What does Make's AI do for a small business?
Make does more than pass data between apps. The AI inside it summarizes, sorts, drafts and even decides. Here is what is there, and where I would stop.
40 new emails. Someone has to read each one, work out what it needs and pass it on. It is not hard work. It is boring work, and it eats hours.
Make is a tool that connects your apps. The AI inside it turns it from something that moves information into something that also understands it.
What is Make, in brief?
You build a scenario on a visual canvas: blocks (modules) joined by a line. An email arrives, a row is written to a sheet, a Slack message goes out. According to Make's site it has more than 3,500 ready-made apps, and if an app is missing you can reach it through an HTTP module. No code.
And the AI? It is one more block in the chain. You put it in the middle, and the information passing through is no longer just copied, it is understood.
What AI capabilities does Make have?
The simple ones are the AI Toolkit: ready-made blocks that summarize text, sort it into categories, detect language, analyze tone (positive or negative), extract details and translate. Exactly what a customer email needs: a one-line summary and a label: order, complaint or question.
Picture an email that arrives at 23:40. By morning it is already summarized, sorted and waiting for you in Slack.
For documents and images there is the AI Content Extractor. It reads PDF and Word files, images and recordings, and returns clean text. According to Make you can pull details from an invoice or a receipt, or transcribe a recording. It is available on paid plans and still in beta.
And for those who want one more step, there are AI agents. An agent gets instructions (a prompt) and picks which tool to use on its own: send an email, search documents, update a record. You can feed it your own knowledge, such as FAQs or business procedures. Make offers them in open beta, available on all plans with Make's AI Provider.
When an agent, and when a regular scenario?
Make itself gives a simple rule of thumb. A process with fixed rules, like updating stock after an order, needs no AI at all. A fixed-rule process with a text step, like summarizing or translating, fits an AI block. Keep agents for tasks that need judgment, like sorting incoming requests.
How do you pick a task for an agent? Make answers that too: something you would trust to an intern.
Can you build this by talking?
Almost. Maia, Make's AI co-worker, takes a description in your own words and builds a scenario on the screen. She asks clarifying questions along the way, and you see every block and can change it. You can also ask her to fix a scenario that got stuck.
Maia is in public beta. On the Free plan she comes with a 30-day trial, and on paid plans she is available, with usage consuming credits.
What if I already work with Claude or ChatGPT?
Make's MCP server lets AI assistants such as Claude and ChatGPT run scenarios in your account. You ask in the chat, and the scenario runs. The assistant you already know gets hands.
How much does it cost?
According to Make's pricing page there is a Free plan with no time limit and 1,000 credits per month. Each action in a scenario uses one credit. AI actions with Make's AI Provider are measured by the amount of text that passed through them. On paid plans you can connect your own provider, such as OpenAI or Anthropic: then it is one credit per action, plus payment to the provider itself.
Prices of the paid plans change, so check them on the site.
Where should you stop?
AI makes mistakes. Whoever connects it to a process is responsible for the result. So at first a person approves every answer before it reaches a customer. Do not give an agent sensitive data, financial decisions or matters with legal weight, and Make itself advises avoiding that. And name an owner for every scenario.
Start small: one task, sorting emails for example. When it works, expand. And if it does not, you failed small, which is exactly the idea.