> ## Documentation Index
> Fetch the complete documentation index at: https://docs.asktable.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Canvas

> The Canvas is AskTable's AI-powered visual data analysis canvas. Build a complete workflow from data acquisition to analysis presentation using natural language and a node-based approach, making analysis visual, reusable, and collaborative.

## What is a Canvas

A Canvas is AskTable's visual data analysis canvas. Through AI-driven natural language interaction and a node-based approach, you build a complete workflow from data acquisition to analysis presentation.

**Core value:** Make the data analysis workflow visual, reusable, and collaborative.

## Why you need a Canvas

Traditional data analysis faces three challenges:

1. **Opaque process** - SQL queries, data processing, and chart generation are scattered and hard to trace
2. **No reusability** - Each analysis requires rewriting code, leading to redundant work
3. **Difficult collaboration** - Analysis logic is hidden in code, making it hard for team members to understand and take over

The Canvas solves these problems through AI-driven, node-based design:

* Describe your needs in natural language, and the AI automatically creates the corresponding analysis nodes
* Dependencies between nodes automatically build the data flow paths
* The entire Canvas is a complete analysis workflow that can be saved, shared, and reused

## Quick start

### Create your first Canvas

1. Click "Analytics Canvas" in the left navigation bar
2. Click the "Create Canvas" button
3. Once inside, describe your analysis needs in natural language in the chat box on the Welcome page, for example:
   * "Query this month's sales data and generate a trend chart"
   * "Import this Excel file and analyze sales by region"
   * "Search for China's 2024 GDP growth rate"
4. The AI (Autopilot) automatically understands your intent, creates the corresponding nodes, and renders them on the canvas

## Node types

The Canvas supports five node types:

| Node                | Description                                                                               | Use cases                                                   |
| ------------------- | ----------------------------------------------------------------------------------------- | ----------------------------------------------------------- |
| **Data node**       | Connect to internal enterprise databases and query and analyze data with natural language | Business data analysis, operational reports                 |
| **Chart node**      | Visualize data as charts                                                                  | Trend charts, proportion charts, comparative analysis, etc. |
| **Python node**     | Write custom Python code to process data                                                  | Complex calculations, data cleaning, custom analysis logic  |
| **Excel node**      | Import Excel / CSV files for instant analysis                                             | Ad hoc data analysis, external data import                  |
| **Web Search node** | Search public market data, industry reports, and trend analysis                           | Market research, industry analysis                          |

## Canvas editor

### Interface layout

* **Top toolbar** — Canvas naming, saving
* **Central canvas** — Node arrangement area, displaying nodes and their dependencies as a ReactFlow graph
* **Right panel** — Node details (data table / chart view) + conversation history (your interaction history with the AI)

### Ways to create nodes

* **Welcome chat input** — When the canvas is empty, describe your needs in natural language right on the Welcome page, and the AI creates nodes automatically
* **Right-click menu** — Right-click on an empty area of the canvas and choose "Query data", "Web Search", or "Import Excel"
* **Floating chat input** — Click the chat button on an existing node and describe your needs in the floating input; the AI creates a new node based on the context
* **File drag-and-drop** — Drag `.xlsx`, `.xls`, or `.csv` files directly onto the canvas to automatically create an Excel node

### Node operations

* **Drag nodes** — Move nodes on the canvas to adjust the layout
* **Click a node** — View data results, charts, and conversation history in the right panel
* **Node toolbar** — Each node offers quick actions: create chart, follow up, create Python node, rerun, delete
* **Multi-select** — Select multiple Data / Python / Excel nodes to "Follow up" or "Create Python node" in one click

### Automatic edge generation

The connections (edges) between nodes are **generated automatically based on data dependencies** — no need to manually drag and connect ports. When one node references another node's data, the system automatically draws a connection between them.

## Use cases

### Scenario 1: Periodic business reports

Describe your needs in natural language, and the AI builds the analysis workflow automatically:

1. Enter "Analyze this month's sales data, calculate year-over-year and month-over-month changes, and generate a trend chart"
2. The AI creates a Data node to query the data and a Chart node to generate the visualization
3. Next time, just rerun the Canvas to update the results automatically

### Scenario 2: Market trend analysis

Combine multiple data sources:

1. Enter "Search 2024 industry reports and compare them with our business data"
2. The AI creates a Web Search node to fetch industry data, a Data node to query your own data, and a Chart node to compare and present them

### Scenario 3: Ad hoc data exploration

Quickly analyze uploaded files:

1. Drag an Excel file onto the canvas
2. Enter "Analyze sales by region and generate a bar chart" in the floating chat box
3. The AI creates analysis nodes and charts based on the file data

## Relationship with other features

**Canvas vs Chat**

* Chat: One-time Q\&A, suitable for ad hoc queries
* Canvas: Reusable workflow, suitable for periodic, multi-step analysis

**Canvas vs Report**

* Report: Static document that presents analysis results
* Canvas: Dynamic workflow that continuously updates data

**Canvas vs Quick Query**

* Quick Query: Quickly query a single metric
* Canvas: Complex analysis workflow with multi-step processing

## Frequently asked questions

**Q: Can a Canvas be rerun?**
A: Yes. Click "Rerun" in the node toolbar to re-execute that node and update its results.

**Q: How do I share a Canvas with team members?**
A: A Canvas belongs to a project. Project members can view and edit it according to their permissions.

**Q: Is the data in a Canvas updated in real time?**
A: It depends on the data source. Data nodes query the latest data on each run, while Excel nodes and Web Search nodes need to be rerun manually.

**Q: How many nodes can a Canvas have at most?**
A: We recommend no more than 20 nodes. Split overly complex workflows into multiple Canvases.