Make.com Claude Integration: Add Anthropic's AI to Any Scenario

Integrating Claude into Make.com scenarios unlocks powerful AI automation without code. From content generation and email classification to document analysis and customer support — Claude handles intelligent tasks while Make.com handles the orchestration. This guide covers the complete technical setup and the best workflow patterns.

MJK Supplies · Jun 10, 2026 · 11 min read

ShareX in f ↗

Make.com Claude Integration: Add Anthropic's AI to Any Scenario

Why Claude + Make.com

Make.com is the best visual automation platform for non-technical users. Claude is the best AI model for reliable, high-quality reasoning and instruction following. Together, they enable sophisticated AI automation accessible to any business user.

Claude's strengths that make it ideal for Make.com workflows:

200K context: Process entire documents without chunking
Precise instruction following: Returns exactly the format you specify
Low hallucination: Doesn't invent information you didn't provide
JSON responses: Reliable structured output for routing and data storage

HTTP Module Setup

Make.com doesn't have a native Claude module (as of 2026). Use the HTTP module:

Add "HTTP → Make a Request" module
Configure:

- URL: https://api.anthropic.com/v1/messages - Method: POST - Headers (add each individually): - x-api-key → your Anthropic API key - anthropic-version → 2023-06-01 - content-type → application/json - Body type: Raw - Content type: application/json - Request content:

{ "model": "claude-sonnet-4-6", "max_tokens": 1024, "messages": [ { "role": "user", "content": "{{1.your_input_field}}" } ] }

Storing the API Key Securely

Don't hardcode your API key in the HTTP module:

Make.com → Settings → Custom Variables
Add variable: ANTHROPIC_API_KEY with your key
Reference in the module header: {{ANTHROPIC_API_KEY}}

Or use Make's "Custom Auth" connection for centralised credential management.

Extracting Claude's Response

After the HTTP module, Claude's response JSON looks like:

{ "content": [ { "type": "text", "text": "Claude's actual response here" } ], "usage": {"input_tokens": 100, "output_tokens": 50} }

In the next module, reference the text as: {{HTTP_module_number.data.content[].text}}

To parse JSON from Claude (when you ask it to return JSON): Add "Tools → Parse JSON" module after the HTTP call. Set the JSON string to the content text field.

System Prompts

For consistent AI behavior, add a system prompt to every Claude call:

{ "model": "claude-sonnet-4-6", "max_tokens": 1024, "system": "You are an email classification assistant. For each email, return ONLY a JSON object: {\"category\": \"support|sales|billing|spam\", \"urgency\": \"high|medium|low\", \"summary\": \"one sentence\"} with no other text.", "messages": [{"role": "user", "content": "{{1.subject}}\n\n{{1.body}}"}] }

Dynamic Prompts

Build prompts from Make.com data using text fields:

Use a "Text aggregator" or simply build the prompt string inline:

{ "messages": [{ "role": "user", "content": "Write a personalised cold email for:\nName: {{contact.name}}\nCompany: {{contact.company}}\nTitle: {{contact.title}}\nOur product: [Product] helps [use case]. Write a 4-sentence email." }] }

Make.com's data mapping fills in the {{variables}} from previous modules.

Complete Workflow Examples

Lead enrichment and personalisation:

HubSpot: new contact created
Clearbit HTTP: enrich with company data
Claude HTTP: "Generate 2 personalised talking points and 1 likely objection for this lead: {{contact_data + company_data}}"
Parse JSON response
HubSpot: update contact notes with AI insights

Content article from brief:

Airtable: new content brief row
Claude HTTP: generate 2000-word SEO article from the brief
Claude HTTP (second call): generate meta title, description, social captions
WordPress: create draft with all content
Slack: notify content team

Support email AI draft:

Gmail: new email to support@
Airtable (search): find matching knowledge base articles
Claude HTTP: draft response using knowledge base context
Gmail: create draft reply
Slack: notify support team

Choosing Model Tier

Optimise cost by using the right Claude model:

Task	Model	Reason
Simple classification	claude-haiku-4-5	Fast, cheap
Email drafts	claude-haiku-4-5	Good enough, affordable
Complex analysis	claude-sonnet-4-6	Quality matters
Long documents	claude-sonnet-4-6	200K context
Research synthesis	claude-sonnet-4-6	Best reasoning