CATO Chat

Overview

CATO Chat is an AI-powered interface that allows you to query datasets, generate visualizations, create reports, search biomedical literature, and perform statistical analyses — all through natural language. The assistant executes Python and R code in a secure sandbox, produces publication-quality figures, and can generate downloadable file artifacts like CSV, Excel, and DOCX reports.

Starting a Chat Session

1. Open your project and navigate to the Chat tab
2. CATO will greet you and list available datasets
3. Type your question or request in the input field
4. Press Enter to send your message

Chat Modes

Plan mode is recommended for complex multi-step analyses where you want to review the approach before execution. Toggle it with Shift+Tab or press Esc to dismiss.

Referencing Datasets with @-Mentions

You can reference specific datasets directly in your messages using the @ symbol. Type @ followed by the dataset name to get an autocomplete dropdown of matching datasets.

Slash Commands

Type / in the input field to access quick commands. An autocomplete dropdown will appear with available commands:

Example Prompts

Here are effective ways to interact with CATO across different capabilities:

Visualizations Panel

When CATO generates charts or plots, they appear both inline in the conversation and in the Visualizations side panel on the right. The panel opens automatically when new figures are created.

• Gallery view — All figures from the current conversation in a scrollable grid
• Click to navigate — Click any figure in the panel to scroll to its corresponding message
• Edit — Click the pencil icon to open the Figure Editor (Base plans and above)
• Version badge — Figures that have been edited show a version badge with dropdown to browse history
• Download — Download individual figures as PNG files
• Auto-scroll — The panel scrolls to new figures as they stream in

Editing Figures

CATO includes a no-code Figure Editor that lets you customize visualizations directly in chat. Click the pencil icon on any figure in the Visualizations panel or inline in the conversation to open the editor. This feature is available on Base plans and above.

Editor Tools

Properties Panel

When you select an element, the Properties panel appears on the right. Edit text content, font family, font size, weight, and style. Change fill color, stroke color, stroke width, and opacity. The color picker includes an Okabe-Ito colorblind-friendly palette (8 colors) and tracks your recent color choices.

Undo, Redo, and History

• Undo/Redo — Ctrl+Z / Ctrl+Shift+Z (up to 50 steps)
• Save — Creates a new version. Each save is a numbered snapshot with an optional edit summary
• Version history — Click the version badge on a figure to browse all versions and revert to any previous one
• Reset to Original — Reverts all edits while preserving the full version history

Export

• PNG — High-resolution export at 2x DPI (300 DPI equivalent)
• SVG — Native vector format, ideal for publications

Additional Keyboard Shortcuts

File Artifacts

CATO can generate downloadable files directly in the chat. These appear as clickable cards in the conversation with file type, name, and size information.

Biomedical Tools

CATO provides access to over 200 biomedical tools across 20+ categories, available directly in chat:

Simply ask in natural language — CATO will automatically select and invoke the appropriate biomedical tool for your query.

3D Molecular Viewer

Ask CATO to visualize a protein or molecule (e.g., "Show me the 3D structure of PDB 1CRN") and it renders an interactive 3D viewer directly in the chat. Supported formats include PDB, SDF, MOL2, XYZ, and CIF. You can rotate, zoom, and switch between display styles:

Large structures (>32KB) stream progressively with a loading indicator. The viewer uses chain-based coloring by default and supports spin animation and custom background colors.

Code Execution Sandbox

CATO executes code in a secure sandboxed environment. Both Python and R are supported with a comprehensive library of pre-installed packages.

You can ask CATO to use a specific language: "Use R to create a ggplot of..."or "Write a Python script that...". CATO will show the executed code alongside the results.

Understanding Responses

CATO provides structured responses that may include:

• Text explanations — Context, interpretation, and methodology
• Code blocks — The Python or R code that was executed
• Visualizations — Charts and plots rendered inline
• File artifacts — Downloadable CSV, Excel, DOCX, or PPTX files
• Follow-up suggestions — Recommended next steps for your analysis
• Tool results — Outputs from biomedical tools and literature searches
• Assumptions — After analyses, CATO states key assumptions: method choice, data handling, grouping, and thresholds

Interview and Follow-Up System

For complex or ambiguous requests, CATO may ask clarifying questions before proceeding. This helps ensure that the analysis matches your specific requirements. The questions appear as a structured form in the conversation. Select your answers and click Submit to continue.

Once you submit answers, the question form clears from the screen immediately and CATO proceeds with your analysis. If you navigate away after submitting, CATO continues processing in the background. The questions will not reappear when you return.

After completing an analysis, CATO will suggest relevant follow-up questions to help you explore further.

Session Management

Chat sessions are saved automatically and persist across browser sessions. You can:

• Start a new session using the + button in the conversation history
• Switch between previous sessions via the conversation sidebar
• Sessions retain all messages, figures, and file artifacts
• CATO remembers context from earlier in the same session

Background Stream Persistence

If you navigate away from Chat while CATO is generating a response, generation continues as a background task. When you return to the conversation, the chat automatically reconnects to the live stream and replays any buffered events. The streaming indicator (cancel button and progress display) appears as soon as background generation is detected.

If the response finishes before you return, the completed message is loaded from the database when you navigate back. You can cancel a background response at any time using the cancel button, which appears as soon as background generation is detected.

Sharing Conversations

You can share any chat conversation with others by generating a public link. Click the share button in the conversation header to open the sharing dialog.

• Read-only access — Recipients see the full conversation including messages, figures, code blocks, and file artifacts
• Token-based links — Each share link has a unique token for access control
• Revoke anytime — Delete a share link from the same dialog to revoke access instantly

Peer Review Phases

For complex analyses, CATO may use an internal peer review process. When this happens, the chat journal displays structured phase containers showing the progression through four stages:

Each phase shows the work done at that stage, helping you understand how CATO arrived at its final answer. A revised analysis section may appear with improved results after the verification step identifies issues.

Chat Keyboard Shortcuts

Literature-to-Chat Bridge

Papers curated in the Literature Intelligence module can be sent directly to Chat for AI-powered analysis. When sending papers, you choose a prompt type that determines how the chat input is pre-filled. The literature packet loads automatically and the chat input is disabled until loading completes to prevent empty sends.

Prompt Types

When sending papers to Chat, choose a prompt type that matches your intent:

Quick-Action Buttons

When papers arrive in Chat, quick-action buttons are also available:

Example Prompts with Literature

Tips for Effective Prompts