Ask ChatGPT about QuickBooks customers
Load QuickBooks into your data warehouse first, join customers to what they were billed and credited, then ask ChatGPT who owes what. ChatGPT reads a Data Mart you published through OWOX, not the QuickBooks account.
%20scale(1.25)%22%20fill%3D%22%23fff%22%20d%3D%22M12%200A12%2012%200%200%200%200%2012a12%2012%200%200%200%2012%2012%2012%2012%200%200%200%2012-12A12%2012%200%200%200%2012%200zm.642%204.1335c.9554%200%201.7296.776%201.7296%201.7332v9.0667h1.6c1.614%200%202.9275-1.3156%202.9275-2.933%200-1.6173-1.3136-2.9333-2.9276-2.9333h-.6654V7.3334h.6654c2.5722%200%204.6577%202.0897%204.6577%204.667%200%202.5774-2.0855%204.6666-4.6577%204.6666H12.642zM7.9837%207.333h3.3291v12.533c-.9555%200-1.73-.7759-1.73-1.7332V9.0662H7.9837c-1.6146%200-2.9277%201.316-2.9277%202.9334%200%201.6175%201.3131%202.9333%202.9277%202.9333h.6654v1.7332h-.6654c-2.5725%200-4.6577-2.0892-4.6577-4.6665%200-2.5771%202.0852-4.6666%204.6577-4.6666Z%22%2F%3E%0A%3C%2Fsvg%3E%0A)
QuickBooksConnectedA/P Aging Summar…A/R Aging Summar…AccountsBalance Sheet re…
ChatGPTRealmReportReportDaily 09:00QuickBooks Endpoints
26 QuickBooks endpoints you can import. Each lands in its own table in your data warehouse.
A/P Aging Summary report
A/P Aging Summary as of report_date (window end): one row per vendor with aging buckets.
11 fieldsTime seriesA/R Aging Summary report
A/R Aging Summary as of report_date (window end): one row per customer with Current, 1-30, 31-60, 61-90, 91+ and Total columns.
11 fieldsTime seriesAccounts
Chart of accounts, active and inactive.
20 fieldsBalance Sheet report
Balance Sheet as of the run window end (Assets, Liabilities, Equity by account).
12 fieldsTime seriesBill payments
Payments against vendor bills; joins to Bill through Line[].LinkedTxn.
23 fieldsBills
Vendor bills (payables).
25 fieldsCash Flow report
Statement of Cash Flows for the run window.
12 fieldsTime seriesCompany info
One row per company: names, country, fiscal-year start.
15 fieldsCredit memos
Credits issued to customers (reduce receivables; negative revenue in reconciliations).
23 fieldsCustomers
Customers, including inactive ones and sub-customers, with open balance and a flattened billing address.
28 fieldsDeposits
Bank deposits, including the ones that group undeposited customer payments.
21 fieldsGeneral Ledger Detail report
General Ledger: every posting line per account for the run window, with opening balance and totals.
12 fieldsTime seriesInvoice changes (CDC, incl. deletes)
Change data capture for Invoice: every invoice created, updated or deleted since the run window start, full payload, deleted ones with status = Deleted.
9 fieldsTime seriesInvoices
Invoices with header amounts, customer and Line[] as JSON (SalesItemLineDetail.ItemRef, Qty, UnitPrice, Amount).
36 fieldsItems
Products and services with price, purchase cost and the three account links.
25 fieldsJournal entries
Manual and adjusting journal entries; the postings the transaction tables miss.
15 fieldsPayments
Customer payments.
21 fieldsPreferences
One row per company: home currency, multi-currency flag, fiscal-year and tracking settings.
17 fieldsProfit and Loss report
Profit and Loss per run window, one column per month (summarize_column_by=Month), rows nested by account group.
12 fieldsTime seriesPurchases
Expenses paid directly (cash, check, card) without a bill.
26 fieldsRefund receipts
Money refunded to customers.
