For an accountant, Uber lines are rarely a pure category question. The same merchant carries business travel, personal rides and food orders, and the right coding depends on what the client tells you. An AI assistant can sweep the whole list in one pass and prepare the grouping, but the judgement stays with you.
Why this task suits an assistant and why it does not
What suits it: finding every row that matches a merchant across months of imported statements, summing them, and presenting them in groups. What does not suit it: deciding purpose. A ride to a client site and a ride home from a family dinner look identical in a statement. Use the assistant as a fast sorter and keep the purpose decision on your side.
A working method
- Confirm the client's statements are all imported and the right account was selected for each one. Each transaction keeps its account, which matters when a client has a personal and a business card.
- Connect the assistant read-only first. The connection lets you pick the scope and disconnect at any time, and for client data that is the right default.
- Ask for all rows with Uber in the narration, grouped by month, with account, amount and any receipt attached.
- Send the list back to the client with a short question for each doubtful row, or work from their receipts.
- Apply your categories, then add a rule so the following months arrive pre-tagged.
- Record what you changed and why, in your working papers.
A worked example with made-up figures
A client's year-to-date records show 22 Uber-related lines totalling ₹6,840. After the client confirms purposes, 15 lines come to ₹4,920 as business travel and 7 lines come to ₹1,920 as personal. Check: 4,920 + 1,920 = 6,840. The totals tie, so nothing was dropped.
| Group | Lines | Amount | Treatment to decide with client |
|---|---|---|---|
| Business rides | 15 | ₹4,920 | Travel, with receipts kept |
| Personal rides | 7 | ₹1,920 | Owner drawings or personal, per the client's setup |
| All Uber lines | 22 | ₹6,840 | Reconciles to the statement rows |
The tax treatment of each group is for you and the client to settle under the rules that apply to them; those change, so confirm the current position with the official source. Categories in the app can carry GST rate, ITC-eligible and tax-deductible settings, and its reports estimate GST inside spending, but those are estimates to verify against invoices.
How do I keep an audit trail?
- Keep receipts attached to their transactions rather than in a separate folder.
- Export the all-transactions file before and after the bulk change and keep both.
- Make changes in small batches so a wrong batch is easy to isolate.
- Note any rows you reclassified after client feedback.
- Remember deleted transactions go to a recycle bin for 30 days, which helps with accidents but is not an archive.
What about client confidentiality?
Statements contain names, account numbers and balances. SpendixAI's own AI categoriser, when switched on, uses a normalised merchant name and an amount band rather than your name, account numbers or full statement text, and it can be turned off per feature. A separate assistant connected by MCP sees what the scope you granted allows, so grant the narrowest scope, tell the client you use one, and disconnect when the job ends.
Closing checklist
- All statements imported, none duplicated.
- Row count and total reconcile to your own sum.
- Purpose confirmed with the client for business versus personal.
- Rule added for the next period.
- Scope reduced or the connection removed once finished.
Useful ideas start with a clearer picture of your own spending.
Explore your reports

