home / use cases / email and inbox / order-emails-to-system

Purchase orders that type themselves in

An Aivell watches the orders mailbox, reads each purchase order — PDF, scan, photo or plain text — matches products and prices against your database, enters the order, and confirms back to the customer with the order number. Unknown items and odd prices pause for a human. Nothing ever leaves your office.

Watch the demo

starts when: A purchase order lands in your orders@ mailbox

The problem

The order intake desk is a translation bureau. Customers send purchase orders however their own systems produce them — a generated PDF, an Excel sheet, a photographed printout with quantities circled in pen — and someone on your side retypes every line into your system. Line by line, code by code, all day.

It’s the purest case of the 62% of working time that goes to repetitive tasks: zero judgment, total attention. And attention slips — a 100 becomes a 1,000, the -B variant ships instead of the -A, and unwinding one wrong delivery eats the margin of ten right ones. Meanwhile the customer waits a day just to hear “order received”.

How your Aivell does it

The order is read the moment it arrives. Attachment or email body, typed or photographed — the Documents & Vision tool reads it onboard, and pulls customer, lines, quantities and dates into structure.
Every line is checked against your catalog. Product codes are matched first, free-text descriptions second, and quoted prices are compared with your price list. Discrepancies surface now, not at the loading dock.
Doubt stops the line — automatically. An unknown item, a price that disagrees, a total above your threshold, a customer on credit hold: the chore says pause, so the order waits for a yes. You approve from your phone; it resumes.
The clean ones go straight in. Order rows land in your database, the email is marked processed, and a guardrail keeps the tool strictly append-only — it can create orders, never rewrite prices or delete history.
The customer hears back immediately. Order number, delivery estimate, their language, minutes after sending. The confirmation your best clerk sends on a slow day — on every order, on every day.

Your thresholds, your escalation names, your definition of suspicious: all of it sits in eight lines of plain language you can edit over coffee. The guardrails make sure that whatever the mail contains, only clean, approved orders ever reach your system.

The chore, as you'd write it.

No flowcharts, no code — a chore is just a message to your Aivell, in your own words. It shapes it into a solid, guarded procedure and follows it to the letter. This one:

Giulia Aivell · enter orders from email ● active

watch orders@ for new mail; an order can be a PDF, a photo, or just text in the body

read the attachment onboard; pull customer, items, quantities and requested dates

match every line against the product database — codes first, descriptions second

unknown product, or a price that disagrees with the list? pause and ask Davide

orders above €5,000, and any customer on credit hold, wait for approval

insert the order rows and mark the source email as processed

confirm back with the order number and delivery estimate, in the customer's language

unreadable attachment? ask the customer for a resend, and put Davide in cc

You

Got it — I'll start right away, and check with you whenever something needs a human.

Giulia Aivell

The tools it uses.

Connected with single-click passkeys — no copied tokens, no OAuth hell.

Email watches orders@, confirms back to customers

Documents & Vision reads order PDFs, scans and photos onboard

Database inserts order rows, looks up products and prices

Sensible guardrails.

Every tool can be limited to exactly what this chore needs. For this one, you might set:

Database Can insert order rows and read products and prices. Updating prices or deleting anything is blocked.

Email Confirmations go only into the ordering customer's thread; mail to anyone else needs approval first.

Questions, answered.

Our customers send orders in every imaginable format. Really all of them? +

If a human could read it, yes — a tidy PDF, a spreadsheet, a scan of a printout, a phone photo of a handwritten list. The vision model runs onboard the box, so the crumpled photo gets the same treatment, and the same privacy, as the clean PDF.

What if a product code is wrong or discontinued? +

It tries the description against your catalog; if the match isn't certain, the order pauses and a human decides. The chore's standing rule is simple — nothing ambiguous ever gets written into the system, because a wrong shipment costs more than a five-minute pause.

Do our order flows and price lists touch any cloud service? +

No. Reading, matching and database writes all happen on the box, inside your network. Who buys what from you, at which negotiated price, is commercial information — it stays in the building.

Can it enter orders into our management software rather than a bare database? +

If the software exposes its database or an import file, your Aivell writes there directly. If the only door is the web screen your clerks use, the Browser Agent clicks through that screen the same way they do.

Related.

Invoice data extraction, without the typing

An on-premise AI colleague reads every supplier invoice — PDF, scan or photo — and books it into your sheet or database. Invoices never leave your office.

documents and vision read →

Quote requests collected, priced and drafted

Specs gathered in chat or email, prices looked up in your own database, a draft quote ready for approval — margins never leave the office.

chat, leads and customer service read →

Cross-system sync without the middleware

A webhook fires in one system, your Aivell writes the record correctly into the other — on-premise glue between your tools, with guardrails on every insert.

data, spreadsheets and reporting read →

Aivells for manufacturers

An on-premise AI colleague for manufacturing back offices: turns order emails into system entries, works supplier portals, watches stock, reads shop-floor photos.

industry read →

What is an Aivell?

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.