Use case #14 · CPA
CPA prototypes vendor spend reconciliation in Claude Cowork, then moves it to a private cloud for client data
- Close/bookkeeping
- Intermediate
- Claude Cowork, Claude on AWS Bedrock (custom app), QuickBooks, OneDrive and SharePoint connectors
- Source checked · state CPA society blog
The short version
What they did
Automated the month-end vendor spend reconciliation: match invoices against contracts, flag variances and build a summary. Proved the workflow in Claude Cowork on sample data, then rebuilt it as a custom application calling Claude through AWS Bedrock for client data.
Result (as claimed)
“The setup took 15 minutes.” “Using sample vendor data (273 invoices, 14 contracts), Cowork extracted terms from a new contract PDF, compared spend against contract values, flagged three variances and built a reusable dashboard.”
Prompt
I’m an accountant testing a month-end vendor reconciliation in Claude Cowork with sample data. I’ll attach [invoice CSV] and [contract PDFs] for [month]. Compare invoiced spend with the terms for each vendor.
1. Extract rates, spending caps and contract dates, citing the page and passage for each term.
2. Total the month’s invoices by vendor and compare them with the applicable terms. Show the calculation and invoice numbers behind each possible variance.
3. Return a findings table and a one-page summary for the close file, written in plain English.List missing contracts, unreadable terms and questions separately. Ask about unclear billing periods or units instead of guessing. Don’t treat a monthly total as exceeding a contract cap without checking the cap’s period. Don’t change invoices, contact vendors or describe a flagged variance as confirmed overbilling.
Do it yourself
These steps are my suggested way in, not Sharmila Damarapati’s setup.
Level: Intermediate with an agent like Claude Cowork. Building a custom app on a cloud platform is Advanced.
What you need
- Sample or anonymized vendor invoices as a CSV and two or three contract PDFs.
- A chat assistant or agent that reads files, such as Claude Cowork.
- A vendor reconciliation you already did by hand, to compare.
Start simple
- Build a small sample set: a few vendors, one contract each, a month of invoices. Plant one overbilling and one expired contract.
- Paste the case’s Prompt with the files.
- Check the extracted contract terms against the PDFs.
- Confirm it found both planted variances, with the right invoice numbers.
Take it further
- Ask for a reusable summary layout and save the instructions so each month runs the same way.
- Move to client data only on a business plan or a private setup your firm approves, as the source did.
- Connect to your accounting system’s exports so invoices arrive without manual downloads.
Check your work
- Recompute the total spend for two vendors yourself.
- Read the rate and cap terms in the contract. Extraction errors here change every variance.
- Variances can lead to vendor disputes. Confirm them with source invoices before raising them with the client.
Before you rely on it
- This is general information about how accountants use AI, not professional guidance for your engagements.
- You stay responsible for the work under your professional standards, including review and sign-off. An AI tool’s output is a draft, not evidence.
- Before you send client data to an AI tool, read the tool’s data and training terms, and get any client consent you need. For tax return information, that includes IRC section 7216.
Source checked. The link, author, date and quote were confirmed against the source. The result itself has not been reproduced. Prompt adapted for this library, inspired by the case. Business size: Firm (Damarapati LLC); size not stated. Results are quoted as the source claimed them and never improved. Spot an error or a broken link? Email charlie@cbarmorecpa.com.