Manufacturing automation: quotes, chemical surveys, molds

Manual work stays in dealings with customers

Japan’s manufacturing workforce fell from 10.46 million in 2024 to 10.33 million in 2025. Among small and medium-sized manufacturers, the staffing diffusion index stood at minus 17.9 in 2025, meaning more firms report too few staff than too many. In a survey shown in METI’s briefing on the 2026 White Paper on Manufacturing Industries, 70.9% of manufacturers named labor shortages as a condition affecting their business, second only to rising raw material and resource prices (fiscal 2025).

Inside the company, orders, production, and stock can be managed in a production management system, and drawings and bills of materials in PLM or drawing management systems. Industries also have shared systems for dealing with customers. The auto industry has JNX, a common network with 2,286 member companies (April 2026). The electronics industry has JEITA’s EDI standard, ECALGA.

Even so, only 16.4% of manufacturers exchange data with other companies in their supply chain (2025 survey), and METI says data exchange between companies has made no progress and has barely changed in two years. Where EDI doesn’t reach, business runs through each customer’s web EDI or portal, email, and fax. JEITA, the electronics industry association, publishes guidelines for web EDI so that it does not cause undue inconvenience or poor usability for the companies that use it. In a 2024 survey of manufacturers’ order staff by AI inside, an AI OCR vendor, 73.3% of those who still exchange paper documents said some of their work involves receiving documents by fax, and 60.4% of those enter faxed documents into their order system by hand.

Work	Screens and systems used	What stays manual
Orders and delivery date changes	Customers’ web EDI and portals, production management system	Signing in to the portals of customers outside EDI and copying orders and date changes
Drawing revisions	Customers’ portals, internal PLM or drawing management	Noticing revised drawings posted on a customer’s portal, pulling them in, and telling the engineer
Quotes	RFQs by email or portal, quoting system, production management system	Picking materials and quantities off drawings, and looking up past quotes and costs in separate screens before entry
Chemical content surveys	Customers’ procurement portals, the chemSHERPA data creation tool	Collecting requests, building an answer file for each part, and submitting it the way each customer asks
Quality complaints	Customers’ quality portals (8D reports and similar)	Entering cause and corrective action reports stage by stage in each customer’s format
Price negotiations and molds	Email, Excel ledgers, purchasing system	Tracking suppliers’ requests to negotiate, and orders for molds held at suppliers, against the ledger

Chemical content surveys differ by customer in format and channel

Electronics and automotive parts makers receive surveys from customers, part by part, on the chemical substances their products contain. Many customers use chemSHERPA, the industry’s common format, but it is rarely the whole story. In a 2021 survey of chemSHERPA users by the industry consortium JAMP, 96.90% of the 3,062 upstream and midstream respondents who receive surveys were asked to use chemSHERPA, 74.95% to provide non-use certificates, and 55.52% to fill in customers’ own forms. JAMP concluded that customers’ own forms and non-use certificates were increasing.

How requests arrive differs by customer, too.

• Hamamatsu Photonics sends requests by email as a rule
• Fujitsu sends requests through its ProcureMART procurement portal. Suppliers load the requester information from Fujitsu’s request data, build a chemSHERPA file for each item, and attach the compliance guarantee as a pair of Excel and PDF files
• Canon sends its own additional forms alongside chemSHERPA, to survey what chemSHERPA cannot cover

Each regulatory change means surveying again. The chemSHERPA substance list is revised twice a year. The CMP Consortium says that each time regulated substances are added or changed, companies must re-survey and pass information to trading partners by email and other means, and that this workload has become a problem.

To reduce it, the CMP Consortium launched CMP on October 1, 2026, a platform that passes chemical content information to designated partners along trading relationships, and aims for several thousand participating companies in 2028. CMP and chemSHERPA will run side by side, complementing each other, for the time being. Trades that move to CMP connect through data. Requests through the portals and email of customers that have not moved still mean manual work.

Quotes mean reading drawings and finding past records before entry

RFQs arrive by email or through customer portals, with drawings and specifications attached. Estimators pick out the material, dimensions, quantity, and surface treatment from the drawing, look up past quotes and actual costs for similar parts, and then enter it all in the quoting system.

In a Kanagawa Prefecture DX demonstration project, the foundry Koiwai reported that quoting and casting design had become work tied to individual veteran employees and had not been passed on to younger staff. Customers were asking for quote replies in three days instead of a week. Finding similar shapes among 10,000 or more sets of 3D data took inexperienced staff a full day. The company trialed a system that uses AI to find similar shapes.

Dedicated systems like this are one way to search for similar shapes and estimate costs. Either way, the work around them remains: reading the request, finding past records, and entering the quote. Estimators set the price.

The amended Subcontract Act calls for records on price talks and mold storage

An amendment passed in May 2025 renamed Japan’s Subcontract Act, now known as the Toriteki Act, and it took effect on January 1, 2026. Manufacturers are on the receiving side with their customers, and on the ordering side when they outsource machining or mold making. Whether the act applies depends on thresholds for capital and number of employees.

The amendment bans setting prices unilaterally without negotiating (Article 5(2)(iv)). The guide by the Japan Fair Trade Commission (JFTC) and the SME Agency says that ignoring a request to negotiate, or repeatedly putting off negotiation, also counts as refusing to negotiate. A supplier presenting a quote calculated at a raised unit price counts as a request to negotiate. Such requests may arrive as a revised quote attached to an email.

