Retail and e-commerce automation: listings, labels, reviews

Marketplace product pages, food spec sheets, and testimonials on product pages. Tools handle orders and stock; AI and staff handle the checks that remain.

  • Runs on your own PCs and servers
  • Use Azure OpenAI, Amazon Bedrock, Google Vertex AI, or an in-house model
  • Its engine is Descarty's open-source Dagu (GPL-3.0)

Common problems

  1. Staff open each marketplace page one by one to see how products registered through a multi-channel tool actually appear

  2. Wrong comparison prices, wrong countries of origin, and testimonials not marked as ads have led to orders against online stores

  3. Staff read suppliers' food spec sheets and copy ingredients and allergens onto product pages by hand

Tools handle orders and stock; checking what pages show is left

According to Japan’s Ministry of Economy, Trade and Industry (METI), the consumer e-commerce market reached 27.6 trillion yen in 2025, up 5.5% from the previous year. Online sales accounted for 9.80% of consumer sales of goods. Among goods, food, beverages, and alcohol was the largest category, at 3.33 trillion yen.

Selling on Rakuten Ichiba, Amazon, Yahoo! Shopping, and your own site means handling orders, stock, and product registration in each storefront’s admin screens. Multi-channel management tools take this on. Next Engine imports orders through order APIs, order emails, and CSV data, keeps stock in sync across every store, and publishes and updates product pages in all stores at once. I’LL’s CROSS MALL likewise syncs stock across online stores, registers products in bulk, and manages orders. Amazon and Yahoo! Shopping publish APIs for sellers covering orders, inventory, products, and more.

What remains is checking what each storefront actually shows, copying supplier spec sheets onto product pages, and checking where the testimonials and posts quoted on product pages come from. All of it means reading content and making a judgment.

Work Screens and systems used What stays manual
Orders, stock, shipping Multi-channel tool, marketplace APIs, warehouses and WMS Mostly automated by multi-channel tools
Registering and updating products Multi-channel tool, each marketplace’s admin screens Registration can be done in bulk. Checking how each storefront shows it remains
Checking product page claims Product pages on Rakuten Ichiba, Amazon, Yahoo! Shopping, and your own site Comparing prices, comparison prices, country of origin, and “No. 1” claims with the product master and supporting evidence
Food labeling Supplier spec sheets (paper, PDF, Excel), the online store’s product admin Copying ingredients and allergens onto product pages and tracking spec sheet revisions
Quoted testimonials and posts Your site and product pages, the log of commissioned posts Matching quotes against the log and checking they are clearly marked as ads
New product registration Head office product master, each trading partner’s entry form Entering the same product information into each company’s form separately
Competitor prices Competitors’ product pages Opening pages and recording prices and promotions

Orders over product page claims reach online stores too

In fiscal 2025, Japan’s Consumer Affairs Agency (CAA) issued 13 cease-and-desist orders and 10 surcharge orders (339.4 million yen in total) under the Act against Unjustifiable Premiums and Misleading Representations, and gave 388 instances of administrative guidance. Its monitoring of online ads alone led to guidance asking for 324 claims to be corrected.

The orders include claims on online stores:

  • Tsuruha Group Merchandising showed a “regular price” next to the sale price for 79 products on its own online store. Those products had not actually been sold at that price for a considerable recent period (November 28, 2025)
  • Iris Plaza and Daiyu Eight, in their stores on the Qoo10 marketplace, showed the country of origin as “domestic” for 101 and 113 products made outside Japan (November 5, 2025)

In each case, dozens to over a hundred products were covered at once. Even when the right values are in the multi-channel tool, the pages each storefront actually shows need checking separately.

Stealth marketing rules reach whoever publishes commissioned posts

Since October 1, 2023, stealth marketing has been regulated under the same Act. The CAA describes it as hiding the fact that something is an ad. The rules apply to the business that supplies the product or service (the advertiser), not to the influencers it commissions.

In fiscal 2025, the CAA approved four commitment plans and gave seven instances of guidance under the stealth marketing rules. A commitment plan is a correction plan that a business suspected of a violation submits and the CAA approves, a procedure introduced on October 1, 2024. The approved plans included ones over customer stories for frozen meal delivery.

On October 20, 2026, the CAA issued a cease-and-desist order against giobai. The company had asked people to post on Instagram in return for free products and payment, then placed excerpts on its own site as customers’ voices. The CAA found that the company did not make clear that it had requested the posts.

Before quoting testimonials or posts on product pages, you need to check them against your log of commissioned posts and confirm they are clearly marked as ads.

For food, spec sheets and product pages must match

Japan’s Food Labeling Standards govern what appears on packaging and do not apply to what online stores post. Yet online shoppers cannot see the packaging and rely on what the site says. The CAA’s guidebook recommends showing the label items required by the standards on online stores as far as possible.

That information comes from suppliers’ spec sheets. According to the guidebook’s supplement, stores receive it in one of three ways: on paper or as PDFs, in spreadsheet formats, or by data exchange. Paper and PDFs must be read by eye and typed in, and even spreadsheet data sometimes ends up being typed in by hand. The CAA recommends data exchange, which causes fewer human errors. If you can receive data that way, start there.

After publishing, spec sheet revisions must be tracked, and the supplement recommends checking for updates on a set schedule. On April 1, 2026, cashew nuts were added to the allergens that must be labeled. There is a two-year transition period, and the CAA asked prefectures and other local authorities to tell food businesses to recheck their ingredients and manufacturing methods.

Passing product information along is still largely manual. METI describes manufacturers, wholesalers, and retailers entering product information into each company’s own forms as a “bucket brigade by hand,” and in November 2024 set up a study group on a shared product master.

