“Zenbuyommon LLM Option Powered by Azure OpenAI Service”
Free trial version available for businesses until September 29th

Cinnamon AI has developed our original AI-OCR “Flax Scanner,” which can automatically recognize characters and extract text from both handwritten and printed documents, and specializes in reading characters from images of technical documents in the manufacturing and construction industries. We have provided "Zenbuyommon" as an AI model that has been developed.

We will start offering "Zenbuyommon LLM Option Powered by Azure OpenAI Service" (hereinafter referred to as "Zenbuyommon LLM Option"), which links large-scale language models as an option, from August 1, 2023.

Screen image of extracting information from OCR result text using "Zenbuyommon LLM Option"

Maintenance Resilience TOKYO2023(July 26th to July 28th/Tokyo Big Sight) We will have a demonstration.
Please stop by the Cinnamon AI booth.
https://www.jma.or.jp/mente/exhibit/index.html

"Zenbuyommon" is Cinnamon AI's AI-OCR that supports the digitization of paper documents in the manufacturing and construction industries and promotes the conversion of knowledge into assets. We handle a wide range of documents including industry terminology, specifications, reports, inspection records, drawings, experiment/test data, instruction manuals, manuals, specialized magazines, research materials, contracts, etc. It can read kanji, hiragana, katakana, alphanumeric characters, symbols and units, and achieves high reading accuracy of 91% or higher* for both handwritten and printed characters.
*Our test data value.

The newly released "Zenbuyommon LLM Option" allows users to utilize document knowledge by linking "Zenbuyommon" with Azure OpenAI Service to provide various information processing support for the contents of technical documents through chat. A typical use case is the ability to easily extract specific information from document content, which previously required individual development. In addition, when using ChatGPT via a browser, information leakage due to input data being used for learning is a security issue, but with "Zenbuyommon LLM Option", it is linked with ChatGPT via API. , the input data is never used for training. Therefore, it is possible to use large-scale language models for business purposes in environments where input data is not used for learning.

Until now, Cinnamon AI has introduced many AI-OCR systems for the manufacturing industry. Many companies in the manufacturing industry still store large amounts of paper documents, and they may not be able to access the information they need unless they go to a warehouse or archive to look for it. Additionally, even if documents are converted to PDF, they can only be searched by file name and cannot be searched by the content of the text, making it difficult to effectively utilize the large amount of stored knowledge. .

``Zenbuyommon'' converts all the characters written in a document into text data, making it possible to search efficiently including the contents of the text. The ``Zenbuyommon LLM Option'' enables further utilization of accumulated knowledge by processing data-based content using a large-scale language model.

Image of knowledge utilization by introducing “Zenbuyommon”

■Looking for free trial
We will be offering free trial versions of ``Zenbuyommon'' and ``Zenbuyommon LLM Option Powered by Azure OpenAI Service'' for businesses until September 29th (Friday).
・Trial version application URL: https://go.cinnamon.ai/zenbu_yomumon_LP1.html
・Download all materials: https://go.cinnamon.ai/wp_zenbu_yomumon_dl.html

■Basic price plan (2 types)
① Monthly usage
A plan that allows customers to upload/download files themselves on the web screen.
[Initial cost] 0 yen
[Monthly] 50,000 yen (excluding tax) *Up to 1000 pages per month.
[LLM option Powered by Azure OpenAI Service] ★
10,000 yen per month (excluding tax)

② Spot use
A plan in which we receive the target image file, perform OCR processing at our company, and then return it all at once.
【Basic charge】
2 million yen (excluding tax) + number of pages processed x 6 yen (excluding tax)
*Spot use until August 31, 2023
If you apply, we will run a campaign where the basic fee will be half price.
*The application date is the contract conclusion date.

■File formats compatible with “Zenbuyommon” input/output *Same as LLM option

• inputImage import (full text AI-OCR processing) is possible from 6 types of file formats.
• File formats that can be processedPDF/JPG/JPEG/PNG/TIFF/TIF
• outputOutput is possible in 3 types of files depending on the purpose

– Excel (lmage line report)
This is the OCR reading output result of the image file and text. with images and text
You can check the OCR results.

– Excel (line report)
The OCR reading output result is text only.

– PDF (with transparent text)
This is a PDF file with OCR-read transparent text* added to the original image file.

* The information of the original image file is added to the searchable PDF file and output as a PDF file.
This makes it possible to perform full text searches from OCR reading results in addition to images.

Output file image of the result of converting to text with "Zenbuyommon"