Lesson 10: Connecting OpenAI/ChatGPT API — Setup + Pricing Explained

Beginning

Everything in this course up to now has been about moving data from one place to another — an email into a spreadsheet, a spreadsheet into Slack, a webhook into Sheets. Useful, but mechanical. Nothing you've built so far could actually read something and understand it, or write something new on its own.

That changes in this lesson.

By connecting Make to OpenAI's API, your scenarios stop being simple messengers and start having something closer to judgment. An email can be summarized before anyone opens it. A customer message can get a drafted reply. A pile of raw notes can turn into a clean paragraph. This is the lesson where "automation" starts to feel like "AI automation" — the thing this entire course is actually named after.

The setup takes a bit more care than the apps you've connected before, mostly because billing works differently here. Walk through it carefully once, and every AI automation for the rest of this course builds on exactly this same foundation.

In this lesson, you'll learn:

• How to create an OpenAI developer account, separate from a regular ChatGPT subscription

• How API pricing actually works, and why it's usually far cheaper than people expect

• How to generate and secure your OpenAI API key

• How to connect Make to OpenAI and send your first real prompt

Core Concept: Renting Intelligence According to the Word

If you use ChatGPT’s website or app, you are already familiar with its standard pricing model: a subscription to the service, which lets you engage in unlimited chatting with the AI. With the ChatGPT API, the service is more akin to a phone company: you pay for the service by the token.

A token is not exactly a word – it is more like a “blob” of text, with one hundred blobs counting as about 75 words in English text. Everytime ChatGPT has to process a question or response, it has to “eat up” a certain number of tokens.

What many inexperienced API users do not expect is that this may well end up costing significantly less than one may think. The article estimates that even summarizing a short email using the API would likely cost only a few fractions of a cent. Even performing hundreds of such minor automatizations per month would likely only add up to several dollars – significantly less than “a fancy subscription to an AI chatbot”, as the article puts it.

Two things should be noted regarding the use of the API:

One’s subscription to ChatGPT does not entitle one to use the API, and vice versa: one does not need a ChatGPT subscription to use the API, but one needs to have an API key. The two services are billed seperately.

One pays for the service upfront, like a prepaid phone plan. One has to “reload” the API “wallet” in order to continue using it, like buying a new top-up voucher for a mobile phone. For the sake of budgeting, it may be useful to think of the API as a service that deducts money from a prepaid plan every time one uses it and sends a request and receives a response from the AI.

Tool Walkthrough: Connecting OpenAI to Make

Step 1 — Create an OpenAI Platform Account

Go to platform.openai.com — note the word "platform," this is the developer side, separate from the regular chat.openai.com interface. Sign up with an email, or continue with an existing Google account.

Step 2 — Add Billing Credit

In account settings, find the Billing section and add a small amount of credit — $5 is more than enough to experiment with for weeks. Without this step, API requests will fail with a quota error, even on a freshly created account.

Step 3 — Generate Your API Key

Navigate to the API Keys section and create a new secret key. Give it a name you'll recognize later, like "Make Automation." Copy the key immediately — OpenAI shows it to you exactly once and will not display the full key again afterward.

Step 4 — Add an OpenAI Module in Make

Back in Make, create a new scenario, click the "+", and search for "OpenAI." Select a basic action like "Create a Completion" or "Create a Chat Completion," depending on which version Make's app list shows you.

Step 5 — Connect Your Account

When prompted for a connection, paste in the API key you generated in Step 3. This links Make to your OpenAI account the same way you've linked Gmail, Slack, and Google Sheets in earlier lessons.

Step 6 — Send a Test Prompt

In the module's prompt field, type a simple instruction — something like "Summarize this in one sentence: [paste a short paragraph here]." Click Run once.

Step 7 — Read the Response

Open the module's output. You'll see the AI's generated text sitting inside the response data, ready to be mapped into any module after it — a Sheet, a Slack message, an email, anything you've already learned to build.

Real Example: The Inbox That Summarizes Itself

Picture a small business owner who receives dozens of customer emails a day, many of them long and rambling. Reading each one in full eats up real working time. By combining what you already know — Gmail's trigger from Lesson 4 — with an OpenAI module in between, every incoming email can be automatically condensed into a single clear sentence before anyone even opens their inbox.

This is a genuinely common freelance request, and it's a perfect example of the sort of job Phase 3 is building you toward: not flashy, not complicated once you understand the pieces, but the kind of time-saver a client will happily pay for every single month.

Common Mistakes

Mistake 1: Skipping the billing setup

An API key alone isn't enough — without credit added to the account, every request fails with a "quota exceeded" error, even moments after the account was created. This is the single most common first-time stumbling block.

Mistake 2: Treating the API key casually

Unlike a webhook URL, an exposed OpenAI key means someone else can spend real money on your account. Never paste it into a shared document, a public repository, or a screenshot that goes out unedited.

Mistake 3: Writing a vague prompt and blaming the result

The quality of what comes back depends heavily on how clearly the instruction is written. A prompt like "fix this" produces a weaker, less predictable result than "rewrite this paragraph to sound more formal, in under 40 words." Specific instructions get specific results.

Try This Yourself

Before moving to Lesson 11, complete the loop on your own:

• Create your OpenAI platform account and add a small amount of billing credit

• Generate an API key and connect it inside a new Make scenario

• Send one test prompt of your own choosing and confirm you get a generated response back

Key Takeaways

• OpenAI's API is billed separately from a ChatGPT subscription, using a prepaid, pay-as-you-go model

• Usage is measured in tokens, and typical automation tasks cost far less than most beginners expect

• Billing credit must be added before any API request will succeed

• An OpenAI API key should be protected at least as carefully as a password, since exposure can lead to real charges

• Clear, specific prompts produce noticeably better and more consistent results

Frequently Asked Questions

Do I need a ChatGPT Plus subscription to use the API?

No — they’re separate. You can use the API without an account on ChatGPT, and having a subscription doesn’t grant you any credit for the API.

What happens when I run out of my API credit?

Incoming requests will be denied until you add more credit — and there’s no hidden “overage” charge if you go over your plan limit. You’re always in control of how much you spend.

Which OpenAI model should I use for simple automation tasks?

For most simple, conversational use cases — summarizing, short-form replies, formatting — there’s a significantly cheaper and faster model available in OpenAI’s portfolio, leaving the largest and slowest model for more complex tasks.

Next Lesson Preview

With OpenAI connected, Lesson 11 puts it to real use: building a complete AI Auto-Responder for Emails — reading an incoming message, generating a reply with AI, and sending it back automatically.

Navigation:

← Previous: Lesson 9 — Understanding APIs & API Keys (Beginner-Friendly)

→ Next: Lesson 11 — Building an AI Auto-Responder for Emails

🏠 Back to Course Hub