How to Connect Claude API to Make.com: Step-by-Step Guide (2026)

If you’ve landed here, you probably want the same thing I wanted a year ago: get Claude answering questions, summarising emails, or extracting data inside a Make.com scenario — without writing a single line of code.

Good news: it takes about 10 minutes.

Bad news: the official docs skip over three things that will cost you an afternoon if you don’t know about them. This guide covers the full setup — and the stuff that breaks.

Who I am, briefly: I run Make.com scenarios in production for my own chauffeur business (Ospipo). Everything in this guide, I’ve done a dozen times. No theory, no filler.

Table of Contents

1. What you’ll need
2. Step 1: Create an Anthropic account
3. Step 2: Add credit to your Anthropic account
4. Step 3: Generate a Claude API key
5. Step 4: Create the Make.com connection
6. Step 5: Build your first Claude module
7. Common errors and how to fix them
8. What does it cost in practice?
9. Next steps

What you’ll need

Before you start, make sure you have:

• A Make.com account (free tier works fine to test)
• An email address you can verify
• A credit or debit card (you’ll need to add at least $5 credit to Anthropic — there is no free tier for the Claude API)
• About 10 minutes

That’s it. No coding experience required.

Step 1: Create an Anthropic account

Go to console.anthropic.com and sign up. You can sign in with Google or use email + password.

After sign-up you’ll land on the Anthropic Console dashboard. This is where you manage everything — billing, API keys, usage, rate limits.

If you’ve signed up before to use Claude.ai (the chat interface), that’s a different account type. The Console is for developers and API usage. Same login works for both, but billing is separate. Claude.ai is a subscription; the API is pay-as-you-go.

Step 2: Add credit to your Anthropic account

This is the step most tutorials skip, and then people wonder why their API key doesn’t work.

The Claude API is not free. You pay per token (roughly: per word of input and output). You need to add credit before you can make a single API call.

Here’s how:

1. In the Anthropic Console, click your profile icon (top right)
2. Click Billing → Plans
3. Click Buy Credits and add at least $5 (the minimum)
4. Add your card and confirm

$5 is enough for thousands of small API calls — we’ll cover real-world costs in the section below.

What happens if you skip this step: Your API key will be created fine, but every call from Make.com will return an error like “credit balance is too low” or “your account does not have API access.” It’s a pure billing issue, not a code issue.

Step 3: Generate a Claude API key

Now the important bit.

1. In the Anthropic Console, click your profile icon → API Keys
2. Click Create Key
3. Give it a descriptive name. I recommend something like make-com-production or make-scenario-1. This helps later when you have multiple keys and need to revoke one without breaking everything.
4. Click Create Key
5. Copy the key immediately and paste it somewhere safe — a password manager, or for now a Notes app. Anthropic shows it once. If you close the dialog, the key is gone forever and you’ll have to generate a new one.

The key looks like this: sk-ant-api03-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx. Treat it like a password — anyone with this key can spend your credits.

Step 4: Create the Make.com connection

Now we switch to Make.com.

1. Log into make.com and open any scenario, or create a new one
2. Click the big + button to add a module
3. In the search box, type Claude or Anthropic
4. You’ll see “Anthropic Claude” in the list — click it
5. Choose the module Create a Prompt (this is the simplest one to start with). There’s also Create a Message and Simple Text Prompt — we’ll cover those in a separate post.
6. When the module opens, the first field is Connection. Click Create a connection
7. In the dialog: - Connection name: give it something recognisable, e.g. Claude Production - API Key: paste the key you copied in Step 3
8. Click Save

If the key is valid and your Anthropic account has credit, the connection will save without error. If it fails, it’s almost always a billing issue — see Common errors below.

Step 5: Build your first Claude module

With the connection made, fill in the module:

• Model: pick claude-haiku-4-5 to start. It’s the cheapest, fastest Claude model — good enough for most automation tasks. You can switch to Sonnet or Opus later if you need more reasoning power.
• Max Tokens: set this to 1024 for testing. This caps the response length and protects you from runaway costs.
• System Prompt: optional — this is where you give Claude its role. Example: You are a helpful assistant that summarises customer emails in one sentence.
• Messages: add one message with Role: user and Content: type or map any test text, e.g. “What is 2 + 2?”

