Find what you've uploaded
Browse, filter, and search across documents, notes, labels, and chats, and pick documents from plain English.
Once you've uploaded more than a few documents, the workspace gets dense. docAnalyzer gives you four browse surfaces (one each for documents, notes, labels, and chats) plus a natural-language picker that finds documents from a description.
The four lists
Every type of entity has its own page:
| Page | What's in it | What you can do |
|---|---|---|
| Documents | Everything you've uploaded | Search by name, filter by label or page count, multi-select, bulk delete |
| Notes | Notes you've written or saved from chat | Filter by label, sort by recency, multi-select, bulk delete |
| Labels | Your label list | Open New Chat on a label, spawn a chatbot, combine labels, delete |
| Chats | Every chat thread | Filter by dataset type (single doc / labelset / mixed / note / dataset-free), search by title |
All four share the same controls: text search, filters (by label, type, date range, and more depending on the page), multi-column sort, and multi-select for bulk actions. Pinned items always sort first.
The lists are workspace-scoped. Switching workspaces switches what you see. Multi-selection doesn't persist across page navigation: it's a per-session working set, not a saved view.
Selection shortcuts
- Click a checkbox to toggle a single row.
- Shift-click a second checkbox to select every row between the two, inclusive of both. The first row you clicked stays as the anchor, so successive shift-clicks keep extending the range from the same starting point.
- Select all at the top of the list selects every row that matches your current filters, not the full workspace. Clear filters first if you want everything.
Pick documents with natural language
On the documents page, the AI Search button next to "Select all" opens a modal. Type a plain-English description of what you want ("contracts with renewal terms longer than a year", "documents about Q3 sales performance", "anything that mentions employee onboarding") and docAnalyzer returns the matching set.
Apply the result with one of four set operations against your current selection:
- Add (union with your current selection)
- Keep only matches (intersection)
- Remove (subtraction)
- Replace (substitution)
Stacking searches lets you build a precise working set fast: first find X (replace), narrow with Y (keep only matches), add Z. Once the selection looks right, open a Focus on it or run a workflow.
See Pick documents with natural language for the longer version.
Info panel: details on what you select
Click any row in any list and the right-hand info panel switches to a detail view of that entity. Documents show metadata, label assignments, and the stages they went through during processing. Notes open in an inline editor. Labels show their member list. Chats show their dataset and title.
The info panel is for inspection, not for full reading. To read a document in full, open it in the document viewer.
What's next
- Organize with labels: turn a growing pile into something searchable.
- Combine sources into a dataset: feed a working set into a Focus.
- Pick documents with natural language: the deep dive on the AI Search button.