The engine
Ask for a number. Get the number, the source and the date.
Pacioli is Acerbo.AI's finance engine. It connects, read-only, to the systems you already run — ERP, bank, CRM, time tracking, Drive — and answers the questions that currently cost a day of work: what's the margin on this project, when does cash run out, which contracts expire in the next 90 days, which of the three files is the right one. No data warehouse to build first, no migration: the engine works on the files and exports your team already produces.
The playground below runs on Alba Servizi S.r.l., a fictitious company, synthetic data as of 4 September 2026.
Try the playground
Ask Pacioli a question about Alba Servizi S.r.l.'s data. Every answer follows the four-part format: Number, Source, Cut-off, Gap.
Alba Servizi S.r.l. · fictitious company · synthetic data as of 4 September 2026 · read-only
Pick a question above or write your own.
Three promises
It reads. It doesn't write.
Read-only access to every source system. It never touches the ledger, edits a file or sends a payment. Your team stays the only one making decisions.
Every number has a pedigree.
Every answer comes with its source, its last-updated date and what's missing. When two files disagree, it tells you which one wins and why.
It knows what it doesn't know.
If the data isn't there, it says so. It doesn't fill gaps or estimate when it should measure. That's the difference between an assistant and a finance tool.
In the tool you already use
You ask ChatGPT or Claude. The engine answers.
Pacioli plugs into the AI assistants that support the MCP standard — Claude, ChatGPT, Microsoft Copilot and others. Open the chat you already use, ask in plain language, and the answer comes from your own data, with source and date. Nothing to learn, no new dashboard. Try the groups already live in the playground above.
Monday morning — cash
Before the board — margins
Month-end — reconciliations
coming soon to the playground- Which version of the June 30 trial balance is final, and where is it?
- Reporting and statutory accounts show two different revenues for the Consulting BU: which prevails and why?
- List August bank movements with no matching entry in the cash book.
- Before I send the board pack, compare today's file with the one approved yesterday.
Sales oversight — renewals
coming soon to the playground- Which contracts expire in the next 90 days and what are they worth?
- Which 2025 customers haven't billed anything in 2026?
- Current recurring base vs twelve months ago, by product line.
- Which annual licences need renewing by year-end, and who owns them?
People — hours and utilisation
coming soon to the playground- How many hours did the team log in July, and on which projects?
- Team utilisation vs capacity, last three months, by person.
- Which tasks exceeded their hour budget, and by how much?
- What do internal projects cost us compared to billable ones?
Tax and debt
Every answer comes with number, source, cut-off and gaps. When the question is a decision — "should we renegotiate?", "should we hire?" — the engine prepares the numbers and stops. The rest is yours.
Every answer has four parts. Always.
A number without a source is an opinion. Pacioli answers in the same format every time, so your CFO can take it to the board without redoing the maths.
Question — What's the margin on project PD-0412 and how much does it weigh on the Consulting BU?
Number — Budget margin 31.4% (revenue €86,000, cost €59,000, EBITDA €27,000). That's 25.0% of the Consulting BU's budgeted revenue.
Source — Budget cost PD-0412 v3.xlsx, updated 28 Aug 2026.
Cut-off — Data as of 4 September 2026 (the demo dataset's cut-off).
Gap — August timesheets not yet available: budget margin, not actual. After overhead allocated by revenue share (€20,312) the fully-loaded margin drops to 7.8%.
Number · Source · Cut-off · Gap. If one of the four is missing, the answer isn't finished.
One tool for every question your CFO gets on Friday at 6pm.
Pacioli is a set of specialised tools, one per process. We switch them on one at a time, starting with the one that hurts most. The demo playground has the eight tools highlighted above live.
Cash flow & forecast
Cash plan
anchored to the real bank balance, not the mechanical sum of due dates. 13-week requirement, instalment sustainability, coverage from signed contracts.
Treasury dashboard
balances by bank and by entity, expected receipts, outflows, monthly snapshots. From the cash book, daily.
Revenue forecast
by business unit, line, project. Separates earned revenue from pass-through, so turnover doesn't inflate margin.
Renewals & expiries
recurring base, contract calendar, customers lost year on year, licences about to lapse. Churn before it's churn.
Reporting & variance
Project margin
revenue, cost, margin per project, read from the budgets your team already fills in.
Fully-loaded margin
opens the overhead "bucket", separates technical costs from G&A, allocates them to business units. Find out which lines are actually profitable.
Hours & cost vs budget
staff allocation, utilisation, billable vs non-billable, overrun tasks. From time tracking, live or from exports.
Close & reconciliations
Source registry
the map of where statements, document lines, ledger movements and cash book actually live. And the rule for which one prevails when they disagree.
Actual payments
"has invoice X been paid?" answered from the reconciled bank statement, not from a two-month-old ledger.
Release check
before a report goes out, an automatic comparison of the delivered file against the approved one. Cell by cell.
AP/AR & working capital
Receivables & payables schedule
expected receipts and payments by week, exposure by counterparty, tax instalments falling due.
Dunning
who's late, by how much, with what history. Ready for sales, not for legal.
