Accounting firms
Chat with your clients' spreadsheets and documents, without uploading them to ChatGPT
An assistant connected to your firm's folders answers questions about a client's spreadsheet or document and transforms it as you ask: match, filter, group or summarise. It does the maths by running operations on the file, tells you where each figure comes from, and nothing goes through personal accounts.
Amud team · Reviewed on October 29, 2026 · 3 min read
In an accounting firm, many questions are answered by opening a spreadsheet: the general ledger a client sends, the bank statement, the quarter's list of invoices. Which suppliers went up the most this year? Which invoices have the wrong VAT? Which payments in the statement have no invoice? Each answer means filters, pivot tables and formulas, and half an hour from someone who knows how to build them.
It's work that lends itself to delegation: the routine is clear and the question fits in one sentence.
What you can ask
- "In this client's sales spreadsheet, who are the ten customers who billed most this year, and how does that compare with last year?"
- "Match the bank statement against the list of issued invoices and tell me which payments have no invoice."
- "Of these purchase invoices, which have a VAT rate that doesn't fit the item?"
- "Summarise this lease: parties, term, rent and indexation clauses."
- "Group this quarter's expenses by supplier and give me a spreadsheet with the summary."
How it works
- 01
Pick the client and the file
Or just name it: the assistant finds it in the client's folder.
- 02
Ask in plain language
As you'd ask a colleague, with no formulas or special instructions.
- 03
It calculates on the file
Figures come from running operations on the data, not from what the model estimates.
- 04
It shows the result and where it came from
Which columns, filters and operations it used, so you can check at a glance.
- 05
For a transformation, it creates a new file
A spreadsheet or CSV ready to use or load into your software, with the original untouched.
Why not just paste it into ChatGPT
- Client data ends up in a personal account, with no data processing agreement. We explain this in is it safe to use AI with my clients' data?, which applies just as much to an accounting firm.
- Figures may come "from memory". A language model mentally adding hundreds of rows makes mistakes. It has to be made to calculate on the file and show how it did it.
- It doesn't know your folders. Every time, someone has to find the file, upload it and explain whose it is.
For the full comparison, see what a digital employee adds compared with ChatGPT.
Where the data lives
The assistant works inside the firm's own accounts: your folders, your email and your permissions. It uses models through business services that don't train on the data, with EU AI providers and pseudonymisation when needed. Every query is logged with who made it. If the firm prefers nothing to leave its network, it can also run on an in-house AI server, knowing what you gain and what you give up.
What it doesn't do
- It doesn't replace the adviser's judgement: it answers and prepares, and the adviser decides what to do with the result.
- It doesn't send anything to the client on its own.
- It doesn't change the original files.
Where to start
In a free assessment we look at which questions you most often ask about clients' files, where those files live and which software you use. From that, we tell you how we'd set it up, at a fixed price with no lock-in.
Frequently asked questions
Which file types does it work with?
Spreadsheets (Excel, CSV), PDF and Word documents, and exports from your accounting software. If a file is very messy, it says so before giving a result.
Does it change the client's files?
No. When you ask for a transformation, it creates a new file and leaves the original as it was.
What if it gets a figure wrong?
Every result comes with how it was obtained: which columns, filters and operations. That makes it quick to check, just as you'd check a colleague's work.
Can it query the accounting software directly?
If the software has a stable way to get data out (an API or scheduled exports), yes. If not, it works with the exports you leave in the client's folder.
Keep reading
Industry
Automation and AI for accounting firms: invoices, client documents and deadlines
In an accounting firm, every quarter brings the same race: asking for invoices, waiting for them, posting them by hand and balancing before the deadline. We automate that chain on top of your accounting software, with a digital employee that collects, reads, checks and chases so your team spends its time advising.
Question
Is it safe to use artificial intelligence with my clients' data in a law firm?
Yes, if the tool is chosen for it: a provider with a data processing agreement that doesn't train on your data, data processed in the EU, limited and logged access, and a lawyer who reviews everything before it leaves the firm. A free chat doesn't meet that bar.
Question
We already use ChatGPT at work: what does a digital employee add?
ChatGPT helps whoever uses it, when they ask. A digital employee does a whole task when work arrives: it reads the email, checks your software, prepares the reply and asks for approval before sending it. They don't compete: most companies use both.
Blog
A local AI server for your business: what it can really do and where it falls short
Running an AI model on your own server is easier and cheaper than it was two years ago, and for many tasks it is enough. But it isn't ChatGPT or Claude in your office. We cover the hardware you need, what it is good for, where the gap with frontier models shows and when it pays off.
Which task would you like off your plate?
Tell us how your team works. In a free 30-minute session we'll tell you what can be automated, how much you'd save and what isn't worth it.
Book a free meeting