Bill One Implements New AI-Assisted Invoice Reconciliation Feature 'Reconciliation Assist'
Sansan Inc. has implemented a new feature called 'Reconciliation Assist' for its accounting AX service 'Bill One', which uses AI to assist in reconciling invoices with delivery and acceptance data. This aims to significantly reduce the burden on staff caused by discrepancies in wording and product specifications, thereby improving the efficiency and reducing the reliance on individual expertise in reconciliation tasks.
📋 Article Processing Timeline
- 📰 Published: June 9, 2026 at 20:00
- 🔍 Collected: June 9, 2026 at 11:21
- 🤖 AI Analyzed: June 12, 2026 at 16:52 (77h 31m after Collected)
Sansan Inc., a provider of AX services that transform the way people work, has launched 'Reconciliation Assist,' a new feature for its accounting AX service 'Bill One.' This feature utilizes AI to assist in the reconciliation of invoices with delivery and acceptance data.
The function identifies discrepancies, including synonyms and variations in wording, between item names on invoices and corresponding delivery/acceptance data from core systems. It then presents items that are determined to be the same. By referencing these results, staff can perform reconciliation tasks, significantly reducing the burden of manual visual checks.
Furthermore, Sansan is initiating a process to accumulate the results of reconciliations and corrections made by staff as training data for continuous AI learning. This will enable the AI to evolve and accurately identify and present items that were previously judged based on individual experience, thereby eliminating subjective decision-making by staff.
This new feature enhances 'AI Automatic Reconciliation' (Note 1), which was launched in November 2025, and Sansan has filed a patent for it.
'AI Automatic Reconciliation' Service Site: https://bill-one.com/ap/function/podo/
■Background of Feature Implementation
Many companies perform monthly reconciliation tasks to verify if the contents of received invoices match delivery and acceptance data exchanged with business partners on a line-item basis. Due to variations in product master databases across companies, discrepancies in product names occur, such as 'apple' versus 'ringo' (apple in Japanese). Sansan's verification, covering approximately 30,000 invoice lines, found that over half of the line items contained such variations, a trend particularly strong in industries like wholesale and food with complex supply chains. In some transactions, staff must reconcile thousands of invoice line items, accurately distinguishing between variations in wording and minor differences in model numbers or specifications through visual inspection, which has been a significant burden.
Bill One developed this feature to eliminate the burden of reconciliation tasks caused by these data discrepancies and minor differences in model numbers and specifications.
■Overview of Reconciliation Assist Feature
The new feature evaluates both the semantic similarity and the string proximity of the text in the target invoice and delivery/acceptance data. This allows the AI to identify variations in wording and accurately distinguish even minor differences in model numbers and specifications. Items that match in amount and are determined to be the same are displayed as 'candidate matches' in the reconciliation results.
For example, it can identify discrepancies in item names like 'Kuroge Wagyu Beef 50KG/5 boxes' versus 'F Domestic Beef 50kg x 5' (where 'Kuroge Wagyu' and 'Domestic Beef' are identified as similar), and differences in product numbers such as 'Marker H007J-BL-1' versus 'Marker H007J-RD-10P.' As a result, 'Frozen Kuroge Wagyu Beef 50KG/5 boxes' and 'F Domestic Beef 50kg x 5' would be judged as the same and displayed as 'candidate matches,' while 'Marker H007J-BL-1' and 'Marker H007J-RD-10P' would be judged as different products and excluded from 'candidate matches.'
Subsequently, staff can perform reconciliation while referring to the AI's presented results, eliminating the effort and risk of misidentification associated with visually distinguishing variations in wording and minor differences in model numbers or specifications.
Image of Invoice and Delivery/Acceptance Data Reconciliation Results
■Future Development
In 'AI Automatic Reconciliation,' all reconciliation results, including corrections made by staff to AI suggestions and manually linked reconciliation records on the screen, will be accumulated as training data. This will enable the AI to evolve and accurately identify and present items based on past performance, even for reconciliations where judgment relied on the experience of staff familiar with the transaction and where invoice wording alone was insufficient. This process aims to structurally eliminate reliance on individual expertise. Bill One will focus on development to automatically reduce the workload through simple daily operations like linking invoice details with delivery and acceptance data.
