n8n Review (July 2026): Features, Pricing & Verdict

n8n review: the short version

n8n is a fair-code workflow automation platform built in Berlin by n8n GmbH, designed for technical teams that want code-friendly nodes, self-hosting, and AI agent tooling rather than a no-code consumer experience. After reviewing the official documentation, the 2.x release notes, the public pricing page, the Trust Center, and customer case studies, my conclusion is simple: n8n is one of the strongest automation platforms for engineers in July 2026, and one of the worst fits for non-technical marketers who just want a button.

The product has clearly crossed a threshold this year. The general availability of the Microsoft Agent 365 integration in May 2026, the SAP Joule Studio partnership, and steady weekly releases documented in the changelog make it a credible backbone for serious AI infrastructure — not just a Zapier clone with a code node bolted on.

“n8n is a beast for automation. self-hosting and low-code make it a dev’s dream. if you’re not automating yet, you’re working too hard.” — Anderoav, builder (n8n pricing page testimonial, 2026) (n8n.io/pricing)

That quote is a fair summary of the platform, but it leaves out the cost: self-hosting shifts ops work to you, and the cloud pricing is in euros, not dollars.

What is n8n?

n8n is a node-based, fair-code workflow automation platform that runs as either a managed cloud service or a self-hosted Docker deployment. It lets you chain APIs, databases, and AI models together on a visual canvas, with optional JavaScript or Python code at any step. The name is a contraction of “nodemation,” coined by founder Jan Oberhauser in 2019 (GitHub README).

The platform was first released on Product Hunt on July 24, 2019, and has since grown into one of the most-starred open-source repositories in the automation category (Product Hunt).

Who should buy n8n in 2026 — and who should not

n8n is a buy for technical teams that want control. Developers, platform engineers, security operations teams, and AI builders who need HTTP flexibility, code steps, and self-host economics. It is a skip for non-technical marketers or solo operators who just want a two-step Zap. If your primary apps are mainstream SaaS with pre-built connectors and your team lacks engineering bandwidth, Zapier or Make will ship faster.

Buy n8n if you are:

• A developer or platform engineer building internal automations across APIs, databases, and AI models.
• A security or IT operations team that wants SOAR-style workflows with self-hosted data control.
• A startup or mid-market company running AI agents that need human-in-the-loop approvals.
• A team that needs to integrate with systems that have no off-the-shelf connector.
• An organization with a data residency or GDPR constraint that prefers EU cloud or self-hosting.

Skip n8n if you are:

• A marketer or operations user who wants a simple “if this, then that” tool with zero code.
• A solo founder who needs an automation running in the next 30 minutes with no setup.
• A team that lacks anyone willing to debug JSON paths and read execution logs.
• A buyer who needs live chat support on day one — the Business plan is self-serve forum support.

How we scored n8n

aiuncovers reviews are scored on a 100-point model across practical value, product quality, ease of use, pricing value, differentiation, transparency, and support documentation. The n8n evaluation is documentation-and-source-based, conducted in late July 2026 against the official 2.x release line, with current pricing verified on n8n.io/pricing and release notes cross-referenced against docs.n8n.io/changelog. The scores below reflect an editorial reading of the official materials — they are not the result of hands-on testing.

n8n score breakdown

Dimension	Weight	Score	Notes
Practical value	25	22	Solves real engineering problems; high ceiling.
Product quality	20	17	Stable releases; rare rough edges in AI nodes.
Ease of use	15	9	Steeper curve than Zapier for non-developers.
Pricing value	15	13	Execution pricing is friendly; EU currency adds friction.
Differentiation	10	10	Self-host, code nodes, MCP, and AI in one tool.
Transparency	10	8	Fair-code license and visible source; some pricing detail is locked behind sales.
Support documentation	5	4	Excellent docs; forum-only support on cheaper plans.
Total	100	83	Strong technical pick, not a generalist tool.

How n8n pricing works in 2026

n8n uses execution-based cloud pricing — you pay per completed workflow run, not per individual step or task inside the run. The published tiers in July 2026 are Starter at €20/month (2,500 executions), Pro at €50/month (10,000 executions), and Business at €667/month (40,000 executions), with an Enterprise tier sold by contact (n8n.io/pricing). Annual billing saves 17%. The free Community Edition on GitHub is self-hosted and free forever under the Sustainable Use License.

This pricing model is the single most important difference between n8n and its closest competitors. In a Zapier or Make workflow, every node that fires inside a run counts as a billable task. In an n8n workflow, the entire run from trigger to last node counts as one execution, regardless of how many steps it contains.

