n8n and Make build it. Cedric just does it.
Two canvases for building automations, one made for engineers and one for everyone else. Cedric is the third option, a teammate you ask instead of a diagram you draw. Facts read on n8n.io and make.com on 11 September 2026.
- n8n is the canvas for engineers. It takes code in any step, runs on your own servers if you want, and traces every step an agent took.
- Make is the canvas for business teams. It connects 3,000+ apps by drag and drop, and its assistant Maia drafts a scenario from a description.
- Both need the automation designed before its first run. That pays off for work that repeats the same way every time.
- Cedric starts from a sentence. You say what you need, and he works out the steps and comes back with the result.
- The meters differ. n8n counts full workflow runs, Make counts each module action, and Cedric counts credits on a plan with no seats.
What each one actually is
n8n calls itself an "AI Workflow Automation Platform", and its pitch is depth: "Build visually, go deep with code, connect to anything". It lists over 500 integrations and writes for technical teams in IT Ops, Security Ops and DevOps. You can run it on your own infrastructure with Docker, and the full source sits on GitHub under a fair-code licence. For a team whose data must stay in-house, that last point often decides it before Make is even on the table.
Make calls itself "The visual AI automation platform" and claims 3,000+ pre-built apps and 400+ AI integrations. You build by prompt, by drag and drop or over MCP, and Maia drafts automations and agents from a plain conversation. Founded in 2015 and now part of Celonis, Make says over 400,000 organizations use it. It aims at marketing, sales, operations, finance, IT and HR rather than at engineers.
Cedric is a teammate you hand the job to, with nothing to design first. You write to him in Slack, Microsoft Teams, Discord, Telegram or the browser, the way you would write to a colleague. He reads the mail, the calendar, the CRM and the channels, does the work, and reports back in the thread. Before a step sends, changes or spends anything, he asks you.
Side by side
Read on n8n.io and make.com on 11 September 2026. Both move quickly, so check their own pricing pages before quoting any figure.
| n8n | Make | Cedric | |
|---|---|---|---|
| What it is | A workflow canvas with code in it | A visual automation platform | A teammate you ask in chat |
| How the work gets made | You build the nodes and the branches | You drag modules, or Maia drafts the scenario | You describe the outcome, he plans the steps |
| Who it is aimed at | Technical teams: IT Ops, DevOps, Security Ops | Marketing, sales, ops, finance, IT and HR | Whoever owns the job, in plain words |
| Before the first run | Design the workflow | Design the scenario | Connect the apps he should read |
| When a case falls outside the plan | It takes the branch you built | It takes the route you built | He reports the case and asks what to do |
| Apps | 500+ integrations | 3,000+ apps, 400+ AI integrations | 3,200+ in the catalog, 15 with playbooks |
| Self-hosting | Yes, Docker, source on GitHub | Hosted | Hosted, nothing to install |
| What the meter counts | Full executions, not steps | Each module action in a scenario | Credits on a plan with no seats |
| Entry price | 20 EUR a month for 2,500 executions | Free to 1,000 credits, then $9 for 10,000 | 100K free credits, then $19 for 40,000 |
| Where you work with it | The n8n editor | The Make canvas | Your team chat, or the browser |
n8n lists Starter at 20 EUR, Pro at 50 EUR and Business at 667 EUR a month, billed annually. Those cover 2,500, 10,000 and 40,000 executions, with unlimited users on each. Make lists Free at 1,000 credits, then Core at $9 and Pro at $16 for 10,000 credits a month. Three meters count three different things, so no single number compares them.
The same job, three ways
Four unpaid invoices, one question
Grace Adeyemi9:14 AM
@Cedric which invoices are still unpaid?
CedricAPP9:15 AM
Four, CHF 9,310 in total. Keller Bau is 41 days late on CHF 3,150, and Lang SA 33 days on CHF 2,980. I drafted a polite reminder for each, with the invoice number and amount. Studio Arn and Weiss & Co are under 30 days, so I left them alone.
In n8n
A schedule trigger, a billing node, a code step for the 30-day cutoff, a mail node. An engineer builds it once and it runs every Tuesday.
In Make
A Tuesday scenario that pulls open invoices, filters past 30 days and sends the email. Maia can draft it, and each module action spends a credit.
In Cedric
One question in #finance. He read live billing, drafted two reminders in Grace's tone, and holds both until she says send.
How to tell which kind of work you have
Which product is better matters less than one plain question. Does the work you want to automate have a fixed shape?
If it does, build it. A form that always creates the same record and posts the same message will run ten thousand times without a surprise. A canvas is the right home for that: cheap, predictable, and a year later the diagram still tells you what it does. n8n and Make both do this well, and Make's app list is hard to beat.
The invoice chase sits in between. The list has a fixed shape, but the replies do not. One client writes "paid on Friday", another asks for a copy, a third wants to talk. On a canvas each of those becomes a branch someone draws and then maintains. Cedric reads every reply, reports it in the thread, and drops the paid invoices from next week's list.
Where each one fits
n8n fits when the automation has to be owned like software. Self-hosting with the source available, code inside the canvas, and a meter that ignores step count are real advantages, and none of them are marketing.
Make fits when breadth and ease matter most. It has 3,000+ apps, a canvas a marketer can drive, a free plan with no time limit, and Maia to draft the first version.
Cedric fits work that changes from one run to the next and belongs to the whole team rather than to one builder. Nobody has to learn a canvas or keep a scenario up to date.
Most teams have both kinds of work. A canvas for the plumbing and Cedric for the judgment is a common setup, and a sensible one.
FAQ
Yes. Tell him "every Tuesday at 9" once and he runs then, posting where you say. Each run reads the situation fresh instead of replaying fixed steps. That suits a report or a chase list better than a payment run.
It depends on the shape of the work, because the three meters count different things. A fixed flow that fires thousands of times is cheapest on n8n's per-run meter. Short scenarios with few modules fit Make's free and $9 tiers. Irregular jobs a person now does by hand fit Cedric's credits.
Maia is the closest thing in this category, and worth a try. It builds you a scenario, which you review and then run. Cedric does the job himself and brings back the result.
Cedric is hosted, so there is nothing to install, patch or upgrade. If your data has to stay on your own servers, n8n's self-hosted community edition is the pick of the three.
Yes, and many teams should run both. Let the canvas move records between systems on a schedule. Bring Cedric the requests where someone has to read the situation and say yes.
Several. Our list of n8n alternatives checks six on their own sites and sorts them by why you might be leaving.
Ask for the work instead of drawing it.
100,000 free credits, no card. Start with this week's unpaid invoices.
100K credits ($50) through Slack, Teams, Discord, Telegram or WhatsApp, 50K ($25) with Google. No card needed.