An Aivell sits in your accounting firm and does the clerical layer — reads every invoice and receipt onboard, chases clients for missing documents before deadlines, sends payment reminders, keeps the tax calendar. Inference runs on a box behind your firewall, so client financials never touch anyone's cloud.
Clients hand over a year of receipts as a folder of crumpled paper and phone photos — and someone junior loses a week typing it into the ledger.
Bank statements, signed engagement letters, VAT registers, payroll summaries — every filing season starts with three rounds of "gentle reminder" emails per client.
VAT returns, withholding payments, annual accounts, estimated-tax instalments — dozens of deadlines per client, each with a penalty attached and each depending on documents the client hasn't sent.
Their receivables become your problem — clients expect the firm to nudge their debtors, and your own fee notes go unpaid while you do it.
The automation you'd love exists — in the cloud. Uploading client ledgers and payslips to a third-party service is a conversation you'd rather not have.
An accounting firm sells judgment, but most of its hours go to transport: moving numbers from a PDF into the ledger, moving documents from a client’s inbox to the file, moving the same reminder from a template to twelve clients. Employees across industries spend 62% of their time on repetitive tasks; in a firm at filing season it feels like more.
The rituals are always the same. The shoebox of receipts, photographed badly. The client who swears the bank statement was sent last week. The engagement letter signed by nobody. The tax calendar with forty deadlines a quarter, each one gated on a document you don’t have yet. And your own fee notes, chased last because client work comes first.
The obvious fix — cloud automation — fails at the first question a partner asks: where does the client data go? An Aivell answers it by construction. It is a box in your office; the models run on it; a client’s payroll file is read a metre from where it is stored, and nothing leaves your network.
So the boring layer gets done, your way. Invoices and receipts are read onboard — scans, photos, handwriting — checked and booked. The document checklist is compared to reality and clients get chased, politely, on your schedule, in their language. Deadlines produce reminders before they produce penalties. Fee notes get dunned without you having to feel awkward about it.
Every chore is written in plain language, like a briefing to a new junior: which clients get the gentle tone, what amount needs a partner’s yes, when to stop chasing and escalate. Guardrails make the limits physical — the Aivell can append to the ledger but not rewrite it, email known clients but nobody else. It works through the night before the deadline. It just never leaves the office.
on the 1st of each month, read the deadlines sheet and list every client with a filing due within 45 days
check each client's folder on the share against their document checklist
email each client a list of what is still missing, in their language
one polite reminder after 7 days, a firmer one after 14 — never more than three
clients marked "handle with care" get a draft for Marta to approve instead
anything still missing 10 days before the deadline goes on Marta's desk as urgent
log every request and reply in the chasing sheet
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.
An on-premise AI colleague requests, tracks and chases client certificates ahead of the deadline — every send logged, nothing processed in the cloud.
Overdue invoices from your own database become polite, escalating reminders — sent by an on-premise AI colleague and stopped the moment payment lands.
Employees snap a receipt and email it. A private AI colleague reads the photo on the box, categorizes the spend and books it — the image never leaves your network.
The monthly close as an orchestrated checklist — figures pulled, reconciliations checked, owners chased — by an AI colleague that keeps the books onboard.
Incoming paperwork classified, renamed and filed to the correct folder by an AI colleague that reads each page inside your office — no document uploaded anywhere.
An on-premise AI colleague chases unanswered emails and missing client documents — politely, on a schedule, stopping the moment the reply arrives.
On the box, in your office. Reading, extraction and reasoning all run onboard — a client's ledger, payslip or bank statement never crosses your firewall. That is on-premise processing by design, which is also the easy answer when a client's DPO asks where their data went.
Yes, structurally. You remain the only holder of client data — there is no cloud processor to add to your register or your engagement letters. Processing stays local, which is privacy by design in the GDPR Article 25 sense, and your professional secrecy duty never leaves your hands.
If your package exposes a database or an import format, the Aivell writes to it directly. If it is web-only, the Browser Agent fills the forms exactly as a junior would — same screens, same fields, with guardrails on what it may touch.
It classifies the way you brief it to — the chore is plain language, so your firm's rules go in as written. Anything ambiguous is not guessed; it lands in a pile for a human with a note on why.
Guardrails, enforced on the Email tool itself. You can restrict it to reply-only, to known client addresses, or require approval for anything new. These are hard limits on the tool, not instructions it might forget.
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