IFS.ai Sales Quotation Automation

IFS.ai Intelligent Document Processing for Sales Quotation uses AI to extract data from uploaded Requests for Quotation (RFQs) and pre-fill Sales Quotation fields in IFS Cloud, reducing manual data entry for Sales Coordinators.

Manually keying in customer, header, and line data from each incoming RFQ is time-consuming and error-prone, especially for sales teams handling a high daily volume of quotation requests.

In practice, this means a Sales Coordinator can upload a customer's RFQ as a PDF or image, have the system read and extract the relevant data, review and refine the auto-populated header and line details, and create the Sales Quotation - instead of typing everything manually.

The feature provides pre-filled Sales Quotation header and line fields based on RFQ content, supporting faster, more accurate quotation intake and quicker response to customers. Output quality depends on the layout, legibility, and language of the uploaded document.

Who This Feature Is For and Why It Matters

This feature is designed for roles involved in the sales quotation intake process, including:

• Sales Coordinators responsible for registering incoming customer RFQs into IFS Cloud
• Sales Representatives responsible for following up on customer requests and turning them into quotations
• Sales operations leads responsible for high-volume quotation processing across multiple sites

Use this feature when you need to:

• Enter RFQ header and line data into a Sales Quotation quickly
• Reduce manual data entry errors when transcribing customer documents
• Process a high volume of incoming RFQs efficiently
• Create Sales Quotations for the main site and, where needed, for additional sites in one guided flow

By using this feature, you can:

• Have customer, header, and line data extracted and filled in automatically from PDF/image RFQs
• Spend less time on manual entry and more on customer engagement
• Review and adjust extracted data before the Sales Quotation is created
• Improve responsiveness and customer satisfaction through faster quotation turnaround

Where to Find It

Access this feature from the Scan Sales Quotation assistant in IFS Cloud.
Page: Scan Sales Quotation assistant
Product Area: Supply Chain Management
Navigation Path: Sales → Quotation → Scan Sales Quotation
Location on page: File upload section in Step 1 ("Scan Quotation") of the assistant

How the Feature Uses Data

When you upload an RFQ and trigger scanning, the assistant sends the document to the IFS.ai Machine Learning Service, which uses the Microsoft Document Intelligence Invoice (Prebuilt) Model to read the document and extract relevant information to populate fields on the Scan Sales Quotation assistant.

The feature reads:

• Customer-identifying data such as Tax ID / VAT Number (used as the primary key to match the customer against the Association Number), Supplier Name, Customer Name, Shipping Address Recipient, and Supplier Address Recipient
• Quotation header data (e.g., site context, coordinator, customer reference)
• Quotation line data presented in tabular form, including Sales Part identifiers, Sales Quantity, and unit of measure
• Delivery and other line-level details that can be inferred from clear, structured text

It also uses existing IFS Cloud reference data to match and categorize extracted values, including:

• The Customer master (active and prospect customers, with Tax ID as primary match)
• The Sales Part master and Sales Part Cross References, for line item identification
• The Site master, to validate the main site and any additional sites selected for the quotation

The feature does not use:

• Data from other Sales Quotations or unrelated records
• Customer documents to train external models — processing stays within the customer's Azure tenant boundary
• Information from expired customers or inactive Sales Parts (these are excluded from suggestions)
• Column header text in line tables (e.g., "Qty kgs" header content cannot be interpreted as UoM)

Supported Documents

• Supported formats: PDF and common image formats (BMP, JPG, PNG, and similar)
• Documents should be machine-printed, legible, and contain only one Request for Quotation per file
• Quotation lines must be in a tabular format with clear column separation; UoM and similar attributes should be in their own column rather than embedded in headers
• English-language documents are preferred and produce the best extraction results
• Only the pages with relevant header and line information should be scanned; multi-page terms and conditions should be skipped to avoid slow processing or timeouts

Not supported / unreliable: handwritten quotations, free-form emails, unformatted documents, multi-RFQ files, low-resolution or skewed scans, shadowed images, and faxed pages

Understanding the Output

After processing, the feature populates the Scan Sales Quotation assistant based on the content of the uploaded RFQ:

• Header fields (e.g., Site, Coordinator, Customer identification with Tax ID/Association Number, Scanned Association Number) are populated based on extracted data
• Line fields (e.g., Sales Part, Sales Quantity, UoM) are populated from the document's quotation-line table
• If the Tax ID cannot be found or matches multiple customers, candidate customers are surfaced based on Shipping Address Recipient, Supplier Address Recipient, Supplier Name, and Customer Name, with similarly-named customers shown at the top of the list
• Expired customers and inactive Sales Parts are excluded from suggestions and not shown in the list
• All populated fields are presented for review and edit before the Sales Quotation is created - the system is assistive, not autonomous
  Not all fields may be populated. Results depend on the quality, structure, and language of the uploaded document.