26 fieldsSales by Customer report
Sales by Customer Summary for the run window.
12 fieldsTime seriesSales receipts
Sales paid at the time of sale (no receivable).
27 fieldsTransfers
Money moved between balance-sheet accounts.
16 fieldsVendor Expenses report
Expenses by Vendor for the run window.
12 fieldsTime seriesVendors
Vendors, including inactive ones, with open payable balance.
19 fields
Why ChatGPT needs your data warehouse, not the QuickBooks account
ChatGPT reading QuickBooks directly sees one customer record at a time. A balance only means something beside the invoices and credit memos behind it, and those are linked in a warehouse, where the data analyst has already joined them on the customer.
- A customer record holds one open balance, not the activity that made it
- The data analyst joins invoices to credit memos on customer_id
- ChatGPT answers from a published Data Mart, with the same definitions every time
QuickBooksCustomers, balanceChatGPT
Data warehouseJoined by idStep 1: Load QuickBooks into your data warehouse first
Create a Data Mart with QuickBooks as the source and choose where the data lands. The QuickBooks connector is a custom connector built in OWOX Data Marts, and the data goes into any popular data warehouse.
- Customers is one of 26 endpoints, each in its own table
- Authorize QuickBooks Online with OAuth and give the company (realm) id
- Open the QuickBooks connector for your data warehouse below
QuickBooksData MartGoogle BigQuery
AWS Athena
Snowflake
AWS Redshift
Databricksquickbooks.customerStep 2: Link QuickBooks customers to invoices and credit memos
In your data warehouse the model joins Invoices and Credit memos on customer_id, so a question about a customer's balance can be answered from what was billed and what was credited.
- Invoices and credit memos join on customer_id
- A customer's balance sits beside the invoices and credits that make it up
- Definitions are written once, so the answer does not drift
Step 3: Connect ChatGPT to your QuickBooks data through OWOX
In ChatGPT, open Plugins, choose Install plugin, find OWOX Data Marts and sign in. ChatGPT then sees the Data Marts you published, and only those.
- Only published Data Marts are visible, nothing else in your data warehouse
- The Data Mart you published is the one it reads
ChatGPTConnectorsOWOX MCPAddConnectedStep 4: Ask ChatGPT about who owes you money
Ask in plain words, for example which customers carry the largest balance. ChatGPT finds the Data Mart that answers it, runs the query and reads the result from your data warehouse.
- The answer reads Customers in your data warehouse, not the QuickBooks API
- Balances can be read beside invoices and credit memos
ChatGPTShow balance by display_nameget_relevant_data_marts_by_promptquery_data_martdisplay_namebalanceactiveStep 5: Reuse QuickBooks customer answers as files
Optional: in one click, export the answer or its source data to Google Sheets, CSV or XLS to check it or share it.
- The export carries the same numbers as the answer
- Share it with a colleague who does not use ChatGPT
Google SheetsCSVXLSOne click
- 0.41.0Dates from Google Sheets arrive as dates, not text
- 0.41.0Rotating a connector key is one change: Sources share reusable Credentials
- 0.40.0Connecting a plan from Google Sheets takes fewer guesses
- 0.39.0Reload a year of history in one backfill, not twelve
- 0.39.0Long Facebook Ads imports no longer fail on a corrupted response
Why OWOX for QuickBooks in ChatGPT
Letting ChatGPT call the QuickBooks API gets one record into a chat. OWOX also keeps the data yours, joined and defined once, so the answer is one you can check.
- You own the data. QuickBooks lands in your own data warehouse, in tables you control.
- Balances with their cause. Customer activity joins on customer_id, so ChatGPT can say why a balance is what it is.
- Every endpoint available. 26 endpoints land as tables, from customers to report snapshots.
- Answers you can trace. You can see which Data Mart an answer came from, and one click exports its source data.
- Versioned, with rollback. The manifest is open in the connector builder, and every change is a saved version you can roll back.
You are booked. The invitation is on its way to , with the calendar invite and the meeting link.