Image of AI Automatic Reconciliation Machine Learning Process
Note 1: Sansan Inc. 'Bill One Launches "AI Automatic Reconciliation" to Reconcile Invoices and Procurement Data by Total Amount and Line Item' (Announced November 27, 2025)
https://jp.corp-sansan.com/news/2025/1127.html
(End)
■'Bill One': Creating 'Eliminatable' Tasks and Transforming Company-Wide Work Styles
Bill One, provided by Sansan Inc., is an accounting AX service that solves challenges in various business operations such as invoice reception, expense settlement, and accounts payable management. By leveraging AI and technology to create 'eliminatable' tasks within company-wide business processes involving documents like invoices and receipts, it transforms the work styles not only of accounting departments but of the entire company.
https://bill-one.com/
■Sansan Inc. Company Overview
With the mission 'To foster innovation from encounters,' Sansan Inc. provides AX services that transform the way people work. Its main services include the business database 'Sansan,' the business card app 'Eight,' the accounting AX service 'Bill One,' the transaction management service 'Contract One,' and data quality management 'Sansan Data Intelligence,' offered both domestically and internationally.
Established: June 11, 2007
URL: https://jp.corp-sansan.com/
Location: Shibuya Sakura Stage 28F, 1-1 Sakuragaokacho, Shibuya-ku, Tokyo 150-6228, Japan
Capital: 7.35 billion JPY (as of February 28, 2026)
Business Description: Planning, development, and sales of AX services that transform work styles.
Sansan https://jp.sansan.com/
Eight https://8card.net/
Bill One https://bill-one.com/
Contract One https://contract-one.com/
Sansan Data Intelligence https://jp.sansan.com/sansan-data-intelligence/
The function identifies discrepancies, including synonyms and variations in wording, between item names on invoices and corresponding delivery/acceptance data from core systems. It then presents items that are determined to be the same. By referencing these results, staff can perform reconciliation tasks, significantly reducing the burden of manual visual checks.
Furthermore, Sansan is initiating a process to accumulate the results of reconciliations and corrections made by staff as training data for continuous AI learning. This will enable the AI to evolve and accurately identify and present items that were previously judged based on individual experience, thereby eliminating subjective decision-making by staff.
This new feature enhances 'AI Automatic Reconciliation' (Note 1), which was launched in November 2025, and Sansan has filed a patent for it.
'AI Automatic Reconciliation' Service Site: https://bill-one.com/ap/function/podo/
■Background of Feature Implementation
Many companies perform monthly reconciliation tasks to verify if the contents of received invoices match delivery and acceptance data exchanged with business partners on a line-item basis. Due to variations in product master databases across companies, discrepancies in product names occur, such as 'apple' versus 'ringo' (apple in Japanese). Sansan's verification, covering approximately 30,000 invoice lines, found that over half of the line items contained such variations, a trend particularly strong in industries like wholesale and food with complex supply chains. In some transactions, staff must reconcile thousands of invoice line items, accurately distinguishing between variations in wording and minor differences in model numbers or specifications through visual inspection, which has been a significant burden.
Bill One developed this feature to eliminate the burden of reconciliation tasks caused by these data discrepancies and minor differences in model numbers and specifications.
■Overview of Reconciliation Assist Feature
The new feature evaluates both the semantic similarity and the string proximity of the text in the target invoice and delivery/acceptance data. This allows the AI to identify variations in wording and accurately distinguish even minor differences in model numbers and specifications. Items that match in amount and are determined to be the same are displayed as 'candidate matches' in the reconciliation results.
