Complex order processing,
fully automated
FAX, email, Excel, paper … no matter where it arrives from, the AI reads it and consolidates it into your core system.
Orders, of course — plus invoices, applications, inspection records and every other document.
ARCHITECTURE
Our vision for building a competitive foundation
Every IPLoT product and service takes a place in this shared OS of competitive strength. Yomitoru-kun covers the highlighted part (Activity-DB layer).
How IPLoT supports this model
The areas covered by each IPLoT product and service within the model above.
| Layer | Products & services in charge |
|---|---|
| Structured-UI layer (the window) | Brain Pro / Shiftly / System development |
| Knowledge layer (what AI is taught) | iPro Manual / AI Strategist Training / Knowledge Loop Consulting |
| Activity-DB layer (store the data) | AI Data Analysis / Yomitoru-kun / Kakikomu-kun / Wholesale core system |
| Agent layer (where AI acts) | IPRO-kun / Menta-kun / System development |
| Designing & applying the whole structure | IT consulting (overall design) / HRobo (all layers for HR & talent) / Knowledge Loop Consulting (full rollout) |
Yomitoru-kun covers: Activity-DB layer
HOW IT WORKS
Receive, read, consolidate. Fully automate your order processing
FAX, email, Excel, paper, PDF, chat — no matter which entry point it arrives from, the AI reads it and consolidates it into your core system. It flows automatically from receipt to loading, and the only things people touch are "confirming fuzzy matches" and "exception handling."
TARGET
Sound familiar?
Order data arriving from scattered entry points is read and consolidated into one. From AI parsing to core-system integration, we automate it all.
FAX, email, LINE, Excel — entry points are scattered and you cannot keep up
Every client sends things differently, and you handle each one by hand.
You are at the mercy of each client's Excel layout
Every time a new client is added, you have to relearn the transcription rules.
Transcription errors into the core system never go down
Digit shifts and mis-entries keep causing order troubles.
Month-end closing piles up and runs late every month
Waiting on transcription turns the whole team into a bottleneck.
INPUT
Full support for every entry point
FAX, email attachments, Excel, paper, LINE / chat, Google Drive, scanners, PDF. No matter where it comes from, feed it to Yomitoru-kun and it is consolidated into a single workflow.
FAX
Receipts & orders
Email attachment
Gmail / Outlook
Excel
Client-specific layouts OK
Paper documents
Scanners & MFPs
LINE
Official-account linkage
Slack / Teams
Files sent over chat
Google Drive
Auto folder sync
Parses any PDF
"Our clients use this way of sending things too" — just ask. We adapt flexibly to adding new entry points.
POINT
Three reasons Yomitoru-kun is chosen
─ MULTI INPUT
Whatever the entry point, Yomitoru-kun picks it up
FAX, paper, email attachments, Excel, LINE / Slack, Google Drive — order data arriving through any channel is consolidated into a single workflow. "A different way of handling every client" disappears.
─ AI LAYOUT
Even Excel that differs per client is parsed automatically by AI
With no template registration, the AI reads field positions and column-name variations from context. Even as new clients are added, there is no need to reconfigure each time.
─ AUTO INTEGRATION
Automated all the way to core-system integration
From master matching (client IDs, product IDs) to output via Excel / CSV / API — all automatic. The only things people touch are "confirming fuzzy matches" and "exception handling." You can leave 99% of order processing to the AI.
FEATURES
Feature list
The feature set running in the field today. You can combine the features you need to match your operations.
─ INPUT
Every entry point, into one
Multi-channel intake
From FAX, email attachments, Excel, LINE / Slack, Google Drive and scanners — all consolidated into a single workflow
Ingest PDF / Office / images, all of it
Excel, Word, CSV and images are auto-converted to PDF via LibreOffice → straight into the OCR pipeline
Google Drive folder sync
New files placed in a designated folder are automatically ingested on a schedule. No manual upload needed
Automatic email intake
Attachments sent to a designated Gmail / Outlook address are automatically ingested
─ EXTRACTION
A high-accuracy, high-speed AI extraction engine
Azure Document Intelligence + Gemini AI
A hybrid extraction pipeline combining coordinate extraction (Azure) with semantic understanding (LLM)
Automatic per-client Excel layout parsing
With no template registration, the AI reads field positions and column-name variations from context
Parallel page processing
Long PDFs are OCR'd page by page in parallel, cutting processing time to 1/N of before
Automatic page-rotation detection
Portrait, landscape and upside-down PDFs are automatically detected for correct orientation before extraction
Asynchronous worker processing
Long-running OCR is processed in the background with Redis RQ, never blocking the UI
─ EDIT & VERIFY
Editing and verification that work in the field
Spreadsheet-style inline editing
Review and fix extraction results in a table. Add, delete and reorder rows
Highlight display
Highlight the relevant spot on the source PDF and edit while looking at the extraction evidence
Per-row image cropping
Cut out only the PDF region matching a row to save and review. Useful for re-OCR and verification
Subtotal verification panel
Automatically reconciles line-item totals against the billed total, instantly catching OCR mistakes and digit shifts
Undo / Redo / History
Keeps the full edit history so you can revert mistakes instantly. Re-running and rolling back OCR is also possible
─ MASTER & MULTI-TENANT
Master management and per-client rules
Master-data management (CSV / API)
Upload and edit masters such as clients, product IDs and delivery destinations via CSV or API. Automatically matched against OCR results
Per-client extraction rules
Configuration management to switch fields, extraction logic and output templates for each client
ID selection modal
When master matching produces a fuzzy match, a UI presents candidates for the user to choose from
─ OUTPUT & OPERATIONS
Core-system integration and operational support
Excel / CSV export
Output formatted to match the import format of your core system or accounting software
API integration
Load data directly into the core system. Custom support for SAP / Oracle and domestic ERPs
Slack notifications
Auto-notify Slack on OCR completion or error, making batch operations easy to monitor
Debug mode
Synchronous processing plus intermediate JSON saving for development and tuning, accelerating early accuracy improvement
History & re-run
Keeps the full history of past OCR runs. Re-running after config changes and diff comparison are possible
USAGE
Use cases
Beyond orders — from invoices, applications and inspection records to contracts, you can use it for any document or paperwork.