“Running workflows with around 100,000 tasks could easily cost over $500 per month on other platforms. On n8n’s Pro plan, you can do the same for about $50.” — Luis Guzmán and Lucas Nilsson, n8n blog (blog.n8n.io/n8n-execution-advantage)

n8n Cloud plans at a glance (July 2026)

Plan	Monthly price (annual)	Executions	Concurrent	Projects	AI Assistant credits	Support
Starter	€20/mo	2,500	5	1 shared	2,300/mo	Forum
Pro	€50/mo	10,000	20	3 shared	up to 13,700/mo	Forum
Business	€667/mo	40,000	(higher)	6 shared	Coming soon	Forum
Enterprise	Contact sales	Custom	200+	Unlimited	Coming soon	Dedicated, SLA

Source: n8n.io/pricing, captured 2026-07-29. Prices are listed in EUR; convert and confirm at checkout.

AI Assistant credits and quotas

The AI Assistant is in early preview on all cloud plans and comes with a monthly allowance of credits. Free trials include 800 credits, which expire when the trial ends; Starter plans get roughly 2,300 credits per month and Pro plans up to 13,700, depending on plan size (n8n.io/pricing). On self-hosted n8n, AI Assistant is coming soon and you bring your own API key, so no credit limits apply.

Business plan overage rules

If you exceed your Business quota, your workflows keep running, but overage charges can apply. n8n publishes an overage price of €4,000 per extra 300,000 executions on the Business plan (n8n.io/pricing). Annual subscribers get a courtesy heads-up at 80% of their annual quota.

What you get for free: the Community Edition

The Community Edition is the free, self-hosted version of n8n, distributed on GitHub under the Sustainable Use License. It includes almost the complete feature set: unlimited workflows, unlimited steps, the visual editor, code nodes, webhooks, AI agent nodes, and 1,500+ integrations (docs.n8n.io/choose-how-to-use-n8n).

You pay only for the infrastructure you run it on, since n8n itself is free. The minimum footprint is small enough that a basic VPS or even a developer laptop can host it for evaluation; sizing for production depends on workflow volume, queue mode requirements, and database choice. There is no time limit and no per-user fee. The trade-off, clearly stated in n8n’s documentation, is that you own the uptime, backups, upgrades, and security hardening (docs.n8n.io/choose-how-to-use-n8n).

When Community Edition is enough

• Personal projects and learning.
• A single team running fewer than a few hundred workflows per day.
• Prototyping an AI agent before committing to cloud.
• Environments where you cannot send data to a third-party cloud for compliance reasons.

When you should pay for Business or Enterprise

• You need SAML SSO, LDAP, or role-based access control.
• You want environments (dev, staging, prod) and Git-based version control.
• You need external secrets management (HashiCorp Vault, AWS Secrets Manager, Azure Key Vault, 1Password).
• You need audit logging, log streaming, or data residency guarantees.
• You have more than one or two engineers and need project-level access boundaries.

Features documented in the 2.x release line

The notes below come from the official n8n documentation, the 2.x changelog, the vendor’s own feature pages, and the Microsoft Agent 365 launch post — not from hands-on use. Where I make a judgement, I flag it as editorial opinion.

Visual editor and code nodes

The canvas is the heart of the product. You drag nodes onto a graph, wire them together, and configure each one in a side panel. Clicking any node shows its raw JSON input and output, which is the closest thing to a debugger you get on a no-code platform. Code nodes support JavaScript and Python (via Pyodide) and are first-class citizens, not hidden behind a paywall. This alone is a major reason developers pick n8n over Zapier, where Code steps have hard 30-second execution limits and a 256MB memory cap (n8n.io/vs/zapier).

AI agents and human-in-the-loop

The AI Agent node, shipped in the 2.x line, lets you combine a chat model, memory, tools, and a system prompt on the canvas. In v2.6 (January 2026), n8n added human-in-the-loop approvals for tool calls — a gated tool cannot execute unless a human explicitly approves the action (docs.n8n.io/changelog). This is a real differentiator for production AI, because it removes the brittleness of prompt-based safeguards.

MCP support

Model Context Protocol is supported on both sides. n8n ships an MCP server so external clients can drive n8n workflows, and an MCP client node so your agents can call any MCP-compatible tool. Recent updates let you connect to servers like Apify, Linear, monday.com, Notion, and PostHog with one click (docs.n8n.io/changelog).

Verified webhooks

As of v2.21 (May 2026), fourteen trigger nodes now verify incoming webhook signatures using each service’s native HMAC scheme, rejecting forged requests with a 401 before any execution starts (docs.n8n.io/changelog). For teams running production automations against GitHub, Stripe, or Twilio, this is a quiet but important security upgrade.