For example, it can identify discrepancies in item names like 'Kuroge Wagyu Beef 50KG/5 boxes' versus 'F Domestic Beef 50kg x 5' (where 'Kuroge Wagyu' and 'Domestic Beef' are identified as similar), and differences in product numbers such as 'Marker H007J-BL-1' versus 'Marker H007J-RD-10P.' As a result, 'Frozen Kuroge Wagyu Beef 50KG/5 boxes' and 'F Domestic Beef 50kg x 5' would be judged as the same and displayed as 'candidate matches,' while 'Marker H007J-BL-1' and 'Marker H007J-RD-10P' would be judged as different products and excluded from 'candidate matches.'
Subsequently, staff can perform reconciliation while referring to the AI's presented results, eliminating the effort and risk of misidentification associated with visually distinguishing variations in wording and minor differences in model numbers or specifications.
Image of Invoice and Delivery/Acceptance Data Reconciliation Results
■Future Development
In 'AI Automatic Reconciliation,' all reconciliation results, including corrections made by staff to AI suggestions and manually linked reconciliation records on the screen, will be accumulated as training data. This will enable the AI to evolve and accurately identify and present items based on past performance, even for reconciliations where judgment relied on the experience of staff familiar with the transaction and where invoice wording alone was insufficient. This process aims to structurally eliminate reliance on individual expertise. Bill One will focus on development to automatically reduce the workload through simple daily operations like linking invoice details with delivery and acceptance data.
Image of AI Automatic Reconciliation Machine Learning Process
Note 1: Sansan Inc. 'Bill One Launches "AI Automatic Reconciliation" to Reconcile Invoices and Procurement Data by Total Amount and Line Item' (Announced November 27, 2025)
https://jp.corp-sansan.com/news/2025/1127.html
(End)
■'Bill One': Creating 'Eliminatable' Tasks and Transforming Company-Wide Work Styles
Bill One, provided by Sansan Inc., is an accounting AX service that solves challenges in various business operations such as invoice reception, expense settlement, and accounts payable management. By leveraging AI and technology to create 'eliminatable' tasks within company-wide business processes involving documents like invoices and receipts, it transforms the work styles not only of accounting departments but of the entire company.
https://bill-one.com/
■Sansan Inc. Company Overview
With the mission 'To foster innovation from encounters,' Sansan Inc. provides AX services that transform the way people work. Its main services include the business database 'Sansan,' the business card app 'Eight,' the accounting AX service 'Bill One,' the transaction management service 'Contract One,' and data quality management 'Sansan Data Intelligence,' offered both domestically and internationally.
Established: June 11, 2007
URL: https://jp.corp-sansan.com/
Location: Shibuya Sakura Stage 28F, 1-1 Sakuragaokacho, Shibuya-ku, Tokyo 150-6228, Japan
Capital: 7.35 billion JPY (as of February 28, 2026)
Business Description: Planning, development, and sales of AX services that transform work styles.
Sansan https://jp.sansan.com/
Eight https://8card.net/
Bill One https://bill-one.com/
Contract One https://contract-one.com/
Sansan Data Intelligence https://jp.sansan.com/sansan-data-intelligence/
FAQ
What problem does 'Reconciliation Assist' solve?
It reduces the burden of manual visual checks for staff by identifying discrepancies in item wording, model numbers, and specifications during invoice and delivery/acceptance data reconciliation.
How does the AI learn and improve its accuracy?
It accumulates staff reconciliation and correction results as training data and continuously learns to achieve high-precision discrimination capabilities.
What industries is this feature suitable for?
It is expected to be particularly effective in industries like wholesale and food with complex supply chains, where wording variations are common.
What is the difference between 'Reconciliation Assist' and 'AI Automatic Reconciliation'?
'Reconciliation Assist' enhances 'AI Automatic Reconciliation' by enabling AI to identify and present more advanced wording variations and model number differences.
What are the benefits of implementing Bill One?
It contributes to the overall efficiency and DX promotion of accounting operations, from invoice reception to expense settlement and accounts payable management.