Order and procurement processing
Ingesting invoices and delivery notes
Digitizing applications and approval requests
Ingesting inspection and examination records
Reading contracts and various documents
Accounting and bookkeeping entry
─ END-TO-END AUTOMATION
fully automate your order processing.
EFFECT
The impact of adoption
The business impact of adopting Yomitoru-kun, in numbers.
─ EFFECT 01
Less transcription work
AI takes over the manual data-entry steps
─ EFFECT 02
Digit shifts & misreads
AI extraction + automatic checks eliminate human error
─ EFFECT 03
Month-end crunch
Immediate processing on receipt removes the closing-day bottleneck
─ EFFECT 04
Time to go live
Start small with a single client
FLOW
How we roll it out
Start small in as little as 2 weeks. We lay out the responsibilities of the customer and IPLoT in chronological order.
STEP 01
~ 1 wk
Mapping INPUT patterns
STEP 02
1 wk
Collecting document samples
STEP 03
1–2 wks
Type classification / AI tuning
STEP 04
1–2 wks
Core-system integration (custom)
STEP 05
2 wks + ongoing
Trial → full operation
─ CUSTOMER
Customer
Inventory the order entry points. Share "which client comes from where"
Provide a full set of real documents and Excel samples
Confirm the meaning of extracted fields (exceptions only)
Share the core-system spec and provide masters
Pilot in the field → move to full operation
─ IPLOT
IPLoT
Discovery and organization. Identify the priority clients and documents
Comprehensive analysis of layout variation and exception patterns. Fix the design approach
Type classification, extraction-rule design, AI tuning, accuracy validation
Master retrieval, matching build-out, output-format alignment
Accuracy and operations monitoring, continuous improvement, support for adding clients
STEP 01
Mapping INPUT patterns
─ CUSTOMER
Inventory the order entry points. Share "which client comes from where"
─ IPLOT
Discovery and organization. Identify the priority clients and documents
STEP 02
Collecting document samples
─ CUSTOMER
Provide a full set of real documents and Excel samples
─ IPLOT
Comprehensive analysis of layout variation and exception patterns. Fix the design approach
STEP 03
Type classification / AI tuning
─ CUSTOMER
Confirm the meaning of extracted fields (exceptions only)
─ IPLOT
Type classification, extraction-rule design, AI tuning, accuracy validation
STEP 04
Core-system integration (custom)
─ CUSTOMER
Share the core-system spec and provide masters
─ IPLOT
Master retrieval, matching build-out, output-format alignment
STEP 05
Trial → full operation
─ CUSTOMER
Pilot in the field → move to full operation
─ IPLOT
Accuracy and operations monitoring, continuous improvement, support for adding clients
PRICING
Pricing
PLAN
Standard
Multiple documents / up to 10,000 pages/mo
- Order-document OCR (multiple clients)
- Excel / CSV output
- Email support
- Operation-status reports
PLAN
Enterprise
Estimated by volume
- Support for every document and input source
- Core-system API integration
- Per-client rules and master setup
- Hands-on rollout and operational support
FAQ
Frequently Asked Questions
Yes. From FAX receipts, paper scans, email attachments, Excel, PDF and Google Drive to purchase orders sent via chat tools like LINE / Slack — it ingests anything, regardless of the entry point. It is designed so that no matter where it comes from, you just feed it to Yomitoru-kun and you are done.
The AI automatically analyzes each client's different layout. Even with no template registered in advance, it reads field positions and column-name variations from context and turns them into data. For clients with stable layouts, you can define dedicated rules to push accuracy even higher.
Yes. Once you register masters such as client IDs, product IDs and delivery destinations via CSV or API, it automatically matches the extracted data and outputs it in a form ready to load into your core system (Excel / CSV / API). A confirmation UI for fuzzy matches is also provided, so only the cases that require human judgment are checked manually.
For printed, fixed-format documents we achieve 95–99% recognition accuracy. Handwriting and low-quality scans are also raised to a practical level with AI correction plus a highlight-confirmation UI. Since you can build a human review workflow for critical fields, the accuracy of the final data can be guaranteed at 100%.
Data is stored and processed in domestic regions, and all communication is encrypted. The document data we read is never used for any external AI training. Access permissions and operation logs can be managed in detail.
A standard rollout takes 2–6 weeks. You can adopt it in stages: accuracy validation on your own document samples → trial operation → full operation. You can start small, beginning with just one client and one document type.
It varies with the volume of processing, the scope of integration and the degree of customization. We offer plans that let you start small. Get in touch and we will provide an estimate tailored to your operation.
