home / use cases / email and inbox / email-to-ticket

From messy emails to clean tickets

An Aivell turns incoming emails into structured tickets — requester, category, priority, status — in your database or sheet. Each new message is checked against a private index first, so duplicates and follow-ups attach to the existing ticket instead of opening a fresh one. Everything runs on a box inside your office.

Watch the demo

starts when: A request lands in your support@ mailbox

The problem

Requests arrive as prose. “The heating in flat 4B is dead again, also the hallway light flickers, my number is…” — one email, two issues, zero structure. Somebody reads it, decides what it is, retypes it into a list, and answers the sender. Then the sender writes again the next morning, a colleague files it as a new item, and now two people are chasing the same boiler.

That translation work — from paragraph to record — is quietly enormous. Administrative overhead of this kind costs firms around $17,000 per employee per year, and the duplicates it produces cost more: double dispatches, contradictory status updates, and requesters who conclude nobody is listening.

How your Aivell does it

  1. The email is read and understood on arrival. Requester, location, what is actually broken — extracted from prose, however rambling, in whatever language.
  2. The index is checked before anything is created. Every past ticket lives in an onboard index. A follow-up or a re-send attaches to the open ticket with its history intact; deduplication is the first step, not an afterthought.
  3. A clean row lands in your system. Category, priority, plain-language summary, status “open”. A guardrail limits the Database tool to inserting tickets and updating status — it cannot delete or wander into other tables.
  4. Urgency follows your written rules. No heating, water, anything hinting at injury — priority high and a direct alert to the colleague you named. The rules are sentences in the chore, not settings in a menu.
  5. The requester gets a number, humans get exceptions. A confirmation with the ticket number goes back in the sender’s language. Closing a ticket requires a human “done” — your Aivell tracks, it doesn’t declare victory.

The result is a request list that is actually true: one ticket per problem, statuses that mean something, and a full audit trail of who reported what and when — none of it ever leaving your building.

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 · turn mail into tickets ● active

read every new message in support@; skip newsletters and out-of-office replies

check the ticket index first — is this a follow-up or a duplicate of something open?

if it is, attach the message to the existing ticket; never open a second one

new issue: create a ticket with requester, building, category and a plain summary

heating, water or electrical failures are priority high — and email Stefan directly

anything mentioning injury or damage is priority high too, whatever the wording

reply with the ticket number, in the sender's language

index the new ticket so tomorrow's duplicates can find it

mark a ticket closed only when the assigned fixer replies "done"

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 reads requests, sends the ticket number back
Knowledge Base indexes tickets so duplicates are caught
Database one row per ticket, status tracked

Sensible guardrails.

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

Database Can insert tickets and update their status column. Deleting rows or touching other tables is blocked.
Email Replies stay in the requester's thread; new outbound mail only to addresses on your own domain.

Questions, answered.

How does it know a new email belongs to an existing ticket? +

Every ticket is indexed in a knowledge base on the box. A new message is compared against that index — same sender, same broken boiler, same rough wording — before anything is created. Matches attach to the open ticket; only genuinely new issues get a number.

We don't have a ticketing system — is a spreadsheet enough? +

Yes. The Aivell writes wherever your rows live — a proper database, or a shared spreadsheet with one line per ticket. You get numbering, status and deduplication without buying and learning a helpdesk product.

Requests contain names, addresses and photos of people's homes. Where does that go? +

Nowhere. Reading, matching and ticket-writing all happen onboard, behind your firewall. Nothing is uploaded to an AI provider, which is exactly what you want when the mail describes someone's flat.

Who decides what counts as urgent? +

You do, in writing. The chore lists the words and situations that mean priority high — no heat, water leak, injury — and who gets alerted. Changing the escalation rule is editing one line, and the audit trail shows every decision made under it.

Related.

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.

Watch the demo

onboard inference · air-gapped · fixed monthly fee