An Aivell answers "where is my order" from your own order data, runs the returns procedure — window check, label, refund rules — and watches competitor prices while you sleep. It works in the customer's language, escalates the angry ones, and processes everything on a box in your back office, not in the cloud.
"Where is my order?" is half the inbox. The answer is always the same lookup — order number, carrier, tracking link — done by hand, per email, while the queue grows.
Window check, reason code, RMA number, label, refund-or-credit decision, a photo if the item arrived broken. Ten minutes per return when done properly — so on busy days it isn't done properly.
Somebody opens twelve competitor pages and a spreadsheet and starts copying numbers. By Wednesday the prices have moved again. On marketplaces, being €2 wrong is invisible; being €2 wrong for a week is a ranking drop.
Marketplace messages come with a response-time countdown and a seller score attached. Miss enough of them over a long weekend and the account health page turns yellow.
Retail admin has a cruel shape: the same event that brings revenue brings the workload. A good campaign means a wave of orders, which means a wave of WISMO emails three days later, which means a wave of returns the week after — precisely when the team is busiest picking and packing. The queue is never long on a quiet day.
Meanwhile the market work never pauses. Competitor prices move daily. Stock runs low without announcing itself. And a fifth of your buyers are shopping at midnight — sites that answer at midnight capture 20–35% more of them than sites that answer Monday morning.
An Aivell absorbs exactly the work that peaks. WISMO answered from your own order data, instantly, in the buyer’s language. Returns walked through your procedure — the real one, with the window check and the damaged-item photo rule — every time, including Saturday at 11 PM. Prices checked nightly instead of Mondays. Stock counted against reorder points before the bestseller sells out.
You write the rules as you’d tell them to a new hire: refund under €50 without fuss, credit-only past 30 days, Giulia sees the serial returners. Guardrails enforce the money rules at the tool level, and anything over the line waits for a tap on your phone.
It all runs on a box behind your firewall — onboard inference, flat monthly fee, no per-message billing that punishes you for having a good Black Friday. Your customers’ order histories stay in the building, and so does the margin data your repricing depends on.
when a customer asks to return, find the order and check the 30-day window
inside the window? issue the rma number, send the label and instructions
item arrived damaged — ask for a photo, look at it, and under €50 refund straight away, no return shipping
outside the window, offer store credit once; never argue twice
third return from the same customer this month? pause and ping Giulia
refunds over €200 wait for approval — prepare everything, then ask
log every return in the returns sheet with reason code and outcome
Your info@ mailbox, triaged by a private AI colleague — classified, routed, and the routine ones answered. Behind your firewall, around the clock.
The Browser Agent reads competitor pages daily, keeps the history and mails a diff — only what changed, never a data dump. Your watch list stays private.
Recurring customer questions answered by email within minutes, from your own documents — on a box behind your firewall. The rest goes to a human.
Purchase orders arriving by email — PDF, scan, photo or plain text — read onboard, checked against your prices, entered into your system, confirmed back.
Customers write in any language; your Aivell replies in theirs and briefs the team in yours. Translation runs onboard, so mail never leaves the office.
Website support chat answering from your indexed manuals and policies, onboard your box — and handing off to email the moment it isn't sure.
Stock levels watched nightly in your own database, reorder emails drafted per supplier — and nothing is sent until you approve it with a tap on your phone.
On the box, in your back office. The Aivell reads order history, addresses and complaints with onboard inference — none of it is sent to a cloud AI. Your purchase data is your moat; it shouldn't be training material on someone else's servers.
Only inside the limits you set, and those limits are guardrails on the tools — not suggestions. A typical chore lets it refund small amounts under clear conditions and prepares everything else for a one-tap approval from your phone.
Chores run in English, Spanish, French and Italian today, and the Aivell replies in the customer's language regardless of which one you wrote the procedure in. One returns procedure, four markets.
No. The Browser Agent visits the public product pages you list, on the schedule you set, and records the prices into your sheet. It alerts you when a competitor undercuts you past your threshold; repricing stays your decision unless you decide otherwise.
An Aivell is an on-premise AI colleague: a small box that plugs into your network, runs its own AI onboard and takes the repetitive work off your desk. Prompts, documents, data — nothing ever leaves your office. Unplug the internet: it keeps working.
You describe each task as a chore, in plain language. Your Aivell turns it into a solid, guarded procedure and runs it in the background — with guardrails on every tool, a complete audit trail, and approvals in your hand when you want them. One-time setup, fixed monthly fee. No tokens, no overages.
onboard inference · air-gapped · fixed monthly fee