For a quick check across the whole catalog, you can connect Katana to Claude or ChatGPT through Katana MCP, or use Katana AI to ask which products are running low.

Stock risk signals work alongside AI Forecasting. They check your whole catalog all the time, based on past sales and current stock, and flag the items at risk of running out. You can then create an order straight away or run AI Replenishment on the item for a closer look. This helps when the signal doesn’t show the full picture, for example, when a seasonal product is near the end of its season or a product is being discontinued, and the forecast may suggest ordering less or not at all. 

Forecasting: adding real-world context 

Most reorder points use a simple formula: average daily sales times lead time in days, plus safety stock. For example, if you sell 10 units a day and your supplier needs 14 days to deliver, you would reorder at 140 units, plus your safety buffer. 

It works as a starting point but has two limits. First, it treats every past sale the same, so a one-time promotion can raise your average as much as a real increase in demand. Second, it only uses the lead time you last entered. If your supplier now takes 28 days instead of 14, your reorder point will not reflect this until you update the number manually. 

These limits are also why the same best sellers tend to run out more than once. Most brands do not notice this pattern in their own data. In Katana’s stockout report, which looked at 375 product brands, 59% of stockouts happened to products that had already sold out earlier that year. 

Dedicated forecasting tools improve on the formula with statistical models that pick up seasonality and trend. Many also let you exclude stockout periods or one-off wholesale orders to avoid skewing the history. Across the wider market, McKinsey estimates AI-driven forecasting can reduce forecast errors by 20 to 50% compared with spreadsheet methods. 

Katana’s AI Forecasting reads up to 13 months of your sales history, with a minimum of six months. It builds a 12-month forecast for each SKU based on the seasonal patterns and trends it finds. You can also add context that your sales history doesn’t show. For example, if you have a campaign planned for spring, you can say so in plain language and the forecast updates to include it. Every number comes with an explanation of how it was calculated so you can check it against your own knowledge. Lead times are still yours to keep up to date, and an accurate lead time is what makes the recommendation reliable. 

“In my mind, I always wanted this system that I could just plug in how much money I think I’m going to make each month, and it would be like: here’s how much you should order.”
– Tom Lewis, Avoda Golf

Replenishment: fast-track to “order ready to approve” 

Once you know what to reorder, most tools suggest a single order quantity for each item. If you use a spreadsheet, you calculate that number with a formula and create the purchase order yourself. Planning software usually pre-fills the PO for you, but the quantity still comes from a single calculation, so if cash is tight this month or you’d rather stock up ahead of peak season, adjusting it is up to you. 

AI Forecasting  gives you three order scenarios for each SKU (minimum, recommended, and maximum) and how many days of stock each one covers. You pick the one that fits your budget and how much stockout risk you’re comfortable with. The tool recommends a safety stock level you can apply to the product in one click, then creates a draft PO or MO with the supplier and destination location already filled in, so you just need to approve it. With Katana MCP, you can also ask for a purchase order in your AI chat and check the details before it is sent. 

Where to start with AI in your planning 

A good place to start is with the products where a wrong order costs you most, like your best sellers or items with long lead times. Run AI Forecasting on those first and compare its recommendations with your usual orders. If you know something the sales history doesn’t, add it as context and see how the forecast changes. 

You can set up Katana MCP in just a few minutes with your Claude or use Katana AI. After connecting, you can check stock and orders for your entire catalog and create purchase orders directly from the chat once you review the details. No matter which tool you choose, you always decide what and how much to order. 

To see how these tools apply to your business, read our guide to AI for inventory management, see how AI Forecasting works, or learn how to run Katana from your AI chat. 

FAQ about Katana’s AI Forecasting

Statistical forecasting uses methods like moving averages or seasonal models to project demand from your sales history. AI forecasting also starts from sales history, but Katana’s AI Forecasting lets you add context the history doesn’t show, like a planned campaign, and explain how each number was calculated.

Yes. AI Replenishment uses your lead times to calculate reorder points and safety stock, so keeping them accurate is what makes the recommendations reliable.

Stockout risk shows a risk badge for every item on your Inventory screen, based on the last 30 days of demand, current stock and lead time, and lets you create an order from the same row. AI Forcasting forecasts demand for the next 12 months for the items you select, using your sales history and any context you add. Clicking an item’s risk badge opens it in AI Forecasting.

Get inventory trends, news, and tips every month

Get visibility over your sales and stock

Wave goodbye to uncertainty with Katana Cloud Inventory — AI-powered for total inventory control