Step 1 (days 1–3): supplier records, service levels and the cash ceiling in Setup
The first step with an Inventory FTE is custom onboarding, because an order proposal is only as good as the supplier record behind it. In the Setup tab, with the ftes.ai team, the owner confirms each supplier's lead time as measured from purchase order history, minimum order quantities and price breaks, landed cost per SKU, the service level per SKU tier, the cash ceiling per month for purchase orders, who approves orders and above what value a second approver is needed.
The baseline is recorded at the same time: stockouts in the last quarter by SKU and days, cash in inventory today, and the share of it in SKUs with more cover than the owner's threshold. The Inventory FTE never asks for what onboarding already captured, so Setup is the supplier and cash rules and nothing else.
Step 2 (days 4–7): review the stock audit in the Stock tab
The second step is the stock audit, which the Inventory FTE runs on connection and reports in the Stock tab. For every SKU it shows current stock, forecast demand with its confidence label, days of cover, the calculated reorder point with its inputs, and the cash held. It ranks two lists: SKUs at risk of stockout within the horizon, by revenue at risk, and SKUs with excess cover, by cash trapped. It flags suppliers whose measured lead time or variance disagrees with what the owner believed.
The owner's job is to read the ranking, correct any supplier or cost input the audit surfaced as doubtful, and set the service levels where the defaults do not fit. Those corrections change the reorder points before the first order is proposed.
Step 3 (days 8–21): approve the first purchase orders in Approvals
The third step is the first purchase order proposals in the Approvals tab, starting with the SKUs at highest stockout risk. Each shows the supplier, the quantity and why, the measured lead time and variance, landed cost, minimum quantity applied, the cash it ties up and when it falls due, the days of cover it provides, and the forecast with its confidence. The named approver approves, edits the quantity, or rejects with a note; orders above the threshold wait for the second approver.
Rejections and edits are the most valuable signal in the first month. An approver who cuts a quantity because a promotion was cancelled teaches the FTE to ask about promotions; one who rejects a supplier teaches it a preference the record did not hold. Each becomes a rule.
Step 4 (days 8–30, daily): watch the risk lists in the morning brief
The fourth step runs daily: the morning brief in the Today tab lists SKUs that reached their reorder point overnight with the proposal ready, SKUs whose stockout risk moved into the horizon, cash committed in approved orders against the ceiling, and dead-stock proposals such as a markdown for approval. Receipts that arrived late or short are listed with the supplier, because they change the lead-time record.
The ceiling is enforced in proposals: the FTE will not propose an order that would take committed cash over it, and it says so when a risk cannot be covered inside the ceiling, so the owner can decide to raise it or accept the risk.
Step 5 (day 30): the accuracy review in Performance and the autonomy decision
The fifth step is the monthly review in the Performance tab. It reports the stockouts that happened, by SKU and days, against the baseline; cash in inventory and in dead stock against the ceiling; forecast accuracy per SKU, comparing what was forecast to what sold, with the SKUs it got most wrong named and a reason where the FTE can see one; and the purchase orders proposed, approved, edited and rejected. It proposes what to change: a service level, a supplier, a reorder point the edits suggest was wrong.
There is no autonomy decision on ordering, because the Inventory FTE never places orders. The decision is whether to let it revise reorder points without approval as inputs move, which most owners grant once the first month's edits are few.
What does done look like after 30 days?
Done after thirty days with an Inventory FTE looks like a reorder point per SKU with inputs the owner has checked, a supplier record with measured lead times and variance, purchase orders that arrive as complete proposals with their cash impact, a daily risk list ranked by revenue and cash, and a forecast accuracy report that names its misses. It also looks like a set of rules the rejections and edits taught it.
Done does not look like a claim of stockouts prevented. The first honest review reports the stockouts that happened, which may include ones already in motion before the role started, and the cover it holds today.