An inventory and replenishment agent that works on your own accounts

Many Amazon brands already have the numbers an inventory agent needs. They sit in Seller Central reports, the Walmart and Shopify admins, the 3PL portal and a buyer's spreadsheet, and someone stitches them together once a week. This page describes the inventory and replenishment agent we build and deploy on a brand's own accounts: what it watches, what it does within your rules, what it sends back to your team, and when another option fits better.

What does an inventory agent do for an Amazon brand?

It does the daily inventory review a good operations lead would do if they had the time. On every run it rebuilds the stock position for each SKU across every place stock sits, compares days of cover with the real lead time to refill it, and decides whether anything needs to happen today. Most days, for most SKUs, nothing does. When something does, it raises an alert or prepares a draft with the evidence attached, so the person who owns the decision can approve it in minutes instead of rebuilding the analysis.

The problems it works on are the ones on our stockouts, replenishment and storage fees pages: top sellers running out at FBA while the 3PL holds weeks of cover, purchase orders placed late because one leg of the lead time was forgotten, and slow stock drifting toward Amazon's aged inventory surcharge. Those pages explain each problem and its fix. This page is about the agent that runs those fixes every day, so they do not depend on someone remembering.

The agent starts from the inventory and replenishment module in Ecomsellertool Growth OS, which tracks days of cover per SKU and channel, sets reorder points from real lead times and drafts POs and shipment plans before you run out. We add the model steps, the connections to your other channels and systems, and the rules your team sets. It is one of the AI agents we build on a brand's own accounts.

Which actions run on their own, and which come back to your team?

Every run ends in one of three outcomes: nothing to do, something the agent handles within your rules, or a decision that comes back to a person with the evidence and a proposed next step. The table below is the default split we start from, signal by signal. Your team moves the lines: a brand with a steady supplier may let small top-up transfers run alone, while another wants to see every shipment. The inventory-risk screen near the top of this page, one account's Growth OS screen, shows the kind of out-of-stock and at-risk flags this table starts from.

The ads flag matters because Amazon Ads says a Sponsored Products ad does not display once the product is out of stock or is not the Featured Offer (Amazon Ads FAQ, checked September 25, 2026). Spending hard on a SKU in its last week of stock speeds up the run-out, and then the ads stop anyway. Our advertising agent reads the flag directly; without it, the alert goes to whoever runs your campaigns. The reasoning is on our inventory-aware ad spend page.

What data does it watch across Amazon, Walmart, Shopify and your 3PL?

An inventory agent is only as good as the stock position it builds, so most of the work in a deployment is connecting the sources correctly. The agent reads each system through the official route that system offers and that you authorize. It keeps days of cover separate for each channel, because a SKU can look healthy in total while it is out of stock where it sells fastest.

The Amazon side is well documented. Amazon's FBA report documentation describes the Manage Inventory Health report as a consolidated view of sales over 7, 30, 60 and 90 days, shipment statuses, estimated storage and long-term storage fees, and aged and excess units, and the Restock Inventory report as recommendations on what to restock, how much and when (checked September 25, 2026). Walmart's WFS inventory API returns available and inbound units, inventory age, days of supply and sales forecasts per SKU, and Shopify's InventoryLevel object tracks available, on-hand, incoming and committed quantities at each location (both checked September 25, 2026).

Two inputs rarely live in any marketplace. The supplier side is one: confirmations and changed ship dates usually arrive by email, so the agent uses a language model to read them and update the expected arrival on the open PO. The other is the bill of materials for kits and multipacks. The warehouse software we built for Cheddy maps kit, multipack and single-pack ASINs to the components held in the warehouse, and the agent works from the same kind of map, so a bundle's days of cover reflects its scarcest part.

How does it decide when and how much to reorder?

The decision has two parts, and only one of them uses a model. The forecast of daily demand per SKU and channel comes from a forecasting model trained on your own sales history, with stockout days, promotions and seasonality marked, so a week out of stock is not read as a week of no demand. Everything after the forecast is arithmetic and rules your team can read and change:

Reorder point = forecast daily demand × full lead time in days + safety stock

Full lead time means every leg from placing the PO to units being sellable where they are needed: production, freight and customs, receiving at the 3PL or AWD, and the inbound leg into FBA, each measured from your own receipts rather than the supplier's quote. Safety stock scales with how much demand and lead time have actually varied for that SKU. The reorder point is compared with stock in every location plus open POs, never with FBA stock alone.

An illustrative run: a SKU sells 25 units a day on Amazon, 8 on Walmart and 5 on Shopify, 38 in total. Production takes 30 days, ocean freight and customs 30, 3PL receiving 3 and the 3PL-to-FBA leg 12, so the full lead time is 75 days and lead-time demand is 2,850 units. Twelve days of safety stock adds 456, for a reorder point of 3,306. Here is the position the agent finds:

The order brings the position up to the reorder point plus 60 days of sales, a cycle length your team sets: 3,306 + 2,280 − 3,200 = 2,386 units, rounded up to 50 cases of 48, or 2,400 units. The transfers go ahead of the PO because FBA and WFS will run out long before new stock lands. To test the numbers on one of your SKUs, the reorder point calculator runs the same formula, and the stockout cost calculator estimates what a gap would cost.

