Beginner5 min

AI Features Overview

AI features help you build faster by generating workflows, forms, and integrations from natural language descriptions.

Available AI Features

Workflow Generation

Describe what you want, get a working workflow:

  • "Send Slack notification when form is submitted"
  • "Sync new Stripe customers to Airtable"
  • "Email weekly reports every Monday"

Form Generation

Create forms from descriptions:

  • "Contact form with name, email, and message"
  • "Event registration with date selection"
  • "Customer feedback survey"

API Import

Import APIs with AI assistance:

  • Paste API documentation
  • AI extracts endpoints
  • Auto-generates actions

Smart Suggestions

AI helps while you build:

  • Suggested next steps
  • Error explanations
  • Optimization tips

How It Works

1. Describe Your Goal

Enter a natural language description:

When a new user signs up, send them a
welcome email and notify the sales team

2. AI Generates

The AI analyzes your request and creates:

  • Appropriate nodes
  • Correct connections
  • Initial configuration

3. Review & Customize

Generated output is a starting point:

  • Review the structure
  • Adjust settings
  • Add missing details

4. Test & Deploy

Finish as normal:

  • Test the workflow
  • Make refinements
  • Publish when ready

AI Models

Available Models

| Model | Best For | Speed | |-------|----------|-------| | GPT-4 | Complex workflows | Slower | | GPT-3.5 | Simple tasks | Faster | | Claude | Detailed explanations | Medium |

Choosing a Model

  • Quick tasks: GPT-3.5 (faster, fewer credits)
  • Complex automation: GPT-4 (more accurate)
  • Explanations: Claude (better reasoning)

Credit System

AI features consume credits:

| Action | Credits | |--------|---------| | Generate workflow | 5-20 | | Generate form | 3-10 | | API import | 5-15 | | Suggestions | 1-2 |

Credits vary by complexity and model.

Best Practices

Writing Good Prompts

Be specific:

Bad: "Make a notification workflow"
Good: "When a form is submitted, send the
      submitter's email and name to #submissions
      channel in Slack"

Include details:

  • What triggers it
  • What data to use
  • Where to send output

Iterating

Start simple, then refine:

  1. Generate basic structure
  2. Test and identify gaps
  3. Add specifics manually
  4. Re-generate if needed

Understanding Limitations

AI generation works best for:

  • Common patterns
  • Well-documented services
  • Clear requirements

May need manual work for:

  • Complex business logic
  • Custom integrations
  • Edge cases

Privacy & Security

Data Handling

  • Prompts sent to AI providers
  • No workflow data in prompts
  • Results not stored by AI

Sensitive Information

Avoid including in prompts:

  • API keys or passwords
  • Personal data
  • Confidential business logic

Next Steps

Tags

aigenerationautomation