Manus to enter receipt items in a household budget
kimura wanted to stop typing individual receipt items into a household account book. After building a similar process with Google Workspace Studio, the author recreated it in Manus: place a new receipt in a designated Google Drive folder, then have the agent read it and write the items into a spreadsheet.
Suggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
When a new receipt arrives in this Drive folder, read its item lines into my household spreadsheet. Preserve the paid total and identify uncertain names, discounts or tax treatment for comparison.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- kimura
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- All 58 items from 13 receipts were entered; 12 totals matched within one yen and one discount-related discrepancy remained.
- Human involvement
- kimura manually restarted the first failed trigger, processed problematic batches individually and compared every receipt total with the sheet.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Connect the designated Drive folder and destination spreadsheet.
- Trigger Manus when a new receipt arrives and enter its item lines.
- Compare paid totals receipt by receipt and retry stalled jobs individually.
kimura wanted to stop typing individual receipt items into a household account book. After building a similar process with Google Workspace Studio, the author recreated it in Manus: place a new receipt in a designated Google Drive folder, then have the agent read it and write the items into a spreadsheet.
The setup used read-only Drive access and spreadsheet writing permissions. Rather than designing a chain of separate steps, kimura supplied the reading instructions and destination sheet and let Manus organize the work.
The test used 13 actual July shopping receipts containing 58 items, from several kinds of stores. All items were entered. When kimura compared each receipt's paid total against the sheet, 12 matched within one yen. The remaining discrepancy involved a discount sticker. Manus also handled tax-exclusive supermarket prices as tax-inclusive item amounts.
The automation was imperfect. The first receipt did not trigger and needed a manual request. A batch of 12 stopped partway through, while individual receipts completed in about two minutes after starting. There was roughly a ten-minute trigger delay, and store or cashier names could be misread. The useful result was itemized household data with a clear reconciliation check, rather than an assumption that successful entry meant flawless reading.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: note.com · kimura
Frequently asked questions
How did the author check accuracy?
Each receipt's paid total was compared with the total written into the household account book.
What happened with a large batch?
Twelve receipts submitted together stopped partway through; one-at-a-time processing completed in about two minutes per receipt.
What kinds of misreads were reported?
A cashier's name could become a product name, and a store could be mistaken for another store. One discount sticker also caused a total discrepancy.
More details
About this account
Manus read receipt files and wrote itemized household account-book rows through a folder-triggered automation.
Firsthand source opened and read. Suggested prompt written by us from the reported workflow; no original task prompt is quoted.
Credit: kimura
Source published
Added to Agents In Use
Source checked
Tools and topics
Manus · Google Drive · Google Sheets
Household account book · Itemized receipts · Drive automation
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.