Listen to the article
The Japanese company’s bespoke agent will compare operational records with purchase orders, delivery notes and invoices, with implementation starting at ¥500,000 and taking at least four weeks.
Gomumari, a Japanese AI implementation company, says it will launch a bespoke data-reconciliation agent on 21 November 2026 for teams that compare operational ledgers with documents such as purchase orders, delivery notes and invoice details.
The service is aimed at production, logistics, order administration and accounting teams. Staff provide the agent with their records, including PDFs and Excel files, and it checks entries against rules agreed with the customer. Gomumari says the system will return one list of discrepancies, with explanations, and another identifying cases that need human review. Where the information is ambiguous, the agent is designed to flag the item rather than decide by guesswork.
The company says initial implementation will start at ¥500,000 and take at least four weeks. Before deployment, it will work with the customer to document the reconciliation rules and test the system against all entries in sample files. The agent can be placed in a customer’s tools or another agreed environment, according to Gomumari. Ongoing support, including adjustments when document formats or rules change, is available for a monthly fee.
Manual checking often persists because the records being compared are not prepared in the same way. Names, codes and document layouts can vary, while teams may also have to re-enter data or assemble information from several spreadsheets before reconciliation can begin. If rules for acceptable differences or split deliveries exist only in staff members’ experience, cases that require judgement can hold up approvals, shipments or month-end closing.
A survey cited in the launch announcement found that 39.8 per cent of 294 employees and business leaders at medium-sized companies who needed to combine data from multiple systems relied on exporting CSV or Excel files and compiling them manually in spreadsheet software. In the same group, 42.2 per cent identified time spent checking differences between figures across departments or systems as a problem, while 40.5 per cent pointed to the need to reconcile different field names and codes. The survey was conducted by WingArc 1st in August 2026; the findings are reported by Gomumari rather than independently verified here.
Gomumari says it has already built a purchasing-document reconciliation system for a customer’s production-management team, which began using it about two weeks after discussions to define the work. The company reports that the customer also used a related process to compare purchase orders with order registrations, and that checks against approved documents produced matching results across the tested items. These are company-reported examples, not independently assessed performance results.
The new agent is one part of Gomumari’s broader service model, which combines implementation work with ongoing support. The company also offers a human-operated reconciliation service: customers send in their records and receive the results, rather than running an agent themselves. Gomumari says that option can be used for one-off or recurring checks, with costs depending on the work involved.
Source Reference Map
Inspired by headline at: [1]
Sources by paragraph:
- Paragraph 1: [2], [3]
- Paragraph 2: [2], [3], [4]
- Paragraph 3: [2], [3], [4]
- Paragraph 4: [2], [3]
- Paragraph 5: [2], [3]
- Paragraph 6: [2], [3]
- Paragraph 7: [3], [5], [7]
Source: Fuse Wire Services


