Make.com Webhook to Claude: Receive, Parse, Respond (Working Example, 2026)

A webhook that receives data, sends it to Claude for analysis, and returns a response — that’s one of the most useful patterns in automation. It turns any external system (a form, a website, a mobile app) into something AI-powered.

This post walks through a complete working example. By the end, you’ll have a webhook URL that takes a JSON payload, asks Claude a question about it, and returns the answer instantly.

What We’re Building

Scenario: a webhook URL that accepts POST requests with any text content. It sends that text to Claude with a summarisation prompt, gets the summary back, and returns it in the HTTP response.

Practical uses: your website can POST a user’s message, get an AI-analyzed response back, and display it. A mobile app can do the same. Another automation platform can call it as a microservice.

Response time: about 2 seconds end-to-end.

Step 1: Create the Webhook Trigger

In Make.com, create a new scenario. For the first module:

1. Click + → search Webhooks → select Custom webhook
2. Click Add to create a new webhook
3. Name it claude-summariser (or whatever makes sense)
4. Click Save

Make.com shows you a URL like https://hook.eu1.make.com/abc123.... Copy this. This is the URL external systems will POST to.

5. Click Re-determine data structure

Now we tell Make what data to expect. Open a second browser tab and use a tool like Postman or cURL. Send a POST request to your webhook URL:

curl -X POST https://hook.eu1.make.com/abc123... \
  -H "Content-Type: application/json" \
  -d '{"text": "Sample text to summarise. The quick brown fox jumped over the lazy dog. This is a placeholder sentence to demonstrate the structure."}'

Back in Make.com, the webhook module should show Successfully determined and you can click the output bubble to see the text field.

Step 2: Send to Claude

Right-click the webhook module → Add module → search Anthropic Claude → pick Create a Prompt.

Configure:

• Connection: your existing Claude connection (or create one with your Anthropic API key)
• Model: claude-haiku-4-5
• Max Tokens: 300
• System Prompt:

You are a summarisation assistant. When given any text, you return a concise one-sentence summary of the key message. Return ONLY the summary — no preamble, no "Here's the summary", just the sentence itself.

• Messages:
• Click Add item
• Role: user
• Content: click the field → map {{1.text}} from the webhook output

Click OK.

Step 3: Return the Response

Here’s the critical bit for webhooks: by default, Make.com just logs the output and ends. To return a response to whoever POSTed the webhook, we need to add a Webhook Response module.

Right-click the Claude module → Add module → search Webhooks → select Webhook response.

Configure:

• Status: 200
• Body: map {{2.content[].text}} from the Claude module (the actual summary text)
• Headers: optional — you can add Content-Type: text/plain if the caller expects it

Click OK.

Step 4: Enable Immediate Response

For the webhook to actually return Claude’s answer (not just a generic “accepted” message), you need to configure this at the scenario level.

Click the Scheduling clock icon (bottom left):

• Make sure the scenario is set to Immediately (not “On schedule”)

Click the ON/OFF toggle to ON.

Now your webhook is live.

Step 5: Test It

Back in your Postman or cURL, send the same request again:

curl -X POST https://hook.eu1.make.com/abc123... \
  -H "Content-Type: application/json" \
  -d '{"text": "Parag is a Sydney-based solo entrepreneur running a premium BMW electric chauffeur service. He uses Claude and Make.com to automate his bookings. In 2026 he launched an education platform teaching other business owners the same stack."}'

Expected response (plain text):

Parag runs a premium electric chauffeur service in Sydney and launched a 2026 education platform teaching others the AI automation stack he built.

That text came from Claude, routed through Make.com, in about 2 seconds. Congratulations — you have a working AI microservice.

Making It Robust

The basic flow works, but production webhooks need error handling. Three patterns I always add:

1. Handle Empty/Missing Input

Add a Router module after the webhook. First branch: if {{1.text}} is empty, go straight to a Webhook Response with status 400 and message “text field required”. Second branch (default): continue to Claude.

2. Handle Claude Errors

Right-click the Claude module → Add error handler → Break → set retries to 3 with a 5-second delay. This catches transient rate limit or server errors automatically.

If Claude still fails after retries, add a Webhook Response returning status 500 with a message, so the caller knows it failed.

3. Cap Input Size

Claude has a token limit. A 100,000-character text will fail or cost a lot. Add a filter after the webhook: only continue if {{length(1.text)}} < 20000. Otherwise return a 413 “payload too large” response.

Extending the Pattern

Once this base webhook works, you can adapt it for dozens of use cases:

• Classification webhook: swap the system prompt to “Classify as spam, sales, or support. Return one word.”
• Translation webhook: “Translate the text to French.”
• Sentiment webhook: “Return positive, negative, or neutral.”
• Extraction webhook: “Return a JSON object with name, email, phone extracted from this text.”

Same structure, different prompt, different response shape. That’s the power of this pattern.

Security Considerations

Public webhooks are, well, public. Anyone with the URL can call them, and each call costs you money in Claude credits. Two things to add before going fully public:

1. Require an Auth Header

After the webhook module, add a Filter: only continue if {{1.headers.authorization}} equals a secret value (e.g. Bearer your-secret-key). Set the same header on any legitimate caller.

2. Rate Limiting

Make.com doesn’t have built-in rate limiting on webhook triggers, but you can set a Max executions per minute in your scenario settings. This caps how many times per minute the scenario fires, protecting you from a flood of expensive API calls.

What This Unlocks

This pattern turns Make.com into a flexible AI backend for your existing apps. Without building any infrastructure, you can:

• Add AI to a no-code app (Bubble, Softr, Glide) via webhooks
• Let your website’s JavaScript call this endpoint for real-time AI responses
• Chain this with other services: “your form submits here → we process → we POST the result to Zapier → email notification”

Next Steps

If you want to see more complex webhook patterns — multi-step processing, branching logic, webhook-based approval flows, secured webhooks — that’s all in the Implementation Blueprint ($29).

For a gentler intro if you’re still new to Make.com, the free Quick Start builds a simpler Gmail-triggered scenario first, then extends it.

---

Questions? Reply to hello@theaigeneration.co.

Last updated: 20 April 2026.