What is an AI employee for an online store?
An AI employee for an online store is a Digital FTE that owns one whole store function rather than answering questions or completing a single task. On ftes.ai the six e-commerce roles are the Product Catalog FTE, the Retention FTE, the Conversion FTE, the Order Support FTE, the Marketplace FTE and the Inventory FTE. Each holds a job description, a set of targets it is reviewed on, and a budget it cannot exceed.
The difference from a store app or a plugin is ownership and accountability. An app runs a rule you configured; a Digital FTE looks at the store every morning, decides what needs doing, proposes it with evidence, and waits in the Approvals tab for the manager's decision. The store owner manages it like a direct report, not like a setting.
Which six store jobs can a Digital FTE own?
The six jobs a Digital FTE can own in e-commerce are the ones every store pays for whether it has one employee or fifty. Each is expensive to do well, boring to do daily, and visible in revenue when neglected. The list below names the role and the function it is accountable for.
- Product Catalog FTE: every SKU page complete, accurate and found, with feed disapprovals fixed at the source.
- Retention FTE: email and SMS lifecycle flows built from order data, every send approved and inside the frequency cap.
- Conversion FTE: the funnel diagnosed, experiments proposed and run on approval, results reported at full sample size.
- Order Support FTE: order-status, returns and exchanges resolved inside policy, sensitive cases escalated with context.
- Marketplace FTE: Amazon and other marketplace listings, Buy Box, sponsored ads inside caps, margin after fees.
- Inventory FTE: per-SKU forecasts, reorder points, purchase orders proposed with their cash impact, every one approved.
How does the approval gate work in a store?
The approval gate on ftes.ai means that every action a Digital FTE takes on a store starts as a proposal in the Approvals tab. A product-page rewrite is shown before and after with every claim mapped to a product field. A lifecycle email is shown exactly as the customer will see it, with the segment size beside it. A purchase order is shown with quantity, lead time, landed cost and the cash it ties up.
The store owner approves, edits or sends the proposal back with a note. Every decision is logged with who made it and what evidence was attached, so the record of why a listing changed or a refund was granted exists the moment it happens. The autonomy ladder lets a manager promote one class of action at a time, such as routine order-status replies, to run without approval once the approval record supports it, and take it back at any time.
Which limits can an e-commerce Digital FTE never cross?
An e-commerce Digital FTE on ftes.ai works inside hard limits that are set by the store owner and cannot be exceeded by the FTE regardless of autonomy level. These limits exist because the cost of a wrong action in a store is immediate and measurable: a price below cost, a refund outside policy, an ad budget spent by noon.
Each role has its own limits. The Retention FTE has a frequency cap and a suppression list. The Conversion FTE has an always-gated list covering price, checkout fields and legal copy. The Order Support FTE has a refund limit and a goodwill limit above which a person decides. The Marketplace FTE has a sponsored-ads cap per marketplace and price floors per SKU. The Inventory FTE cannot send a purchase order; it can only propose one. The token wallet hard-stops every role at zero balance, so no role can overdraw the organization.
What stays with the store owner?
The store owner keeps every decision that shapes the business rather than operates it. Pricing strategy, brand positioning, which products to launch or retire, supplier negotiation and payment terms, the return policy itself, and any customer case that involves judgment or a sum above the limit all stay with people. A Digital FTE prepares the evidence for those decisions; it does not make them.
This is by design, not a limitation to apologize for. A vendor that claims an AI can own pricing strategy or decide a disputed refund is claiming something a store owner should not accept. The value of a Digital FTE in e-commerce is that the six operational functions run every day to a standard, with evidence, so the owner's time goes to the decisions only an owner can make.
How do the six roles work together?
The six e-commerce roles on ftes.ai share one organization, one token wallet, one workspace and one manager, so they work together without a separate integration project. The Product Catalog FTE's rewritten pages are the pages the Conversion FTE tests. The Inventory FTE's stockout risk is visible to the Marketplace FTE before it proposes a sponsored-ads bid on a SKU that is about to run out. The Order Support FTE's claim spikes are evidence the Inventory FTE and the Marketplace FTE can read.
A store does not have to hire all six. Most start with the function that hurts most, usually catalog or order support, and add roles as the approval record builds. Each role reports on its own targets, so a manager reviewing six roles still reads six short briefs, not one long one.
Where should an online store start?
An online store should start with the function whose failure it can already see in its numbers. A store with thousands of SKUs and a Merchant Center full of disapprovals starts with the Product Catalog FTE. A store drowning in order-status tickets starts with the Order Support FTE. A store whose repeat rate is flat starts with the Retention FTE. The right first hire is the one with the clearest baseline to measure against.
Record that baseline before the role starts: catalog completeness, first-reply time, repeat-purchase rate, checkout completion, suppressed listings, stockouts last quarter. The first monthly review compares the FTE's measured numbers against that baseline, and the approval record from the first month decides what, if anything, gets promoted up the autonomy ladder.