Records are required too. The ordering party must create records of each transaction (known as Article 7 records) and keep them for two years. When it changes a price, it records the increase or decrease and the reason. The promotion standards included in the guide say that both parties are each to keep records of their price negotiations.

Mold storage is under scrutiny as well. The guide treats making a supplier store molds without paying storage costs, while not ordering the parts for a long time, as an unjust request for economic benefits. Its examples include having a supplier store molds free of charge when no products made with them have been ordered for a year or more. Where tracking each mold’s use at all times is too burdensome, it also allows checking orders once a year and paying that year’s storage costs together. Of the 16 recommendations the JFTC issued in fiscal 2026 (through October 9, 2026), 10 included having suppliers store molds or similar tools without bearing the storage costs.

What automation reduces is missed negotiation requests and overlooked idle molds, and the work of pulling records together. Staff negotiate and decide prices and what happens to molds.

What to check before automating

• Whether data exchange is enough: where a customer or trade runs on EDI, JNX, or CMP, use that
• Each customer portal’s terms of use: check whether each one allows automated operation (Check the terms before you automate a website)
• Which AI models may see drawings: drawings and specifications from customers fall under confidentiality agreements. Decide first whether they may go to outside AI models, or only to in-house or local models
• Documents the company stands behind: QA staff check chemical content guarantees and non-use certificates before they go out
• How each portal signs in: IDs and passwords, client certificates, and one-time passwords differ by customer portal
• Where people decide: staff set quote prices, conclude price negotiations, and decide whether to take back or scrap molds

What Kitewell does

Descarty’s Kitewell is integration and automation middleware that connects systems through APIs, files, databases, and screens.

• Operates each customer’s portal: AI reads the screen to collect orders, download drawings, and answer surveys. Operations that worked are remembered and repeated the same way on later runs
• Keeps going when screens change: when a remembered operation no longer fits the screen, AI reads the screen again, does the step, and remembers the new operation. If it still cannot proceed, the run stops, and AI looks into the cause and suggests the fix
• Operates desktop apps too: quoting systems, production management systems, the chemSHERPA data creation tool, and other apps without an API are operated by reading the screen. Desktop steps run one at a time on a signed-in, unlocked screen
• Reads drawings and requests: steps can have AI read PDFs and images. You choose the AI model for each step, including in-house and local models
• Handles sign-in hurdles: for sites that ask for a client certificate, it presents the certificate registered on the computer for that site. When a one-time password is required, the run pauses until staff enter it. Passwords are referenced by name and typed in without showing the value to the AI
• People check where it matters: approval can be set on any step, such as before answering a survey or sending a quote. Requests reach Slack, Microsoft Teams, and email, and sending one back restarts the run from the step you choose
• Everything on record: each run keeps its logs, screenshots, and approval decisions
• Runs from email and Excel: it checks email at set intervals and saves attachments. Register an Excel list, such as a mold ledger, as a batch sheet, and it runs row by row and writes each result back to its row. It can also convert Excel files to Shift_JIS CSV for older production management systems’ import screens

The computers that run workflows can be VMs or dedicated PCs in your own cloud, on your premises, or in a closed network. With in-house or local AI models, nothing is sent outside, drawings included. AI models that can read and operate a screen are limited, though, and small local models usually cannot operate one, so the PoC settles which model to use.

For connecting hand-offs between in-house production, sales, and core systems, see on-premises integration.

You can start from one piece of work, such as receiving a customer’s chemical content survey through gathering each part’s data and preparing and submitting the answer. A demo shows how it runs, a free connection test confirms Kitewell reaches your systems, and a one-month PoC measures the result against success criteria agreed up front, before you license it for production. See From evaluation to production for details.

Sources

• Ministry of Health, Labour and Welfare: 2026 White Paper on Manufacturing Industries, summary of the MHLW chapters (May 2026; in Japanese)
• METI: On the 2026 White Paper on Manufacturing Industries (June 2026; in Japanese)
• JNX Center: JNX: Japanese automotive Network eXchange (in Japanese)
• JEITA EC Center: JEITA/EC Center and JEITA Web-EDI guidelines (in Japanese)
• AI inside: More Than 70% of Manufacturing Orders Are Still Received by Fax (August 29, 2024)
• ZF: Problem Solver - Electronic Complaint Resolution
• JAMP: Summary of the June 2021 adoption survey (July 29, 2021; in Japanese)
• Hamamatsu Photonics: How to answer green procurement surveys (chemSHERPA) (October 1, 2018; in Japanese)
• Fujitsu: Fujitsu ProcureMART: answering technical and environmental information surveys (June 30, 2025; in Japanese)
• Canon: Canon green procurement survey answer manual (March 2026; in Japanese)
• CMP Consortium: Changes to the chemSHERPA substance list revision timing and external list versions (June 5, 2026; in Japanese)
• CMP Consortium: CMP, the platform for chemical content and resource circulation information, launches (October 1, 2026; in Japanese)
• Kanagawa Prefecture and Koiwai: Making quoting more efficient with an AI system and building a base for casting design (March 9, 2022; in Japanese)
• JFTC: The Toriteki Act (amended Subcontract Act) (in Japanese)
• JFTC and SME Agency: Toriteki Act guide (November 2025; in Japanese)
• JFTC: Recommendations under the Toriteki Act, fiscal 2026 (in Japanese)