Tax debt
instalment plans rebuilt from actual payments, with what's left and when.
For M&A
Number consistency
dozens of rules that check that memorandum, deck, business plan and statements tell the same story before a buyer sees them.
Data room
release check on every file shared with a counterparty. Integrates with Gordon and Sherlock, Acerbo.AI's M&A OS.
Finance-grade isn't an adjective. It's a list of things the engine can't do.
We built Pacioli for finance teams who have to sign what they present. So the constraints live in the engine, not in the instructions.
Read-only, by architecture.
ERP, CRM, bank, time tracking, Drive: read-only connections. There is no write path.
No autonomous critical decisions.
The engine prepares: proposed reconciliations, flagged exceptions, calculated scenarios. A person approves. Every time.
Full traceability.
Every output carries its source, file version and cut-off date. Every answer is reproducible: same question, same data, same number.
Declared source hierarchy.
When the statements and the management report don't match, the engine doesn't average them: it applies the precedence rule agreed with you and tells you it did.
Release check executed, not recommended.
Before every delivery, an automatic comparison between the approved file and the outgoing one. Verification is a command that runs, not a reminder that gets forgotten.
Your data stays yours.
No training on customer data. Deployment inside your perimeter (your cloud or a dedicated machine) or tenant-isolated SaaS — your choice.
Separation of duties.
Whoever configures the engine doesn't approve the outputs. Whoever approves doesn't post the entries.
Method
- 1
30 days — Diagnosis
We map processes, systems and data sources.
Pacioli — With Pacioli: we build the source registry and switch on a single tool for the most painful process, usually cash. By day 30 you already have your first answer with number, source and cut-off.
- 2
90 days — Pilot
We run the first process end-to-end with controls and human oversight.
Pacioli — With Pacioli: three or four tools on the pilot process. We measure cycle time, exceptions, accuracy. Your team uses the engine every week.
- 3
6-12 months — Scale
We extend the method to the next processes, with stable governance.
Pacioli — With Pacioli: the engine becomes the finance data core. Every recurring question has a tool; every report has a release check.
From the field
Six sources, two clouds, one number.
A B2B services group with several entities and project-based work. Statements and reporting on one cloud, project budgets and cash book on another. Three versions of the same trial balance, none labelled final. A cash plan built by adding up due dates, while the real bank balance said something else.
- 1
Source registry.
Where every document lives and which one prevails. On its own it closed half the recurring questions: they were questions about where, not how much.
- 2
Bank-anchored cash plan.
Real balance as the starting point, tax instalments rebuilt from actual payments, signed contracts as coverage. For the first time the day cash ran out was a date, not a range.
- 3
Fully-loaded margin by business unit.
Overhead parked on a dummy project was reclassified and allocated. Two lines that looked profitable weren't; one that looked marginal was carrying the group.
- 4
Release check.
After a manual review let a file through that differed from the approved one, the comparison became a mandatory command before every delivery.
What changed: questions about numbers stopped triggering a file hunt and started getting an answer with a source and a date. The decisions stayed with people. That's the point.
With the CFO. Or on its own.
Pacioli is included in Acerbo.AI's CFO services. If you already have a finance team and only need the engine, you get it as SaaS.
With the CFO service
The engine comes with the people who configure it, govern it and sign the outputs. For SMEs that want an operational, AI-augmented CFO.
Book a call →Pacioli SaaS
The engine for your own finance team. Read-only connectors to your systems, tools switched on one at a time, isolated tenant. 30-day onboarding with the source registry included.
Request a demoFor firms and partners
Accountants, fractional CFOs, system integrators: an engine to bring to your own clients, with your name on the output.
Let's talk →Priced by active tools. We tell you at the demo, not on the third call.
Request early access
Leave your details and we'll reach out for a Pacioli SaaS demo tailored to your case.
FAQ
Is this another piece of software to install?
No. Pacioli reads the systems and files you already have: ERP exports, bank statements, CRM, time tracking, Google Drive or Dropbox. No migration, no new ERP.
What does "MCP" mean?
It's the open standard AI models use to connect to tools and data (Model Context Protocol). In practice: Pacioli works with the AI assistants your team already uses, and every tool is a defined command with controlled inputs and outputs — not a free-form conversation.
Can it be wrong?
It can read a stale or missing figure: that's why every answer carries a cut-off and its gaps, and why it doesn't decide. It never invents numbers: if the data isn't there, it says so.
Where does my data go?
It stays inside your perimeter (on-premise or your cloud) or in an isolated SaaS tenant. It is never used to train models.
How long to get started?
The first tool is live in 30 days, source registry included. Further tools take weeks, not months.
Do I need a data warehouse first?
No. That's why Pacioli exists: most SMEs have the data, they just haven't organised it. The source registry does the job a warehouse would, without building one.
Does it work with my accountant?
Yes, and they usually like it: they get sharper questions and files that are already reconciled. It doesn't replace the accounting function; it feeds it.
Which ERPs?
Any system that exports to CSV or Excel, plus direct connectors for the main CRMs, time-tracking tools and cloud storage. If yours exports, Pacioli reads it.