Workflows

Learn to build workflows in Draft & Goal: create a workflow, add and connect nodes on the canvas, and use patterns like linear flows, branches, and loops.

What is a workflow?

A workflow is a series of connected nodes that work together to accomplish a task. When you run a workflow:

  1. Data enters through input nodes
  2. Each node processes the data and passes it to the next
  3. Final results are output or saved

Think of it as a production line where each station (node) adds value to the product (data).

Workflow canvas with connected input, integration, code, and LLM nodes flowing left to right

Creating a workflow

You can also describe the workflow in natural language with the Workflow Generator — useful for a first draft before you refine nodes by hand.

Start a new workflow

  1. Click New Workflow in the Studio
  2. Give it a descriptive name (e.g., “SEO Content Audit”)
  3. You’ll see an empty canvas ready for nodes
Workflows list with the New Workflow button highlighted in the top-right corner

Add your first node

Every workflow needs at least one node:

  1. Click Add Node in the left sidebar
  2. Choose a node from the library
  3. Click to add it to the canvas
Node Library sidebar with Inputs tab open, search bar, and basic input nodes listed
Tip

Most workflows start with an input node (like Text Input) or a data source node (like Google Sheets Reader).

Connect nodes together

To pass data between nodes:

  1. Hover over the output point (right side) of a node
  2. Click and drag to create a connection line
  3. Drop on the input point (left side) of another node

Connections show the data flow direction with arrows.

Url node output port connected to the Web Scraper node input on the canvas

Workflow patterns

Linear workflow

The simplest pattern — data flows in a straight line:

Input → Process → Transform → Output

Example: Scrape a URL → Summarize with AI → Save to Google Sheets

Linear workflow with Url, Web Scraper, and LLM nodes connected in a straight line

Parallel workflow

Multiple branches process data simultaneously:

        ┌→ Branch A →┐
Input →│              │→ Merge → Output
        └→ Branch B →┘

Example: Fetch a URL → Extract both text AND images → Combine results

Parallel branches from Short Text and Url inputs through two LLM nodes into a Merge node

Conditional workflow

Different paths based on conditions:

        ┌→ Path A (if true) →┐
Input →│                      │→ Output
        └→ Path B (if false)→┘

Example: Check word count → If < 500, flag for review → Else, publish

IF/ELSE node branching to a Stop Node on match or to a Merge node on the default path

Loop workflow

Repeat actions for multiple items:

Input (list) → Loop [Process each item] → Collect results

Example: List of URLs → Loop through each → Scrape all pages

Loop container running Web Scraper, JSON Path Extractor, and LLM nodes per list item, then merging results

Workflow organization

Folders and smart views

On Builder → Workflows, use the workflow explorer on the left to file and filter workflows — similar to a file explorer tree on the left and the contents on the right.

Smart views (All workflows, Recently viewed, Favorites, My workflows, Dormant) are saved filters across your workspace. Folders let you group related workflows in a tree, with optional subfolders.

ActionHow
Create a folderNew folder at the top of the explorer, or New subfolder from a folder menu
File a workflowRow menu Move to folder, or select multiple rows and use the action bar
Remove from a folderMove to Uncategorized — the workflow is not deleted
Set folder visibilityFolder menu → Visibility & members — Private, Group, or Organization
Workflow explorer with smart views, folder tree on the left, and workflow table on the right
Warning

Folder visibility ≠ workflow access. Sharing a folder only controls who sees the folder structure and who can file workflows there. Each workflow still has its own share settings. A group folder can list four members while a workflow inside it remains private to its owner until shared separately.

See Teams & Collaboration for folder visibility levels and member management.

Naming conventions

Use clear, descriptive names:

TypeExample
Workflow“Weekly SEO Report Generator”
Nodes“Fetch Keywords”, “Calculate Scores”

Visual organization

Keep your canvas clean:

  • Left to right: Start inputs on the left, outputs on the right
  • Group related nodes: Keep similar functions together
  • Use spacing: Don’t crowd nodes together
  • Align nodes: Use the grid to keep things tidy
Workflow canvas organized left to right with inputs, processing nodes, and parallel output branches

Adding descriptions

Document your workflow for future reference:

  1. Click Settings in the top bar
  2. Add a Description explaining what the workflow does
  3. Add Tags to organize your workflows
Workflow Settings modal with Name, Description, Folder, Instructions, and Brand Kit toggle

Workflow settings

Timeout settings

Set maximum execution time to prevent runaway workflows:

SettingDescriptionDefault
Workflow timeoutMax total time30 minutes
Node timeoutMax per node5 minutes

Error handling

Choose what happens when a node fails:

  • Stop workflow: Halt execution immediately
  • Continue: Skip the failed node and continue
  • Retry: Attempt the node again (1-3 times)

Input configuration

Define what inputs your workflow accepts:

inputs:
  - name: url
    type: text
    required: true
    description: "URL to analyze"
  - name: depth
    type: number
    required: false
    default: 1

Running workflows

Manual execution

  1. Click the App button in the top bar
  2. Fill in any required inputs
  3. Click Generate
  4. Watch the real-time progress
App runner with input form fields, cost estimate, and Generate button

Viewing results

After execution:

  • Green nodes: Completed successfully
  • Red nodes: Failed (click to see error)
  • Grey nodes: Not executed

Click any node to see its output in the right panel.

Canvas during a run with green Done nodes, a blue Running node, and output previews on connected steps

Execution logs

Access detailed logs:

  1. Click Run History in the execution panel
  2. See timestamp, duration, and data for each node
  3. Debug issues with full request/response data
Activity Runs tab with status filters, search, and a table of past workflow executions

Workflow templates

Don’t start from scratch — use templates:

  1. Go to Workflows Library in the Studio
  2. Browse by category (SEO, Marketing, etc.)
  3. Click Use Template to create a copy
  4. Customize for your needs
Workflows Library page with category filters and featured workflow cards with Add Workflow buttons
Browse Templates

Explore 50+ ready-made workflows

Version history

Track how a workflow changed over time:

  1. From the workflow list, open row actions → History, or use History from the workflow menu on the canvas
  2. Review changes with timestamps and who made them
  3. Restore a previous version when you need to roll back

Version history is an editing aid — it does not replace folder visibility or Share permissions. See Workflow access & sharing.

Organize your workflow library

Keep the workspace easy to navigate for the whole team:

  • Use folders: Group related workflows under root folders with clear visibility (Organization / Group / Private)
  • Naming conventions: e.g. [Team]-[Project]-[Description]
  • Add descriptions: Document purpose and who owns the App
  • Archive instead of delete: Prefer archival patterns over hard deletes when you may need history later
  • Document complex logic: Use node names and comments so reviewers and teammates understand the flow

For folder visibility and sharing, see Workflow access & sharing. For the explorer UI, see Studio Interface.

Best practices

Start simple

Begin with a basic workflow and add complexity gradually:

  1. ✅ Build the happy path first
  2. ✅ Test with real data
  3. ✅ Add error handling
  4. ✅ Optimize for performance

Test incrementally

After adding each node:

  1. Run the workflow
  2. Verify the output is correct
  3. Then add the next node

Handle errors gracefully

  • Use Conditional nodes to check for empty data
  • Add Stop Nodes for clear error messages
  • Configure retry logic for flaky APIs

Optimize for performance

  • Minimize API calls: Cache when possible
  • Use parallel branches: Run independent tasks simultaneously
  • Limit loop iterations: Add safeguards for large datasets

Next steps