Skip to main content
DeepState documentation

Investigation analysis on a canvas you control

DeepState is a desktop application for mapping investigations: people, organizations, events, places, documents, posts and the connections between them, on an infinite canvas. An AI research assistant searches, scrapes and builds the board alongside you. Everything is stored in a local database on your machine and can be shared with colleagues end-to-end encrypted.

What you get​

How the app is laid out​

The window holds two things: the agent, a chat with the research assistant scoped to the open board, and the canvas, where nodes live. Three layouts arrange them (switch from the bottom bar, Settings → Appearance, or Ctrl+1/2/3):

  • Merged - the agent and the canvas side by side as two panes.
  • Agent - a full-window conversation over a frosted view of the live board.
  • Board - the canvas fills the window and the agent is an ask bar at the bottom.

On the canvas, a toolbar across the top creates nodes and holds Search, Tidy, Fit, Collapse, the narrative filter and the Case file (the board's actors, organizations and narratives). Nodes open in a large node sheet over the board. A timeline can be pulled up along the bottom, and the status bar shows your license tier, connected integrations and the layout switch.

Editions​

PersonalProEnterprise
Boards5UnlimitedUnlimited per seat
Core AI tools (web, Wikipedia, entities, social, narratives, maps, records)✓✓✓
Vision: OCR, image analysis, images in chat—✓✓
Person search, YouTube transcripts, historical newspapers—✓✓
RSS feed creation by the AI—✓✓
Custom JavaScript tools—✓✓
DeepState AI monthly budget$2$8$8 per seat
Collaboration (LAN, relay, folder)✓✓✓
Seat management & usage dashboard——✓

Details in Licensing and Enterprise.

Where to next​

  • New here? Follow the Quick start - it takes about ten minutes and ends with a researched board.
  • Want to know what the assistant can actually do? The tools reference lists every tool with what it needs.
  • Working with others? Read Collaboration before you share anything.