Document Indexing Specifications
The Windchill AI Assistant supports a wide range of document and business object types for both on‑premises and SaaS environments.
Document Types
The following table provides a list of document formats supported by the Windchill AI Assistant plugin:
|
Document Type
|
Document Name
|
MIME Type
|
||
|---|---|---|---|---|
|
CSV
|
Comma Separated Values
|
text/csv
|
||
|
DOCX/DOC/DOCM
|
Microsoft Word
|
application/msword
application/vnd.ms-word.document.macroenabled.12
application/vnd.openxmlformats-officedocument.wordprocessingml.document
|
||
|
EML
|
Email
|
message/rfc822
|
||
|
EPUB
|
Electronic Publication
|
application/epub+zip
|
||
|
GZ
|
GNU Zip compressed file
|
application/gzip
application/x-gzip
|
||
|
HTML
|
Hypertext Markup Language
|
text/html
|
||
|
JSON
|
JavaScript Object Notation
|
application/json
|
||
|
KML
|
Keyhole Markup Language
|
application/vnd.google-earth.kml+xml
|
||
|
MSG
|
Outlook Email
|
application/vnd.ms-outlook
application/msoutlook
|
||
|
ODP
|
OpenDocument Presentation
|
application/vnd.oasis.opendocument.presentation
|
||
|
ODS
|
OpenDocument Spreadsheet
|
application/vnd.oasis.opendocument.spreadsheet
|
||
|
ODT
|
OpenDocument Text
|
application/vnd.oasis.opendocument.text
|
||
|
PDF
|
Acrobat Document
|
application/pdf
|
||
|
PPTX/PPT/PPTM
|
Microsoft PowerPoint
|
application/vnd.ms-powerpoint
application/vnd.ms-powerpoint.presentation.macroenabled.12
application/vnd.openxmlformats-officedocument.presentationml.presentation
|
||
|
RTF
|
Rich Text Format
|
application/rtf
|
||
|
TXT/LOG
|
Plain text
|
text/plain
|
||
|
XLSX/XLS/XLSM
|
Microsoft Excel
|
application/vnd.ms-excel
application/vnd.ms-excel.sheet.macroenabled.12
application/vnd.openxmlformats-officedocument.spreadsheetml.sheet
|
||
|
XML
|
Generic XML and special XML variants such as Word XML or DITA XML
|
application/xml
text/xml
application/vnd.ms-wordml
application/vnd.ms-word2006ml
application/dita+xml
|
||
|
ZIP
|
ZIP compressed file
|
application/zip
application/x-zip-compressed
|
Business Object Types
The following table provides a list of business objects and their associated documents supported by the Windchill AI Assistant plugin:
|
Business Object
|
Primary Content
|
Representation(s)
|
Attachment(s)
|
Annotation(s)
|
||
|---|---|---|---|---|---|---|
|
WTDocument
|
✔
|
—
|
✔
|
—
|
||
|
EPM Document (CAD)
|
—
|
✔
|
✔
|
—
|
||
|
WTPart
|
—
|
✔
|
✔
|
—
|
||
|
Arbortext Dynamic Document
|
✔
|
—
|
✔
|
—
|
||
|
Change Request
|
—
|
—
|
✔
|
—
|
||
|
Change Notice
|
—
|
—
|
✔
|
—
|
||
|
Change Task
|
—
|
—
|
✔
|
—
|
||
|
Problem Report
|
—
|
—
|
✔
|
—
|
||
|
CAPA
|
—
|
—
|
✔
|
—
|
QMS (Quality Management System) Object Types
QMS object types manage quality-related processes across a product lifecycle and during manufacturing. For more information, see Windchill Quality Management Solutions.
The following table provides a list of QMS object types supported by the Windchill AI Assistant plugin:
|
Object Type
|
Purpose
|
Key Attributes
|
Supported by Windchill AI Assistant
|
|---|---|---|---|
|
QMS
|
|||
|
QMS Documents
|
Documents pertaining to quality management
|
Identical to WTDocuments in all aspects
|
Yes
Same level of support as WTDocuments (including attachments).
|
|
CAPA
|
|||
|
CAPA Request
|
Initiates the corrective and preventive action process
|
Lot/Serial Number, Batch Number, UDI
|
Yes (attachments only)
|
|
Regulatory Submissions and Regulatory Compliance Objects
|
|||
|
Regulatory Master
|
Tracks regulatory submissions and compliance
|
Agency, Submission ID, Status
|
Yes
Same level of support as Change Management objects.
|
Indexing Limits
The Windchill AI Assistant answers questions using the text content of your indexed Windchill documents. It uses this content to generate grounded answers and to cite the source documents in its responses. It does not analyze images, CAD geometry, or Windchill metadata as searchable content.
Review the following sections to understand what content the assistant can and cannot use when generating answers.
What is Indexed
The indexing process extracts and processes textual information only. The Windchill AI assistant indexes:
• Text content of the supported document types, including continuous text sections, titles, notes, captions, and tables, as well as any other text that is selectable in the document (such as text in tables, headers, and footers of a PDF).
• Alternative text or metadata that a supported document format stores to describe an embedded image.
• Text contained in a CAD representation, such as a drawing PDF, including annotations that are captured as text.
What is Not Indexed
The following content is not indexed and cannot be used by the assistant to generate answers:
• Images and charts, and other non-textual content. No optical character recognition (OCR) or image analysis is performed.
• Windchill metadata, such as file name, object name, version, folder (location), and context name, and Windchill attributes (IBAs).
• Native CAD geometry and CAD formats outside the supported document formats.
 |
Although Windchill metadata is not searchable, it is still used to ground and reference the assistant's answers. In a follow-up question, you can refer to a previously cited document by its file name, folder, or version to continue the conversation in context.
|
How to Phrase Your Questions
Because the assistant retrieves information from document content, not from metadata, phrase your questions in terms of the information you need, not the document you think contains it.
|
Do
|
Don't
|
|---|---|
|
Ask about the information, topic, or knowledge you need. Example: "What are the safety requirements for the hydraulic assembly?"
|
Ask the assistant to find or search for a specific document by name, folder, or version. Example: "Find the document named HydraulicSpec_v3.pdf."
|
|
Refer to a document that the assistant has already cited (by file name, folder, or version) to ask follow-up questions in context. Example: "In the document you just cited, what does section 4 say about pressure limits?"
|
Rely on metadata such as file name or folder location as the basis for a query.
|
Document Size Limits
The indexing limit for an individual document is 128 MB or 4 million characters, whichever is reached first.
|
|
Documents that exceed the supported limits may cause indexing failures. Use one of the following approaches to avoid such issues:
• Split large documents into smaller files.
• Prevent indexing of large documents by removing the indexing user’s access to those documents. For more information, see Configuring the Indexing User.
|
Indexing Performance
The indexing process depends on several factors, including the performance of your Windchill system, the network connectivity between Windchill and Microsoft Azure, and the capacity of the configured cloud resources.
Cloud resource capacity depends on your deployment model:
• For on‑premises environments, capacity depends on the Azure resources you configured.
• For SaaS environments, indexing uses PTC‑managed, scalable cloud resources.
Additionally, Microsoft Azure service performance may vary by region, time of day, or temporary service disruptions.
As a reference, indexing takes approximately two days for the following dataset:
• 150,000 documents
• Average document size: 500 KB
• Indexable content: 100% full text
|
|
In some cases, indexing may appear to be stalled due to a known Microsoft Azure indexing issue. When this occurs, indexing automatically resumes after approximately six hours. A permanent fix from Microsoft is expected in a future update.
|