Make.com AI Automation: Build Claude and GPT-Powered Scenarios

Make.com is one of the best no-code platforms for building AI-powered automations. Its visual interface makes it easy for non-technical teams to connect AI models to their business tools — without writing code. This guide covers every pattern for adding AI to Make.com workflows.

MJK Supplies · Jun 18, 2026 · 11 min read

ShareX in f ↗

Make.com AI Automation: Build Claude and GPT-Powered Scenarios

AI Integration Options in Make.com

Make.com supports AI integration through:

Native OpenAI module: The easiest way to add AI to Make.com workflows. Supports ChatGPT, DALL-E image generation, and Whisper transcription without configuring HTTP requests manually.

HTTP module for Claude/Gemini/others: Any AI API can be called via Make's HTTP module. Requires manual header and body configuration.

Anthropic module (Make marketplace): Third-party module for Claude API. Check the Make app marketplace for current availability.

OpenAI Module — Quick Setup

Add "OpenAI" module to your Scenario
Create a connection (enter your OpenAI API key)
Choose action: Create a Completion, Create an Image, Transcribe audio
Configure the prompt using data from previous modules

Simple example — email classification:

Trigger: Gmail new email → OpenAI: Create a Completion

System: Classify this email into: support, sales, billing, spam. Return only the category name. User: Subject: {{subject}} Body: {{body}}

→ Router: branch based on classification result

Claude API via HTTP Module

For Claude, use the HTTP module:

Module: HTTP → Make a Request

URL: https://api.anthropic.com/v1/messages
Method: POST
Headers:

- x-api-key: your Anthropic API key - anthropic-version: 2023-06-01 - content-type: application/json

Body type: Raw
Body:

{ "model": "claude-sonnet-4-6", "max_tokens": 1024, "system": "{{your_system_prompt}}", "messages": [{"role": "user", "content": "{{your_user_message}}"}] }

Extract the response: Claude's API returns JSON. Map {{data.content[].text}} to get the response text in the next module.

Common AI Automation Patterns

Content generation: Airtable (new content brief row) → Claude/OpenAI (generate article) → WordPress (create draft) → Slack (notify team)

Email classification and routing: Gmail trigger (new email) → OpenAI (classify: support/sales/billing) → Router → branch to Zendesk / HubSpot / billing system

Lead enrichment: HubSpot (new contact) → Clearbit (HTTP lookup) → Claude (generate personalised talking points) → HubSpot (update contact notes)

Document summarisation: Google Drive (new PDF) → Extract text → Claude (summarise with key points) → Notion (create summary page) → Slack (notify team)

Meeting notes: Zoom (meeting ended trigger via webhook) → Claude (extract decisions, action items, risks) → Google Docs (create formatted notes) → Calendar (create follow-up tasks)

Building an AI Support Bot

Complete workflow:

Trigger: Webhook (receive support request from website chat)
Data Store: Look up conversation history for this session
HTTP (Claude): With system prompt + history + new message
Data Store: Save updated conversation
Webhook Response: Return Claude's response to the chat widget

System prompt for support:

You are a helpful customer support assistant for [Company]. Use the information below to answer questions: {{knowledge_base_text}} If you cannot answer from this knowledge base, say "Let me connect you with our team" and set needs_human: true in your response. Always respond in JSON: {"response": "...", "needs_human": false}

Parse the JSON response: After the Claude HTTP call, parse {{data.content[].text}} as JSON to extract response and needs_human for routing.

AI Image Generation

With the native OpenAI module, generate images automatically:

Trigger (new product added to Airtable)
OpenAI: Create Image

- Prompt: Professional product photo of {{name}}: {{description}}. White background, studio lighting. - Size: 1024x1024 - Style: natural

HTTP (upload image): POST to your CDN or image storage
Airtable: Update record with image URL

Or use Claude to first write a better image prompt, then call DALL-E/Flux:

Claude: "Write a detailed DALL-E image prompt for: {{product_description}}"
OpenAI: Create Image using Claude's prompt

Transcript Processing with Whisper

Automatically transcribe and analyse audio:

Trigger: Google Drive (new audio file)
OpenAI: Transcribe Audio (Whisper)
Claude/OpenAI: Summarise transcript, extract action items
Google Docs: Create structured meeting notes
Email: Send summary to participants

This works for sales call recordings, meeting audio, podcast episodes, and more.

Error Handling for AI Workflows

AI API calls can fail or return unexpected output:

Retry on failure: Right-click the HTTP/OpenAI module → Set up error handling → Retry (3 times with 30s delay)

JSON parse error handling: If Claude returns non-JSON when you expect JSON, add a fallback via error route.

Response validation: After the AI module, add a filter: "Only continue if {{data.content[].text}} contains {" (basic JSON check before parsing).