Capacity shapes the transfers. Amazon sets one FBA capacity limit per storage type each month (standard-size, oversize, apparel, footwear and extra-large), measured in cubic feet. On-hand units plus shipments on the way count against it, and a shipment that would push usage past the limit is blocked. Limits are usually announced in the fourth week of the month, with estimates for the two months after (Amazon's FBA capacity limits page, checked September 25, 2026).

The agent sizes each shipment to the space left in its storage type and ranks SKUs by days of cover when space is short. Open shipments you no longer plan to send still count toward usage, so it lists them for your team to cancel. Extra space through Capacity Manager carries a reservation fee you name, so that request comes back to your team too. For peak season, see our Q4 holiday operations checklist.

What does the agent do about aging stock before the surcharge snapshot?

Aged stock is the other side of the reorder decision. Amazon's aged inventory surcharge applies to FBA units that have been in its network 181 days or longer, assessed on an inventory snapshot taken on the 15th of each month; clothing, shoes, bags, jewelry and watches are excluded from the 181-270 day tiers. Units on a removal or disposal order submitted by 11:59 p.m. PT on the 14th avoid that month's charge, even if they have not left the building yet (Seller Central help, checked September 25, 2026).

Amazon counts age first in, first out across its whole network: a unit sold or removed comes off the oldest stock, whichever unit actually shipped. The agent projects age the same way. It reads the age buckets and per-tier surcharge estimates in Amazon's Manage Inventory Health report, rolls each SKU forward to the next snapshot or two, and flags units about to cross into a charge or a higher tier, above all the step at 271 days, where the rate per cubic foot more than triples.

For each flagged SKU it prices the options: selling through at the current pace, a price cut or deal, moving units to a faster channel, removal, disposal or liquidation. On its own, it holds the SKU out of the next purchase order, because buying more of a slow seller starts the next batch of aged stock.

What it does not do alone is spend money or move stock out. Removal, disposal and liquidation orders, price cuts and deals come back to your team with the estimate behind each option and the date the decision is due. Our long-term storage fees page covers the surcharge tiers and which fix costs least; the screen below is one account's storage-fee view in Growth OS.

When are Amazon's own tools or an inventory app the better choice?

Not every brand needs a custom agent. On September 23, 2026 Amazon announced Seller Assistant workflows that monitor and carry out routine tasks such as restocking, with sellers reviewing and approving actions before they run. It also said sellers can build their own restock agent inside Amazon Quick through its selling partner plugin, which is in beta for sellers in Amazon's US stores (Amazon's announcement, checked September 25, 2026).

Amazon Warehousing and Distribution is the other built-in route. Amazon says AWD auto-replenishment relies on a proprietary data-science model to keep FBA in stock from bulk storage, and that AWD can also send stock to other channels, such as Walmart Fulfillment Services, from the same pool (checked September 25, 2026). Inventory apps are a third route; we checked each one's own page on September 25, 2026.

We are one of the options in this table, so weigh our view accordingly. If Amazon's tools cover your stock and your channels, use them. A custom agent earns its cost when your stock position spans systems your current tools do not connect, or when the rules that decide your orders are specific to your business, such as kit builds, supplier minimums or channel priorities.

What does it take to deploy it on your accounts?

Five steps. Your operations lead takes part in most of them:

1. Diagnose. Start with the free 24-hour diagnostic: connect Amazon with Login with Amazon, no password shared, and get a report within 24 hours of connecting, on business days. We only read data; we never change listings, prices, stock or ads.
2. Connect. You authorize each account through its official API: Amazon's SP-API, Walmart, Shopify, your 3PL's API or file feed, and your ERP or PO system.
3. Map. We record suppliers, lanes, case packs, minimum orders and kit components, and measure each lead-time leg from your receipts.
4. Set the rules. Your team sets floors, safety stock, tolerances and which actions run alone. Everything else comes back for approval.
5. Scope and date. The agent is scoped in an architecture doc with a fixed price and go-live date.

The agent runs on your seller, marketplace and warehouse accounts, not ours, and it is designed to follow Amazon's Agent Policy: it works through the APIs you authorize, acts only within the rules you approve and leaves anything outside those rules to a person on your team. You keep your accounts, your data and the custom code we build; the Growth OS base is licensed to you. We never buy your inventory or sell under our own accounts.

We have built this kind of plumbing before. Cheddy is warehouse software we built that maps kits, multipacks and single-pack ASINs to components. Ecomsherpa is a warehouse management system we built for FBA stock and a local warehouse, covering purchase orders, replenishment quantities, vendor SKUs and components. Amazelligence is rule-based automation we built for purchase-order tracking, with alerts for inventory due and for discrepancies in inbound deliveries. The multi-marketplace layer in Seller7, software we built, pulls each marketplace's data through its own API. None of these is an agent; they are the kind of software the agent builds on.

Ecomsellertool builds and deploys this agent for Amazon-first brands selling on two or more channels. We start from the Growth OS inventory module, connect your marketplaces, 3PLs and suppliers, and add the rules and model steps your operation needs, so your team reviews drafts instead of rebuilding spreadsheets. We have built on Amazon's seller APIs since 2017. Start with the free 24-hour diagnostic to see what it would flag.