Announced details
Merchant-name mapping, address normalization, corporate-name matching, transaction-description classification, fraud-detection text analysis, and structuring data after document OCR. Target industries include card companies, payment processors, personal finance services, financial institutions, real estate, logistics, credit management, anti-social forces screening, CRM, banks, accounting outsourcing, insurance, and securities.
Approximately 95.7% cost reduction versus the OpenAI API, approximately 98% reduction in manual checks, processing throughput of 15.1 cases per second, accuracy of more than 95% for cases with a confidence score of 90 or higher, processing of 1 million cases in approximately 18.4 hours, processing costs of approximately ¥4,500 per 1 million cases, expected annual savings of more than ¥25 million from merchant-name mapping alone and more than ¥40 million including expansion to other operations
Nowcast developed a prototype in an internal AI contest and demonstrated that it had reached a level suitable for application to internal operations. Nowcast processes approximately 300,000 merchant-name mappings per month.
Sumitomo Fudosan Kudan Building, 9th Floor, 1-8-10 Kudan-kita, Chiyoda-ku, Tokyo (Nowcast)
In this announcement the company addresses none of Contract Value / Offering price / Impact on Business Performance.
| Item | Details |
|---|---|
| Names of Products / Services Covered | Local LLM Text Processing Platform |
| Applicable area | Merchant-name mapping, address normalization, corporate-name matching, transaction-description classification, fraud-detection text analysis, and structuring data after document OCR. Target industries include card companies, payment processors, personal finance services, financial institutions, real estate, logistics, credit management, anti-social forces screening, CRM, banks, accounting outsourcing, insurance, and securities. |
| Location | Sumitomo Fudosan Kudan Building, 9th Floor, 1-8-10 Kudan-kita, Chiyoda-ku, Tokyo (Nowcast) |
| Quantitative Effects | Approximately 95.7% cost reduction versus the OpenAI API, approximately 98% reduction in manual checks, processing throughput of 15.1 cases per second, accuracy of more than 95% for cases with a confidence score of 90 or higher, processing of 1 million cases in approximately 18.4 hours, processing costs of approximately ¥4,500 per 1 million cases, expected annual savings of more than ¥25 million from merchant-name mapping alone and more than ¥40 million including expansion to other operations |
| Usage history | Nowcast developed a prototype in an internal AI contest and demonstrated that it had reached a level suitable for application to internal operations. Nowcast processes approximately 300,000 merchant-name mappings per month. |
Items this document does not state (13)
These are items the company does not state in this document. We do not leave them blank, nor fill them in from other sources or by inference.
The reasons the company gives
The company's own explanation, not something inferred from the numbers.An internal AI contest prototype demonstrated that confidential information could be processed with high accuracy without sending it externally, while also reducing costs and manual work. Nowcast aims to apply the platform to the challenge, common across the financial industry, of processing text data with AI when the data cannot be sent externally.