Your First AI Automation
Connecting AI to Your Tools
Before building workflows, you need to connect your automation platform to both AI services and business tools. This lesson covers the connection process and best practices.
Two Types of Connections
1. AI Connections (The Brain)
These power the intelligent parts of your workflows:
| AI Service | What It Does | How to Connect |
|---|---|---|
| Built-in AI | GPT-5.4 Mini included free in Zapier, Make | No setup needed |
| OpenAI (ChatGPT) | GPT-5.4, GPT-5.4 Mini, custom models | API key from OpenAI dashboard |
| Anthropic (Claude) | Claude Sonnet 4.6, Claude Opus 4.6 | API key from Anthropic console |
| Google (Gemini) | Gemini 3.1 Pro, Gemini 3.1 Flash | API key from Google AI Studio |
2. App Connections (The Hands)
These perform actions in your business tools:
| Category | Popular Apps |
|---|---|
| Gmail, Outlook, Mailchimp | |
| CRM | Salesforce, HubSpot, Pipedrive |
| Communication | Slack, Microsoft Teams, Discord |
| Documents | Google Docs, Notion, Airtable |
| Project Management | Asana, Monday, Trello, ClickUp |
| Storage | Google Drive, Dropbox, OneDrive |
Connecting AI Services
Option 1: Use Built-in AI (Easiest)
Both Zapier and Make include AI capabilities at no extra cost:
Zapier:
- "AI by Zapier" action—uses GPT-5.4 Mini
- Available on all plans including free
- No API key needed
Make:
- OpenAI and AI modules included
- "Make AI" features in premium plans
- Maia natural language builder
Option 2: Bring Your Own API Key
For more control, power, or specific models:
STEP 1: Create an account
→ OpenAI: platform.openai.com
→ Anthropic: console.anthropic.com
→ Google: aistudio.google.com
STEP 2: Generate API key
→ Navigate to API Keys section
→ Create new key
→ Copy immediately (shown once)
STEP 3: Add to automation platform
→ Find OpenAI/Claude/Gemini in app connections
→ Paste your API key
→ Test the connection
API Key Security Checklist
Your API key is like a password. Protect it:
| Do | Don't |
|---|---|
| Store in platform's secure credentials | Share in emails or messages |
| Set usage limits in provider dashboard | Leave unlimited spending enabled |
| Rotate keys periodically | Use the same key everywhere |
| Use separate keys for test/production | Commit keys to code repositories |
Connecting Business Apps
Most connections follow OAuth—you just click and authorize:
STEP 1: In your automation platform, click "Add Connection"
STEP 2: Select the app (e.g., Gmail, Slack, HubSpot)
STEP 3: Click "Sign in with [App]"
STEP 4: Grant permissions the automation needs
STEP 5: Connection is saved for all future workflows
Permission Best Practices
When granting access, consider:
| Permission Level | When to Use |
|---|---|
| Read only | When automation just reads data (monitoring, reporting) |
| Read + Write | When automation creates/updates data (most common) |
| Full access | Only if automation needs admin-level actions |
Tip: Most platforms let you create "test" or "development" credentials. Use these while learning, then switch to production when ready.
Managing Connections
Connection Health Dashboard
Check your connections regularly:
| Status | Meaning | Action |
|---|---|---|
| Connected | Working normally | None needed |
| Needs reauthorization | Token expired | Re-authenticate |
| Error | Connection broken | Check credentials, reconnect |
Organizing Multiple Connections
As you scale, you may have multiple accounts:
Example: Marketing Team Setup
├── Gmail (marketing@company.com)
├── Gmail (newsletter@company.com)
├── Slack (Marketing Workspace)
├── HubSpot (Marketing Account)
└── OpenAI (marketing-automation key)
Name your connections clearly to avoid confusion.
Cost Awareness
AI API Costs
AI services charge per use. Typical costs (2026):
| Model | Approximate Cost | Best For |
|---|---|---|
| GPT-5.4 Mini | $0.75 per 1M input tokens | Most tasks, default choice |
| GPT-5.4 | $2.50 per 1M input tokens | Complex reasoning |
| Claude Sonnet 4.6 | $3.00 per 1M input tokens | Long documents, nuance |
| Gemini 3.1 Flash | $0.075 per 1M input tokens | High volume, budget |
Tokens explained: ~750 words = 1,000 tokens. A typical email analysis uses 500-1,000 tokens.
Setting Spending Limits
Always set limits in your AI provider dashboard:
OpenAI Dashboard → Settings → Limits
- Set monthly maximum: $20 (start low)
- Enable email alerts at $10
Anthropic Console → Settings → Spend Management
- Set monthly limit: $20
- Get notified at 80% usage
Quick Connection Checklist
Before building your first workflow:
- Create accounts on your chosen AI providers
- Generate and securely store API keys
- Connect AI services to your automation platform
- Connect essential business apps (email, CRM, Slack)
- Test each connection with a simple action
- Set spending limits on AI services
Next: Time to build. Let's create your first working AI automation step by step. :::