Click OK to save the module, then click the Run once button in the bottom left of the Make canvas.

If everything is wired up, you’ll see the module turn green and the output bubble will contain Claude’s response. Congratulations — you’ve just made your first Claude API call from Make.com.

Common errors and how to fix them

Here are the errors I hit when I first set this up, and what they actually mean.

401 Unauthorized or Invalid API key

Your API key is wrong. Either you copied it incorrectly (extra space at the start or end is a classic), or you’re using a key from a different Anthropic workspace. Go back to the Console, check the key, regenerate if needed.

400 credit_balance_too_low or Your credit balance is too low to access the Anthropic API

You haven’t added credit to your Anthropic account, or you’ve used it all up. Go to Billing → Buy Credits and top up.

429 rate_limit_error: This request would exceed your rate limit

This one trips up a lot of people. Anthropic enforces rate limits by tier — new accounts start on Tier 1, which is fairly low (around 50 requests per minute for most models, as of 2026). Claude returns 429 if you exceed it.

Three fixes:

1. Slow your scenario down. In Make, open the module settings → Scenario settings → increase the sequential processing delay, or set max number of cycles per execution lower. For loops, add a Sleep module between iterations.
2. Upgrade your Anthropic tier. Tier 1 automatically moves to Tier 2 once you’ve spent $5 and the account is 7 days old. Higher tiers happen automatically as you spend more — see Anthropic’s rate limits page for the current ladder.
3. Switch to a lighter model for bulk work. Haiku has a higher throughput than Sonnet or Opus.

500 api_error or 529 overloaded_error

These are Anthropic’s servers having a bad moment — not your fault. Add error handling in Make: right-click the module → Add error handler → use Break with retries. Set it to retry 3 times with 30-second intervals. Ninety-nine percent of the time the retry works.

The connection saves but the module returns nothing

Usually means your Messages array is empty or malformed. Claude needs at least one user message to respond. Check you’ve added a message, picked a role, and put some content in.

What does it cost in practice?

Honest numbers, from my own Make.com bill:

As of April 2026, Anthropic charges (per million tokens, input/output):

• Claude Haiku 4.5: roughly $1 input / $5 output
• Claude Sonnet 4.6: roughly $3 input / $15 output
• Claude Opus 4.7: roughly $15 input / $75 output

(Always check the official pricing page for current numbers — these change.)

For a sense of scale: a typical Make.com scenario that summarises an email is maybe 500 input tokens + 150 output tokens. On Haiku, that’s about $0.0013 per run — roughly one-tenth of a US cent. Run it 1,000 times a month and you’ve spent $1.30.

Where costs get real:

• Large document processing (10,000+ tokens per call)
• High-frequency workflows (every incoming email, every minute)
• Using Opus for everything when Haiku would do

Tip: in Make.com, always set Max Tokens on every Claude module. It’s a hard cap that protects you from a runaway loop draining your credit. I set mine to 1024 unless I specifically need more.

Next steps

Once the connection works, you have the foundation for pretty much any AI workflow in Make. A few useful next directions:

• Parse incoming emails and route them by intent (sales vs support)
• Summarise long PDFs or Google Docs before pushing them to a team
• Generate first-draft replies to customers, with a human approval step
• Enrich CRM records with AI-generated notes from meeting transcripts

Each of those has its own gotchas — prompt structure, chaining modules, keeping costs predictable, adding human-in-the-loop steps. That’s what the Implementation Blueprint course walks through in detail — full Make.com scenarios with Claude, Perplexity, and Gemini wired together, including the patterns I use in production.

If you just want to get your feet wet, the free QuickStart builds one working scenario end-to-end. No email required, name-your-price (including $0).

Either way — you’ve got Claude talking to Make now. That’s the hard part.

---

Questions or stuck on a step? Reply to hello@theaigeneration.co — I read everything personally.

Last updated: 20 April 2026. I keep this post updated as Make.com and the Claude API change.