Microsoft 365 and Agent 365

In May 2026, n8n shipped the Microsoft Agent 365 Trigger node, letting teams build n8n-powered agents that show up as members of your team inside Microsoft Teams, Outlook, and Word, with their own Entra ID identity (blog.n8n.io/deploy-n8n-agents). This is the deepest integration n8n has done with a hyperscaler.

SAP Joule Studio

In 2026, n8n also became a managed environment inside SAP Joule Studio on the SAP Business Technology Platform, meaning SAP customers can run n8n workflows natively inside BTP without spinning up new infrastructure (n8n.io/sap-partnership). For SAP shops, this collapses the procurement and security review that usually blocks new automation tools.

Other features worth noting

• OpenTelemetry tracing for workflow executions (v2.15, March 2026), with span-level detail.
• External secret stores including 1Password, HashiCorp Vault, AWS Secrets Manager, Azure Key Vault, and GCP Secret Manager.
• Custom project roles and finer-grained RBAC (v2.8, February 2026).
• Canvas Groups to collapse and label sections of large workflows (v2.28, June 2026).
• Visual diff in version history (v2.13, March 2026).

Self-hosting nuance: what the docs don’t shout about

Self-hosting n8n is genuinely viable, but it is not a turnkey experience. You will own the database, the backups, the upgrades, the TLS termination, the secret storage, and the queue mode if you want scale. The good news is that the docs and the Community Edition are mature enough that a competent engineer can stand up a production-grade instance in an afternoon.

The fair-code Sustainable Use License lets you self-host and modify the code, but it is not OSI-open-source in the strict sense: there are limits on how you can offer n8n as a hosted service to third parties (GitHub LICENSE.md). For internal company use, the license is friendly. For resale, you need a separate commercial agreement.

Deployment options documented by n8n

1. Docker Compose on a single VPS — the path the official installation docs recommend for getting started quickly.
2. Kubernetes with the n8n Helm chart — documented for production scale, with queue mode and an external PostgreSQL.
3. n8n Cloud — the easiest documented path once you outgrow self-hosting.

What self-hosting does NOT solve

• You still need your own secrets vault and identity provider.
• You need to run your own upgrade cadence, since n8n ships a new minor version most weeks (docs.n8n.io/changelog).
• You do not get the SOC 2 scope that n8n Cloud carries — your auditor will ask about your own setup.

n8n vs Zapier vs Make: where n8n wins and loses

The comparison below is drawn from n8n’s own vs/Zapier and vs/Make pages, plus each vendor’s public pricing page as of late July 2026. Where n8n’s framing favours itself, I have noted it.

Comparison table

Dimension	n8n	Zapier	Make
Pricing model	Per workflow execution	Per task (step)	Per operation; transitioning to credits (2025)
Starting price	€20/mo (2.5K executions)	$19.99/mo (750 tasks)	Free tier; Core plan from ~$9/mo
Self-hosting	Yes, free Community Edition	No	No
Code steps	JS and Python, no limits	JS and Python, restricted sandbox	JS via custom function on Enterprise
Webhooks	Free on every plan	Premium app, Pro plan+	Limited
Custom nodes	Yes, build and deploy	Developer Platform only	Developer Hub with review
Multi-agent AI	Native, with HITL	Basic AI actions	AI modules
MCP support	First-class	None	None
EU data residency	Yes (Azure Frankfurt)	No (US-based)	EU region available
Support	Forum on Starter/Pro/Business; SLA on Enterprise	Tiered, live chat on Team+	Tiered

Sources: n8n.io/vs/zapier, n8n.io/vs/make, and each vendor’s own pricing page (July 2026).

Where n8n clearly wins

• Cost predictability for complex flows. A 50-step AI workflow counts as one execution on n8n but as 50 tasks on Zapier.
• Self-hosting. Zapier and Make are both cloud-only.
• Code and HTTP flexibility. Code nodes on n8n can import npm packages and run shell commands.
• AI tooling. Native LangChain nodes, MCP, and human-in-the-loop are years ahead.

Where Zapier and Make win

• Time to value. Zapier is faster for a marketer to ship a five-step Zap in 20 minutes.
• Ecosystem size. Zapier advertises 8,000+ apps, Make has 2,800+; n8n’s 1,500+ is growing but still smaller (n8n.io/vs/zapier, n8n.io/vs/make).
• Non-developer onboarding. Zapier’s UI is more forgiving; n8n assumes you can read a JSON path.

