The AI assistant is most accurate when it can search your organization’s own content. Indexing converts supported content into searchable information so the assistant can use the right policy, guide, or page when responding.
Not every source or file type is indexed in the same way. Connect the sources you need and keep the content structured, relevant, and up to date.
Where the assistant looks
The assistant searches indexed knowledge. It can use the following sources:
| Source | What is indexed | When to use it |
| Knowledge Management articles | The article body and supported attachments | Policies, how-to articles, and FAQs that you author and review in Applaud |
| Knowledge documents | Files you upload in Knowledge Management | Existing PDF, Word, Excel, and similar documents that you want to make searchable |
| SharePoint | Documents in connected libraries, site pages, and list items | Intranet policies and local procedures that your teams already manage in SharePoint |
For best results, use Knowledge Management articles as the source of truth for HR policies. Use SharePoint for content that your teams already manage there. Use uploaded documents when you need to make a file searchable without creating a full article.
Files that an employee attaches to a conversation can be read during that conversation. They are not added to the shared knowledge base.
Use Knowledge Management articles first
Knowledge Management articles generally give the assistant clearer, more policy-aligned content than a folder of mixed files.
Knowledge Management articles are:
- Structured with clear headings and short sections.
- Version-controlled and regularly reviewed.
- Targeted to specific audiences and countries, such as all employees, managers, or HR.
This structure helps the assistant retrieve relevant content and keep responses aligned with your organization’s policies.
SharePoint content
When the SharePoint knowledge connector is enabled, the assistant can use eligible content from the sites you connect.
It can index:
- Documents in libraries, using the supported file types listed below.
- Site pages, including intranet HTML converted to searchable text.
- List items with useful field content.
The sync process skips content that is unlikely to be useful as knowledge, including temporary Office files, thumbnails, system paths, empty templates, very old items, checked-out files, archives, and unsafe file types such as executables, scripts, and ZIP packages.
Your plan tier determines how many sites you can connect, how often they sync, and whether advanced parsing is available for larger or more complex files. See Plan quotas, sync frequency, and advanced parsing.
Supported content types
Once content is added to the knowledge index, the following formats can be searched.
| Category | File types |
| Documents |
.pdf, .doc, .docx, .ppt, .pptx, .odt, .odp, .rtf
|
| Spreadsheets |
.csv, .xls, .xlsx, .xlsb, .xlsm, .ods
|
| Web and text |
.html, .htm, .md, .txt, .json
|
| Images |
.jpg, .jpeg, .png, .tiff, .tif, .bmp, .gif, .webp
|
SharePoint document libraries typically sync the most common formats, including PDF, Word, Excel, CSV, HTML, Markdown, JPEG, and PNG. Other supported Office formats can be indexed when you upload them through Knowledge Management.
Unsupported content types
Some content cannot be indexed or requires additional preparation.
| Content type | Limitations |
| iFrame content | The assistant cannot read or search content that is available only inside an iframe, even when the surrounding page is indexed. |
| Google Docs, Sheets, and Slides | Native Google files are not indexed. Export them to PDF, Word, or another supported format first. |
| Videos and audio | Video and audio files are not searchable in the knowledge index. Upload a text transcript in a supported format if employees need to search the spoken content. |
| Archives and executables |
.zip, .rar, .7z, .exe, and similar packages are not indexed. |
| Microsoft OneNote and InfoPath | OneNote notebooks and InfoPath forms are not treated as knowledge content. |
Tips for accurate answers
- Prefer a short, well-structured article over a long, unstructured PDF when possible.
- Make titles, headings, and opening paragraphs specific. These help the assistant identify relevant content.
- Keep content current. Publish updates in Knowledge Management or allow SharePoint to sync updated content so outdated information is no longer used.
- Use audience and country targeting on Knowledge Management articles so the assistant can prioritize the right content for each employee.