For example:

• Clean, English, table-based PDF RFQ from a known active customer with a valid Tax ID: customer is matched automatically, header fields and most line items (Sales Part, Quantity, UoM) are populated and ready for a quick review
• Low-resolution scan, multi-column line table with UoM embedded in the header (e.g., "Qty kgs"), or a non-English document: customer may need to be selected manually from similar-name suggestions, UoM may not be recognized, and several line fields may require manual completion

Enabling and Using the Feature

The feature is available when AI Services are enabled in your IFS Cloud environment and the Sales Quotation module is configured for the relevant site(s).

Prerequisites

• AI / Machine Learning Services must be enabled in IFS Cloud
• As of Release 24R2, an entry for Model Name = INVOICE must exist in the Machine Learning Pre-trained Models table, and the corresponding Machine Learning Services configuration must be selected as the Configuration ID
• The Customer must already exist in IFS Cloud (active or prospect)
• The Sales Parts referenced in the RFQ - and, if applicable, the Sales Part Cross References - must already be defined
• For multi-site quotation creation: a Customer must be linked to each company across the connected sites, and Sales Parts must be registered with identical names across those sites

Configuration Requirements

Ensure the INVOICE pre-trained model entry and Machine Learning Services configuration are completed as described in the Prerequisites (required from 24R2 onwards)
Ensure user defaults for Site and Coordinator are set where appropriate, so they can be preset in Step 1 of the assistant

Upload and Extract (Step 1: Scan Sales Quotation)
1.      Go to Sales → Quotation → Scan Sales Quotation
2.      Select the Site that the quotation is scanned for and the Coordinator (if not already preset by your user defaults)
3.      Select the RFQ file (PDF or image) to scan and click Next to trigger extraction via OCR + Machine Learning

Review and Refine (Step 2: Review Order)
4.      Edit scanned quotation header data — complete any header fields that were missing or not recognized, and verify the correct customer was identified before moving on
5.      Edit scanned quotation lines data — complete any missing line data, with particular attention to Sales Part and Sales Quantity
6.      Adjust any incorrectly interpreted values (quantities, parts, customer reference, delivery details) before continuing

Apply (Step 3: Enter Processing Parameters — optional)
7.      Optionally choose whether the quotation should be printed and/or emailed
8.      Optionally enter any additional sites for which the Sales Quotation should also be created, beyond the main site
9.      Click Finish to create the Sales Quotation(s) — with default processing parameters if Step 3 was skipped, or with the parameters you selected in Step 3

The feature does not automatically save data without review. You must review and confirm in Step 2 (and, optionally, Step 3) before the Sales Quotation is created.

You can upload additional RFQs and repeat the process for each. Each uploaded document is processed independently, and the scanned document is securely stored as a DOCMAN document linked to the resulting quotation.

Example Actions

Based on extracted results, you may:

• Create the Sales Quotation for the main site and immediately share it with the customer by email or print from Step 3
• Generate individual Sales Quotations across multiple sites in a single action, once a Customer is linked to each company and Sales Parts are aligned across sites
• Verify the recognized Customer using Tax ID/VAT Number, and disambiguate using Shipping/Supplier Address Recipient, Supplier Name, or Customer Name when the Tax ID is missing or shared
• Correct Sales Part or UoM values when the RFQ table embeds the UoM in column headers or uses non-standard layouts
• Skip pages containing only terms & conditions before uploading, to avoid slow processing or timeouts
• Report repeated misinterpretations or low-quality scan patterns to internal support so they can be reviewed via Solution Manager → Automation and Optimization → Machine Learning → Machine Learning Logs

Providing clear, legible, English-language RFQs in supported formats (PDF or high-resolution images) with well-structured line tables helps improve extraction accuracy.

Important to Know

• Extracted data is a suggestion and must be reviewed before the Sales Quotation is created
• The AI does not guarantee correctness or completeness - it is assistive, non-autonomous, with mandatory human-in-the-loop validation
• The AI does not replace manual review and professional judgment of the Sales Coordinator
• Output quality depends on document quality - unclear, low-resolution, skewed, handwritten, or unformatted documents may produce incomplete results
• The model supports only one RFQ per uploaded document; multi-RFQ files cannot be processed reliably
• Customer, Sales Part, and Cross Reference master data must be set up first; the feature does not create masters automatically
• For multi-site quotation creation, customers and Sales Parts must be consistently configured across sites or affected lines will be skipped and logged in quotation history
• If no data is extracted, verify that the document is in a supported format, contains only one RFQ, is legible, and (ideally) is in English
• Typical response time is around 20–30 seconds, depending on document complexity