How to Build Your First AI Workflow in Make.com Without Writing a Line of Code

What Is Make.com and Why Do AI Practitioners Need It?

Make.com (formerly Integromat) is a no-code visual automation platform that connects apps and services through a drag-and-drop canvas — no programming required. It is best understood as the layer that sits between your AI tools: where ChatGPT, Claude, or Gemini generate content or analysis, Make.com handles what happens next — routing outputs to the right destination, triggering follow-on actions, and eliminating every manual copy-paste step in between.

As of 2026, Make.com supports over 1,500 app integrations and includes dedicated native modules for OpenAI, Anthropic Claude, Google Gemini, and Perplexity. For practitioners who already use AI tools daily, it closes the last mile: your AI does the thinking, Make does the moving.

The gap it fills is real. You might already be using Claude to summarise meeting notes — but someone still has to copy those summaries into Notion, create the Asana tasks, and email the attendees. Make.com automates that entire downstream chain in one scenario you set up once.

 

How Does Make.com's Visual Canvas Work?

Unlike Zapier's left-to-right card interface, Make.com uses a circular canvas. You drag "modules" — each representing a single app action — onto the workspace, then draw arrows between them to define how data flows. The result looks like a flowchart you can read at a glance, explain to a colleague, and debug visually without tracing through a numbered list.

Two terms you need to know from day one:

--- Scenario: A complete workflow in Make.com. One scenario might watch for new Gmail emails, send each one to Claude for triage, then log the result in Notion.

--- Module: A single action in a scenario — one app, one task. "Watch Gmail," "Parse JSON," "Create Notion page" are each separate modules, connected by arrows that pass data between them.

Data that moves between modules is called a "bundle." Each bundle carries the output of one module (for example, an email's subject, body, and sender) into the next module as a live variable — referenced with double-curly-brace syntax like {{1.body}}.

The visual approach is Make's core advantage. According to Make's own usage data, scenarios with 5+ modules are built and debugged 40% faster on the canvas than equivalent automations in list-based tools, because the logic is visible rather than implied.

 

How Do I Connect an AI Model to Make.com?

Make.com has native modules for Claude, GPT-4o, Gemini, and Perplexity. Adding one to a scenario takes three steps: authenticate your API key, choose the model, write your prompt. The prompt can include dynamic variables pulled from any earlier module in your scenario — making each AI call context-aware by default.

Here is the exact process for adding Claude to a Make scenario:

--- Step 1: In your scenario canvas, click the "+" button and search for "Anthropic Claude"

--- Step 2: Select "Create a Message (Chat)" and click "Add connection"

--- Step 3: Paste your Anthropic API key from console.anthropic.com — name the connection and save

--- Step 4: Select your model (claude-sonnet-4-6 is the recommended default for most automation tasks)

--- Step 5: Write your system prompt, then in the "Messages" user content field, reference the output from a previous module using {{1.body}} or whichever variable holds your input data

Try this prompt now — copy-paste it into the Claude module system field for email triage:

--- "You are an email triage assistant. Analyse the email below and return a JSON object with exactly three keys: 'priority' (high / medium / low), 'action' (one of: reply / delegate / archive / follow-up), and 'summary' (one sentence, max 20 words). Return JSON only — no markdown, no explanation. Email: {{1.body}}"

The {{1.body}} is Make's variable syntax that pulls the live email body from Module 1 (your Gmail trigger) into the Claude prompt at runtime. Make replaces it automatically each time the scenario runs.

 

What Are the Most Useful AI Automation Patterns in Make?

After two years of AI integrations maturing on Make.com, five workflow patterns consistently deliver high value for practitioners — these are the automations that recover meaningful hours rather than just seconds per task.

Pattern 1 — AI Email Triage & Draft Responder

Trigger: new email arrives in Gmail. Claude reads and categorises it (sales inquiry, support request, press, spam), drafts a contextually appropriate reply in your voice, and saves it as a Gmail draft for review before sending. Time saved: 30–60 minutes of inbox triage per day for a busy professional mailbox.

Pattern 2 — Content Repurposing Pipeline

Trigger: new blog post appears in your RSS feed. Make.com sends the post title, URL, and excerpt to Claude, which generates three platform-specific social posts (LinkedIn, X/Twitter, Facebook). Make then queues them in Buffer on a staggered schedule. One article becomes a week of social content, automatically.

Pattern 3 — Meeting Notes to Action Items

Trigger: new audio file appears in a shared Google Drive folder. Make.com passes it to OpenAI Whisper for transcription, then sends the transcript to Claude with a prompt to extract action items, owners, and deadlines. Make creates Asana tasks and emails a summary to all attendees. No human edits required.