What to check before automating

  • Whether a multi-channel tool is enough: orders, stock, and product registration and updates can be automated with multi-channel tools and marketplace APIs. Use them where they suffice
  • Marketplace and social media terms: check each marketplace’s terms and guidelines on automated checks of product pages and admin screens. The same applies to checking social media posts
  • What counts as the correct value: decide whether prices, country of origin, ingredients, and allergens are checked against the product master, spec sheets, or sales records
  • Where the evidence for each claim is: gather the sales records behind comparison prices, the surveys behind “No. 1” claims, and the log of commissioned posts, for staff to consult when reviewing findings
  • Who decides: staff decide whether a claim is a problem under the Act and how to correct allergen labeling. AI only points out differences and items that need checking
  • How each system signs in: check each admin screen’s sign-in steps, such as one-time passwords

What Kitewell does

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

  • Reads every storefront’s pages: AI reads the screen and extracts what product pages and admin screens on each marketplace and your own site show. Operations that worked are remembered and repeated the same way on later runs
  • Keeps running when screens change: when a remembered operation no longer fits a marketplace’s screen, AI reads the screen again, does the step, and remembers the new operation. If that fails, the run stops, and AI looks into the cause and suggests the fix. Fix it in the editor, then save and continue
  • Reads spec sheets and compares them: questions to AI and judgments are workflow steps. AI reads ingredients and allergens from spec sheets in PDF or image form and shows where they differ from the product page. The AI model is chosen per step
  • Starts from spec sheets that arrive by email: Gmail, Microsoft 365, and IMAP mailboxes are checked at set intervals, and attachments are saved
  • Runs from an Excel list: register a product list as a batch sheet, and it runs row by row and writes each result back to its row. Runs can also start on a schedule, such as Mondays at 8:00
  • People check where it matters: approval can be set on any step, such as before a page is corrected. Requests reach Slack, Microsoft Teams, and email, and sending one back restarts the run from the chosen step
  • Handles sign-in hurdles: saved browser sign-ins are reused by later runs. 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
  • Every check on record: each run keeps its logs, screenshots of the pages it checked, and approval decisions

You can start from one piece of work, such as listing products through checking how their pages display on each marketplace and fixing them. 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

Frequently asked questions

Is anything left to automate if we already use a multi-channel management tool?
Yes. Orders, stock, and product registration and updates are handled by multi-channel tools and marketplace APIs. What remains is reading and judging: checking the pages each storefront actually shows, reading spec sheets and matching them to pages, and checking where testimonials come from. Kitewell automates that work alongside staff review.
May a tool open and check marketplace product pages?
It depends on each marketplace's terms and guidelines. The PoC checks the terms and automates only what they allow. Items that can be checked in the seller admin screens may be checked there instead.
Does AI decide whether a claim breaks Japan's Act against Unjustifiable Premiums and Misleading Representations?
No. AI points out claims that differ from the product master or spec sheet, wording that needs supporting evidence, and quoted posts that are not marked as ads. Staff decide how the law applies and how to fix the page.

Use cases for any industry

  • All industries
  • Accounting & finance

Read invoices into the accounting system

AI reads incoming invoices, checks them against purchase orders, and records them in the accounting system. Invoices whose amounts differ wait for approval.

  • All industries
  • Accounting & finance

Create invoices and send them by email

Invoices are filled into your template from an Excel billing list and emailed once a person approves. AI first points out amounts far from last month's.

  • All industries
  • Sales

Research a list of prospects

For an Excel list of hundreds of companies, each one's business, locations, and size are collected from its website into its row, with a priority for sales.

  • All industries
  • HR & payroll

Find missing clock-ins

Collect yesterday's clock-ins from the attendance system and, once HR approves, ask employees with missing entries to fix them in Teams.

  • All industries
  • Accounting & finance

Take in emailed invoices and draft the entries

Find unread email with an invoice attached, have AI read the amount, and record a draft ledger entry. Uncertain entries wait for accounting to confirm.

  • All industries
  • IT

Review SaaS accounts

A SaaS account review matches each service's users to current staff. Collect each console's users, match them, and disable leavers only after approval.

  • All industries
  • Accounting & finance

Partner registration that starts on approval

A kintone approval starts the workflow by webhook. It looks up the tax registration number, AI matches it to the request, and a match enters the sales system.

  • All industries
  • Planning & marketing

Collect KPIs from SaaS tools into a weekly report

Collect figures from advertising, sales, support, and other SaaS tools, and produce a weekly report with week-on-week comparisons and likely causes of change.

  • All industries
  • Accounting & finance

Check supplier invoice registration numbers

Check your supplier master's registration numbers monthly through the NTA's Web-API, and list suppliers that were revoked, lapsed, or differ in name.

Prove Kitewell on your own systems

Try it free for 14 days on your own PC. To see it working first, ask for a 30-minute demo. There is also a one-month onboarding (¥500,000 before tax), where we build your first job together.

What the onboarding checks

  • Which method connects each target system: API, file, database, or screen
  • Whether Kitewell can work a system with no API through its screens, as a person would
  • Whether AI relearns a step and processing continues when a screen changes
  • Whether errors are handled as agreed: stop, notify, or retry
  • Whether it runs in your environment, with the AI model you choose
  • Whether processing time and AI usage cost fit daily operation

How the onboarding runs

  1. Week 1Requirements and success criteria
  2. Weeks 2 to 3Build
  3. Week 4Verify on production screens
  4. Final dayReport

One-month onboarding

Length
One month
Cost
¥500,000 (tax excluded). Your staff build, with the developer beside them
If you go on to a license
The onboarding fee is credited to the first year's contract
Without the onboarding
At Team scale, start on Team monthly (¥14,800 per computer a month, tax excluded), and cancel in any month

You can try it first on your own PC, in the free 14-day trial.

See how the onboarding runs and its terms