Documents overview
The Informly document library is the source content your AI answers from, with status tracking, visibility controls, and search across every uploaded file.
Documents are the source content the AI answers from. Anything you want the assistant to know — product manuals, help articles, internal runbooks, FAQs — starts as a document in Informly. Without documents, the AI has nothing to retrieve and will fall back to generic answers. The library is where you upload, organize, and keep that content fresh.
Open the library from the sidebar by going to Documents.
What lives in the library
Every file you ingest becomes a document, regardless of where it came from. The library accepts four content types:
| Type | Notes |
|---|---|
| Best for manuals, guides, and exported articles. | |
| Word | .doc and .docx files. |
| Plain text | .txt files or pasted text. |
| Web pages | Crawled from a URL you provide. |
Each file is capped at 20 MB. For larger sources — books, long PDFs, multi-chapter manuals — split the content into smaller documents so retrieval stays accurate.
The processing lifecycle
A document doesn't become usable the moment you upload it. Informly runs every file through a short pipeline before the AI can retrieve from it.
| Status | What's happening |
|---|---|
| Pending | The file is queued. Nothing is happening yet. |
| Processing | Text extraction, chunking, embedding, and indexing are in progress. |
| Completed | The document is indexed and the AI can answer from it. |
| Failed | Something went wrong. The detail page shows the reason. |
Most documents finish in under a minute. Large PDFs and long web pages take longer. You can leave the page and come back — the status updates live.
During processing, the document is split into chunks — short passages, usually a few paragraphs each. Chunks are what the AI retrieves at query time, not whole documents. Smaller, well-structured source files chunk more cleanly and produce better answers.
Library stats
The top of the library page shows a quick health summary:
- Total documents in the org
- Completed, Processing, and Failed counts
- Storage used against your plan's limit
Use the failed count as a TODO list — every failure means a question the AI can't currently answer.
Visibility
Each document has a visibility setting that controls who in your organization can see it.
| Visibility | Who can see it |
|---|---|
| Public | Everyone in the organization. |
| Private | Only members with explicit access. |
Visibility is independent of whether a document answers in a widget — that's controlled by which documents you assign to the widget. See Documents → Managing for bulk visibility changes.
Searching the library
The library supports two search modes side by side:
- Full-text search matches the exact words you type — useful when you know the document title or a unique phrase.
- Semantic search matches meaning, not words — useful when you remember the topic but not the wording.
Both run against the same library and return the same document cards. See Documents → Managing for details.
Archiving and reprocessing
Two non-destructive actions keep the library tidy without losing work:
- Archive hides a document from the library and stops it appearing in chat. The file and its chunks are preserved — unarchive it any time.
- Reprocess re-runs the full extract-chunk-embed-index pipeline. Use it after you replace the underlying file or change ingestion settings.
Both actions are covered in Documents → Managing.
How documents reach the AI
A document in the library is not automatically available to every widget. You assign documents to widgets explicitly, which gives you per-widget control over what each assistant knows.
Upload the document
Add the file through uploading or web crawl. Wait for the status to show Completed.
Open the widget
Go to Widgets → (your widget) → Documents.
Assign it
Tick the document. The widget will start retrieving from it on the next message.
You can also pull documents in automatically from external systems — see Data sources.
What's next
Upload a document
Drag in a file, paste plain text, or trigger a web crawl.
Crawl a URL
Pull in a public page and turn it into a document.
Manage the library
Search, bulk-edit, reprocess, archive, and download documents.
Connect a data source
Sync documents automatically from Slack, Google Drive, and more.
Core concepts
The vocabulary you'll see throughout Informly — organization, project, widget, document, conversation, skill, voice, and handoff.
Uploading documents
Add documents to your Informly library by uploading a file, pasting plain text, or crawling a URL, and follow the processing pipeline to completion.