Pattern 4 — AI Document Classifier

Trigger: new file uploaded to a shared Google Drive folder. Claude reads the filename and first 300 words, identifies the document type (invoice, contract, proposal, report), and Move File module relocates it to the appropriate subfolder. Eliminates a constant but low-value admin task.

Pattern 5 — Research-to-Brief Pipeline

Trigger: a Typeform submission with a client brief. Make.com runs web searches via Perplexity API, sends all results plus the original brief to Claude to synthesise into a structured research document, and posts the output to a Notion database ready for the team. Turns a 2-hour research task into a 5-minute automated briefing.

 

How Do I Build My First Make Scenario Step by Step?

The fastest path to understanding Make.com is building something real. Below is a complete walkthrough of Pattern 2 — a content repurposing workflow — using an RSS trigger, Claude, and Buffer. Total setup time: approximately 45 minutes on your first attempt.

--- Step 1: Log into Make.com and click "Create a new scenario"

--- Step 2: Add an "RSS" trigger module. Set the RSS feed URL to your blog (e.g. https://yourblog.com/feed). Set polling interval to every 4 hours.

--- Step 3: Add an "Anthropic Claude" module. In the system prompt field, paste this prompt:

--- "You are a social media content specialist. Given the blog post details below, write three platform-specific posts. LinkedIn: professional tone, 120 words max. Twitter/X: punchy, under 280 characters, end with a question. Facebook: conversational, include one relevant emoji. Return as JSON with keys: linkedin, twitter, facebook. Do not include any other text. Title: {{1.title}} Summary: {{1.summary}} URL: {{1.url}}"

--- Step 4: Add a "JSON — Parse JSON" module, connecting it to Claude's output text field. Map the three keys.

--- Step 5: Add three Buffer modules (one per platform), each referencing the appropriate parsed JSON key.

--- Step 6: Click "Run once" to test with your most recent RSS post. Review the output, then click "Activate" to enable the scenario.

Once active, every new blog post automatically becomes three scheduled social posts — without you touching it.

 

Where Does Make.com Fall Short? (Honest Limitations)

Make.com is one of the best no-code automation platforms available in 2026 — but naming its real limitations saves you from building on wrong assumptions.

API costs accumulate inside workflows. Every scenario run that calls an AI model incurs API charges. A scenario triggered 50 times per day with 800-token Claude prompts generates 40,000 tokens daily. At claude-sonnet-4-6 rates, that is roughly USD $0.12/day per scenario. Low individually, but non-trivial across 10+ active automations. Monitor API usage weekly from the start.

Error handling needs deliberate setup. When a module fails — an API timeout, unexpected JSON format, rate limit — Make stops the scenario and logs an error by default. You need to add Make's "Error Handler" module from the beginning, not as an afterthought, to build production-grade resilience into your scenarios.

The free plan is not practical for AI workflows. Make's free tier provides 1,000 operations per month. A scenario that polls every hour uses 720 operations in trigger checks alone before doing any actual work. The Core plan (USD $9/month) is the realistic starting point for active AI automation use.

Real-time triggers require webhooks. Scheduled polling means your workflow might run with a 1–4 hour delay. For truly real-time triggers — like responding to a form submission the moment it arrives — you need to configure a webhook. It is not coding, but it does require 2–3 hours of learning for your first implementation.

 

Try This Now: Build a Gmail to Notion AI Summary in Under an Hour

Before building complex multi-branch scenarios, start with the simplest possible AI automation: Gmail → Claude → Notion. It produces immediate, visible value and teaches you every core Make concept you need for everything more complex.

What you need: Make.com free account, Gmail, Anthropic API key (or OpenAI), and a Notion database with a "Summary" text property and a "Sender" text property.

The Claude prompt to use:

--- "Read the email below and write a 2–3 sentence summary. Start the summary with the sender's main request or question. Do not include any greeting or sign-off. Return the summary only, nothing else. Email: {{1.body}}"

Connect Gmail → Claude → Notion (Create a page). Map Claude's output to the Summary field. Map the Gmail sender field to the Sender property. Run once, check your Notion database, activate.

Come back tomorrow. Every email you received will be there — summarised, searchable, and stored. Once you've seen it work, you will immediately see five other workflows you want to build. That is how every Make.com power user started.

懂AI的冷，更懂你的難 — UD 同行28年，讓科技成為有溫度的陪伴。 The goal of automation is not to replace your judgment — it is to protect your judgment for the decisions that actually need it.

 

Ready to Build AI Workflows That Actually Run?

Make.com is the starting line. UD's team will walk you through every step — from your first scenario to a full automation stack that handles your repetitive tasks while you focus on higher-value work.