Building AI Automations with Make.com: A Product Manager's Practical Guide

Why Make.com Over Zapier for AI Workflows

Zapier is simpler but linear — each zap is a straight line from trigger to action. Make.com uses a visual scenario canvas where you can build branching logic, loops, error handling, and parallel paths. For AI workflows, this matters: an AI module that sometimes returns JSON and sometimes returns plain text needs conditional logic to handle both cases. Make.com handles this natively; Zapier struggles.

Make.com also has native HTTP request modules, better JSON handling, and lower cost per operation — important when you're running automations that call expensive AI APIs thousands of times per month.

5 High-Value PM Automations to Build First

1. Competitive intelligence daily briefing: Trigger: daily at 8 AM. Steps: HTTP request to Perplexity API with competitor research prompts → format output → send to Slack channel. Estimated build time: 2 hours. Time saved: 30 min/day of manual research.

2. User feedback categorisation: Trigger: new feedback form submission (Typeform/Google Forms). Steps: send feedback text to Claude API with categorisation prompt → update Notion/Airtable row with category, sentiment, and priority. Build time: 3 hours. Time saved: 2-3 hrs/week of manual tagging.

3. Meeting notes → JIRA tickets: Trigger: new note in Notion tagged "meeting." Steps: send content to Claude with "extract action items and convert to JIRA ticket format" prompt → create JIRA issues via API. Build time: 4 hours. Time saved: 1 hr/day of post-meeting admin.

4. App store review monitoring: Trigger: daily. Steps: fetch new Play Store / App Store reviews via API → send to Claude to identify bug reports and feature requests → filter high-priority items → send to Product Slack channel. Build time: 3 hours. Impact: no more missing critical user feedback.

5. Weekly product metrics report: Trigger: every Monday 9 AM. Steps: pull last week's metrics from Mixpanel API → send data to Claude with "write a 200-word PM briefing on these metrics" prompt → post to Slack. Build time: 4 hours. Time saved: 45 min/week of report writing.

The Make.com AI Module Setup

Make.com has native modules for OpenAI (GPT-4, GPT-4o) and Anthropic (Claude). To use them:

• Add the "OpenAI" or "Anthropic" module to your scenario
• Connect your API key (stored encrypted in Make's credentials vault)
• Configure the prompt — you can use dynamic variables from earlier steps in your scenario
• Parse the response (usually JSON or plain text) and pass it to the next step

For Claude specifically: use the HTTP module with a direct Anthropic API call if Make's native Anthropic module doesn't have the latest model. The HTTP module gives you full control over request headers and body.

Pricing: What You Actually Need

For most PMs with 5-10 active scenarios: Pro at $16/month is sufficient. Each Make operation counts as one step in a scenario — an automation with 8 steps running 100 times/month uses 800 operations. Monitor your operation usage in the first month to calibrate which plan you need.

FAQ

Make.com vs n8n — which should I choose?

n8n is open-source and self-hostable — better if you have data sovereignty requirements or want unlimited automations at a fixed infrastructure cost. Make.com is fully managed, has a better visual interface, and is easier to use without technical background. For most PMs: start with Make.com. Move to n8n if you hit cost or data residency constraints.

Can non-technical PMs really build these automations?

Yes. Make.com's visual interface is genuinely learnable in a weekend. The AI modules add complexity, but the basics (trigger → AI call → send result) are within reach of any PM who can write clear prompts. Start with one of the 5 automations above. The pattern repeats across all of them.