AI and Search
This section explains the AI product model, from quick PDF chat to workspace-scoped actions and persistent indexed retrieval across folders and workspaces. It focuses on the practical difference between asking one question once, asking the product to do something, and building a reusable document intelligence workflow.
What this section covers
- Immediate Q and A against PDFs.
- Workspace-scoped action execution for file, folder, intake, workflow, and AI operations.
- Persistent indexed retrieval across folders and workspaces.
- Saved searches, alerts, compare mode, and reusable retrieval workflows.
- Operational screens such as indexed files, automations, usage monitoring, and shared chats.
How to choose the right AI surface
Use AI PDF Chat when you need a fast answer from a narrow uploaded document set. Use AI Action Assistant when the request is operational, such as creating a folder, generating a guest upload link, moving files, running indexing, or explaining a plan or permission restriction inside the active workspace. Use AI Document Assistant when you need a repeatable, governed knowledge layer across folders or workspaces. Use compare mode when the question is about differences rather than isolated facts. Use alerts and saved searches when the same information need recurs over time.
Start with
Use conversational Q and A against uploaded PDFs.
Use natural language to run allowed workspace actions safely.
Work with indexed knowledge and recurring retrieval.
Turn retrieval into an operational system.
Manage indexed files, automations, usage, and shared chats.
What good AI workflows look like
- Action requests stay within the current workspace and current user permissions.
- The folder scope is deliberate and business-meaningful.
- Source files are readable and OCR quality is acceptable.
- Saved searches or prompts reflect recurring business questions, not one-off improvisation only.
- External sharing uses shared chats or intake links instead of broad workspace access.