What Most Guides Get Wrong About AI Marketing
Most lists of the best AI marketing tools are just affiliate roundups for glorified text editors. They’ll point you to a dozen different GPT wrappers that help you write social media posts. That’s a fine feature, but it’s not a business system. It’s a commodity. The real value isn’t in one-off content generation; it’s in building automated systems that run without you.
These guides consistently miss the point. They ignore the plumbing. An AI tool that can’t talk to your CRM, your email sender, or your lead database is a toy. A pretty UI is useless if the backend API is flaky or has draconian rate limits. You need tools built for connection, not just for content. This isn’t about finding a better thesaurus. It’s about building machines that do real work, and for that, you need to think like an operator, not a content creator.
The Core Stack: Glue, Brains, and Action
I build every single marketing automation with three components. Glue, Brains, and Action. Get these three right, and you can build almost anything. Get one wrong, and the whole system is brittle.
- The Glue: This is the automation platform that connects everything. For me, it’s Make (formerly Integromat). I used to use Zapier, but I moved everything off it. My main gripe with Zapier is that its branching logic—the “Paths” feature—is absurdly clunky and expensive. Make.com’s visual, flowchart-style interface makes it far easier to build and debug complex, multi-step workflows. Its error handling is also more granular, which is critical when an automation is running on its own.
- The Brains: This is the Large Language Model (LLM) that does the thinking. The default choice is OpenAI’s API, specifically GPT-4o, for its speed and reasoning capabilities. For tasks that require more sophisticated and nuanced writing, I’ll sometimes use Anthropic’s Claude 3 Opus API. People worry about cost here, but they shouldn’t. For most of the automations I run, the API costs are pennies per execution. We’re talking a few dollars a month, not hundreds.
- The Action: This is the final tool that performs the task. It could be an email service provider like Brevo for sending the email, a social media tool like Buffer via its API for posting an update, or your CRM like HubSpot for creating a new contact. The specific tool depends on the job.
Example Build: A Personalized Cold Email Research Agent
Let’s make this concrete. Here’s a real system I built that saves me about 10 hours of manual work every month. The goal: automatically research new leads and write a personalized first line for a cold email.
The entire process is orchestrated with Make.com.
1. The Trigger: New Lead in Google Sheets
The workflow starts when I add a new row to a Google Sheet. The sheet has three columns: `FirstName`, `LastName`, and `CompanyWebsite`. That’s it. Simple.
2. The Scraper: Grab Website Content
Next, Make.com sends the `CompanyWebsite` URL to Apify. Apify is a web scraping and automation platform. My favorite thing about it is the library of pre-built “Actors” that can handle common tasks, like scraping the text off a single page. I use their “Website Content Scraper” actor. It visits the URL and returns the clean text content. Apify’s pricing is fair; the $49/mo plan is plenty for most small-scale operations, though their free tier is generous enough for testing. Honestly, trying to build and maintain your own scraper is a fool’s errand in 2026.
3. The Brains: Generate the Opening Line
The scraped text from Apify is then passed to the OpenAI API via a Make.com module. I use the GPT-4o model with a very specific prompt. The prompt is everything. A lazy prompt gives you generic, unusable garbage.
You are a senior B2B sales development representative specializing in technology sales. Your tone is concise and professional. Given the following text from the 'About Us' or 'Home' page of a company, perform two tasks:
1. Identify the company's primary value proposition or a recent major announcement.
2. Write a single, compelling opening sentence for a cold email that references this specific value proposition or announcement. The sentence should feel personal and demonstrate you've done your research.
Output ONLY the sentence. Do not include any preamble, explanation, or quotation marks.
Here is the text:
{scraped_text_from_apify}
4. The Action: Send the Email
The sentence generated by OpenAI is the output of that step. The final module in my Make.com scenario takes that sentence and plugs it into an email template in Brevo (formerly Sendinblue). The email is then sent to the lead. Brevo’s API is reliable and their deliverability is solid. I find their free plan is a joke for any real volume, but the entry-level paid plan at around $25/month is a good value for what you get.
This four-step process runs in about 60 seconds and costs maybe $0.03 per lead. It turns a tedious manual task into a fully automated system.