The AI Assistant gives better answers when accessing straightforward, structured content. It relies on indexing, the process of turning documents into searchable information that the assistant can use to respond in real time.
When content is indexed successfully, employees get accurate, timely responses based on the right source. However, not all content types are indexed similarly, and some formats aren’t supported. This article explains which content works best with the assistant and how to optimise your content for consistent, reliable answers.
Why Knowledge Management content works best
While you can upload documents for the AI Assistant to index, we recommend connecting it to your Knowledge Management content for the best results. Articles created in the knowledge base are:
- Structured with clear headings and short paragraphs.
- Version-controlled and regularly reviewed.
- Targeted to specific audiences using filters.
This makes it easier for the assistant to process them and helps ensure responses are always relevant, up to date, and aligned with your HR policies.
Supported content types
The assistant indexes the following formats. Any text in these file types becomes searchable:
Content type | Details |
---|---|
PDF documents | The assistant indexes PDFs, making the text searchable and accessible. |
Word documents (.doc/.docx) | Word files are fully supported and analysed for relevant content. |
Plain text files (.txt) | All content in plain text files is searchable. |
CSV files | CSV data is indexed, making text fields easy to search and retrieve. |
Unsupported content types
Some content types can’t be indexed or require extra preparation:
Content type | Limitations |
---|---|
iFrame embeds | The assistant can’t read or search anything embedded via iframes. |
Google Docs | These aren’t indexed directly. Export as PDF or Word for indexing. |
Videos | Not indexed. Upload a text transcript in a supported format to make it searchable. |
Images | Not indexed. Extract any text into a plain text format first. |