The test: one automation, three platforms

Rather than listing features and pricing in a table and calling it a comparison (you can read the marketing pages yourself), I built the exact same automation in zapier, make, and n8n. The workflow: monitor a Gmail inbox for new emails, use AI to summarise the email, and post the summary to a Slack channel. Simple enough to be fair, complex enough to reveal real differences.

Here's what happened.

Zapier: The fastest to set up

I had the Zapier version running in about 8 minutes. That's not a typo. The interface walks you through each step with enough hand-holding that you'd struggle to get lost. Pick Gmail as the trigger, select "New Email," connect your account. Add an AI step (Zapier has a built-in "AI by Zapier" action), tell it to summarise the email body. Add Slack as the final action, pick your channel, map the summary as the message.

Done. Eight minutes. No documentation consulted.

The AI summarisation was decent. Not exceptional. Zapier's built-in AI produces summaries that are functional but generic. You can customise the prompt, which helps, but the output quality sits firmly in "good enough" territory.

The catch is pricing. Zapier's free tier gives you 100 tasks per month with single-step Zaps only. This automation has three steps (trigger + AI + Slack), so you immediately need the Starter plan at $19.99/month. That gets you 750 tasks. If you receive 30 emails a day, you'll blow through that in 25 days. The Professional plan at $49/month gives you 2,000 tasks, which is more realistic.

For a single automation, that's steep. At $49/month, each email summary costs you roughly $0.80. You could hire an actual human to read your emails for that price. Well, almost.

Make: The visual thinker's choice

Make (formerly Integromat) took about 20 minutes to set up. Not because it's hard, but because the visual workflow builder encourages you to think about your automation differently.

Instead of a linear "do this then this then this" flow, Make shows you a visual canvas where you connect modules with lines. The Gmail module connects to a Router (you can split the flow), which connects to an OpenAI module (for AI summarisation), which connects to a Slack module. You can see the entire flow at once, which is brilliant for complex automations but slight overkill for something this simple.

The AI step requires you to connect your own OpenAI API key, which is both a pro and a con. Pro: you get to use the actual GPT-4o model, so the summaries are noticeably better than Zapier's built-in AI. Con: you need an OpenAI account and API key, which adds setup friction. The summaries were more nuanced, picked up on the email's tone, and included better context.

Pricing is where Make gets interesting. The free tier gives you 1,000 operations per month. Our three-step automation counts as three operations per run, so that's about 333 emails per month on the free tier. That might actually be enough for many people. The Basic plan at $9/month gives you 10,000 operations. At three operations per email, that's over 3,300 emails. Far more generous than Zapier.

The total cost including OpenAI API usage: roughly $9/month for Make plus about $2-3/month in API costs for the summarisation. Call it $12/month total, versus Zapier's $49/month for similar volume. That's a significant difference.

n8n: The developer's playground

n8n is the wildcard here because it's open source and self-hostable. You can run it entirely on your own server for free, or use their cloud hosting starting at $20/month.

Setup took about 35 minutes, and I'll be honest, it required some comfort with technical interfaces. The workflow builder is visual (similar to Make), but the configuration for each node assumes you know what you're doing. Connecting Gmail required setting up OAuth credentials manually. The AI node lets you choose from multiple providers (OpenAI, Anthropic, local models) and gives you granular control over parameters like temperature and token limits.

The result was the best of the three. I used Claude's API for summarisation (because I could, and because the output quality is excellent for this task), and the summaries were noticeably superior. More concise, better structured, and they preserved important details like deadlines and action items that the other two sometimes missed.

The self-hosting advantage is enormous. If you run n8n on a $5/month VPS (DigitalOcean, Hetzner, whatever), your only ongoing cost is the AI API usage. No per-task limits, no operation caps, no artificial pricing tiers. Process 10,000 emails a month and your cost is the same as processing 10 (just the API calls).

The trade-off is clear: n8n requires technical ability. If you can deploy a Docker container and configure OAuth credentials, n8n is by far the best value. If that sentence made you nervous, n8n isn't for you.

The AI capabilities compared

This is where these tools are evolving fastest, and the differences matter.

Zapier has "AI by Zapier," a built-in AI action that uses OpenAI's models behind the scenes. You don't need your own API key. The prompts are templated and easy to configure. Quality is decent but you have limited control. They've also added AI-powered "suggested Zaps" that analyse your connected apps and recommend automations. That feature is genuinely useful for discovering workflows you hadn't considered.

Make doesn't have its own AI. Instead, it offers native modules for OpenAI, Anthropic, Google AI, and others. You bring your own API key, which means you get the full power of whatever model you choose, but you also manage the cost and configuration yourself. For power users, this is better. For non-technical users, it's a barrier.

n8n goes furthest. It has nodes for virtually every AI provider, plus the ability to run local models through Ollama integration. You can build AI chains where one model's output feeds into another. You can add vector databases for RAG (retrieval-augmented generation). If you want to build sophisticated AI workflows with branching logic, error handling, and custom processing, n8n is the only one of the three that won't get in your way.

Pricing summary (for our test automation at ~30 emails/day)

| Platform | Monthly Cost | AI Quality | Setup Time | |----------|-------------|------------|------------| | Zapier Professional | $49/month | Good | 8 min | | Make Basic + OpenAI API | ~$12/month | Better | 20 min | | n8n Self-hosted + Claude API | ~$8/month | Best | 35 min | | n8n Cloud + Claude API | ~$23/month | Best | 25 min |

Who should use what

Choose Zapier if: You're not technical, you want things working in minutes, and you don't mind paying a premium for simplicity. Zapier's app library is the largest (6,000+ integrations), so if you need to connect obscure tools, Zapier probably supports them. It's also the safest choice for teams where multiple people need to manage automations.

Choose Make if: You're a visual thinker, you want more control than Zapier offers, and you're comfortable connecting your own API keys. Make's pricing is dramatically better than Zapier's for medium to heavy usage, and the visual builder makes complex workflows easier to understand and debug. This is my recommendation for most small businesses.

Choose n8n if: You're a developer or have access to one. The self-hosting option makes it essentially free (minus API costs), the flexibility is unmatched, and the AI capabilities are the most sophisticated. The learning curve is steeper, but the ceiling is much higher. If you're building anything complex, you'll eventually outgrow Zapier and Make. You won't outgrow n8n.

The honest bottom line

If I'm building a quick automation and I want it done in ten minutes, I use Zapier. If I'm building something that needs to run efficiently at scale, I use Make. If I'm building something complex that involves AI processing, custom logic, or self-hosting requirements, I use n8n.

Most people reading this should probably start with Make. It's the best balance of usability, pricing, and capability. If you find yourself wanting more control, move to n8n. If you find yourself wanting less friction, move to Zapier. But Make is the sensible starting point.