Inventory Analysis & AI Operations

#After onboarding, ask AI to analyze inventory

Once WMS accounts, inventory goods, and warehouse mappings are ready, ask inventory questions from Chat or a connected message channel. You do not need to know the underlying data interfaces. Specify the Shop, SKU, warehouse, and date range; AI can read current inventory and historical demand and turn them into an actionable explanation.

Complete WMS onboarding, goods sync, and warehouse mapping first. An unmapped warehouse exposes only the quantity displayed by the platform, without identifying its authoritative third-party WMS stock source.

#Facts AI can combine

Depending on warehouse type and the data returned by the provider, AI can combine:

• current available, locked, in-transit, total, and unavailable WMS quantities;
• inventory displayed by the Shop platform for each product, SKU, and warehouse;
• platform-operated fulfillment pools, including available, reserved, unfulfillable, and in-transit units;
• daily units, order count, and GMV for each Shop SKU over the requested period;
• the relationship of one Seller SKU across multiple Shops and warehouses.

A platform fulfillment pool can be visible to several Shops while each listing also reports its allocated quantity. They are two views of the same physical units and must not be added together. AI explains warehouse pools and Shop allocations separately; preserve that distinction in manual exports as well.

#Common analyses

Find stockout risk

Analyze the last 30 days of SKU demand and current inventory for every onboarded US Shop. Identify the products most likely to run out first, including available, in-transit, average daily units, and the reason for the risk.

Which SKUs accelerated over the last seven days but have insufficient in-transit stock for the US West warehouse?

Compare Shops and warehouses

Compare inventory for SKU ABC-001 across every Shop and warehouse. Separate WMS physical stock, platform-displayed quantity, and platform fulfillment stock without double counting.

Find Shop warehouse quantities that differ materially from their mapped WMS warehouse and list the mappings that need human review.

Prepare replenishment recommendations

Using the last 60 days of sales, current available stock, and in-transit stock, prioritize replenishment for the next 14 days. State the assumptions and do not place an order.

If the next two weeks follow the recent 14-day average, which products fall below seven days of safety stock?

These results support an operational decision. Promotions, seasonality, Creator-driven spikes, returns, and lead time can change demand, so add campaign plans, procurement lead time, and safety-stock rules before acting.

#How far AI can execute

The current product can read inventory and demand facts and, after an explicit instruction, update warehouse-level SKU inventory in TikTok Shop. Inventory changes affect sales. Before execution, specify the Shop, product, target quantity for every warehouse, and verify that no assigned warehouse is missing for a multi-warehouse SKU.

Replacement dispatch or overseas-warehouse fulfillment depends on whether the WMS provides a supported write operation:

• when a writable dispatch capability is integrated, AI can schedule a replacement using the order, SKU, quantity, and address confirmed by a person;
• when the connection provides read-only inventory and order data, AI can investigate, prepare replacement details, and produce an action checklist, but it cannot create the warehouse outbound order;
• refunds, replacements, and policy exceptions should retain human confirmation.

After execution, ask AI to verify the Shop inventory, WMS record, and order state so you know every system accepted the change.

#Future inventory automation

Warehouse mapping and shared product identity are not one-time administrative work. They are the foundation for later automation. As more WMS write capabilities and inventory policies become available, TK Copilot plans to support:

• continuous safety-stock and discrepancy monitoring;
• proactive warnings before stockout;
• replenishment proposals that combine demand, in-transit stock, and lead time;
• purchase orders or transfer tasks within explicit policies and human-review boundaries.

These are roadmap capabilities, not features enabled for every account today. For now, use Scheduled Tasks to have AI run a recurring inventory analysis and deliver the report. Any write or purchase action remains subject to the controls visible in the current product and the capabilities of the connected provider.

#Recommended operating cadence

1. Check connection and synchronization errors daily.
2. Review stockout risk by Shop and warehouse, not only total stock.
3. Review high-volume, high-value, and long-lead-time goods weekly.
4. Run a separate scenario before a promotion or major Creator activation.
5. Verify the resulting records after every inventory adjustment, replacement, or replenishment decision.