Pros and cons: the honest list

Pros

• Execution-based pricing that rewards complex workflows, not penalizes them.
• Free Community Edition on GitHub that is feature-complete for individual use.
• Code-friendly nodes in JavaScript and Python at any step.
• Native AI agent tooling with human-in-the-loop approvals and MCP support.
• Self-host option for data residency and GDPR.
• Active development — the changelog ships something meaningful every week.
• Strong partnerships — Microsoft Agent 365, SAP Joule Studio, and major hyperscalers.
• Visible source code under a fair-code license (github.com/n8n-io/n8n).

Cons

• Steeper learning curve than Zapier or Make for non-developers.
• Pricing listed in euros, which complicates budgets for US dollar teams.
• Self-hosting shifts ops burden: upgrades, backups, security hardening are on you.
• AI Assistant is still in preview with capped credits and no option to buy more on cloud.
• Smaller integration library than Zapier, though the gap is closing.
• Forum-only support on every plan except Enterprise — there is no live chat.
• Some advanced governance features (SAML, audit logs, multi-main) require Enterprise.

Security, compliance, and trust

n8n Cloud is hosted on Microsoft Azure in the EU (Frankfurt, Germany region) and carries a SOC 2 Type 2 audit, a SOC 3 report, and GDPR compliance (n8n.io/legal/security, trust.n8n.io). Customer instances are logically isolated, data is encrypted at rest with AES256, and OAuth tokens are stored in an encrypted credentials store. n8n runs third-party vulnerability scans at least every 90 days and third-party penetration tests at least annually.

For self-hosted deployments, n8n is transparent that you own the encryption-at-rest setup, the TLS termination, and the secrets management. The security docs are clear about this division of responsibility, and the docs.n8n.io/choose-how-to-use-n8n page spells out exactly what self-hosters must do.

The Trust Center at trust.n8n.io lets enterprise buyers request the SOC 2 report, the pentest report, and self-assessment questionnaires (CAIQ) for both cloud and self-hosted deployments. This is the level of paperwork a Fortune 500 procurement team will ask for, and n8n is one of the few automation vendors that delivers it without weeks of back-and-forth.

Alternatives to consider

• Zapier — best for non-technical teams that need the largest app library and fastest onboarding. Choose Zapier over n8n if your users are marketers, not engineers.
• Make (formerly Integromat) — best visual builder in the middle ground. Choose Make if you want more design polish than Zapier without n8n’s code depth.
• Pipedream — developer-focused, code-first workflows with strong HTTP and API primitives. Choose Pipedream if you are mostly an API engineer.
• Workato — enterprise integration platform as a service (iPaaS) with strong governance and pricing for mid-market and above.
• LangChain or CrewAI — pure agent frameworks without the broader workflow canvas. Choose these if your work is 90% LLM orchestration and you do not need 1,000+ integrations.

The verdict

n8n is one of the best automation platforms for technical teams in July 2026. Execution-based pricing, a free Community Edition, code-friendly nodes, native AI agent tooling with human-in-the-loop, MCP, and serious enterprise partnerships give it a depth that none of its direct competitors match. If your team is comfortable with code, you will get more for less money than you would on Zapier or Make.

If your team is not comfortable with code, you will hate it. The same things that make n8n powerful — the canvas, the expressions, the JSON debug panel — also make it intimidating for the marketers and operations folks that drive most automation tool purchases. For them, the friction outweighs the savings.

aiuncovers score: 83/100. Strong technical pick. Skip for non-developers.

Frequently asked questions about n8n

What is n8n used for?

n8n is used to automate multi-step workflows that connect APIs, databases, AI models, and internal services. Common jobs include webhook-driven data pipelines, AI agents with tool use, scheduled syncs between SaaS apps, IT and security operations (SOAR), and RAG-style retrieval workflows (n8n.io/features, docs.n8n.io/changelog).

Is n8n really free to self-host?

Yes. The Community Edition on github.com/n8n-io/n8n is free forever under the Sustainable Use License and includes almost the full feature set. You pay only for the server you run it on. Some advanced features — SAML SSO, environments, audit logs — require a paid Business or Enterprise license key applied to a self-hosted instance.

How does n8n pricing work in 2026?

n8n Cloud charges per workflow execution, not per step. Published tiers are €20/mo (Starter), €50/mo (Pro), and €667/mo (Business) for 2,500, 10,000, and 40,000 executions respectively. Annual billing saves 17%. Overages on the Business plan are billed at €4,000 per extra 300,000 executions (n8n.io/pricing).

Is n8n better than Zapier?

For technical teams running complex or AI-heavy workflows, yes. n8n’s execution-based pricing, code nodes, MCP support, and self-hosting option are real advantages. For non-technical teams running simple app-to-app Zaps, Zapier is faster to learn and has a larger integration library (8,000+ vs n8n’s 1,500+). The two tools serve different buyers (n8n.io/vs/zapier).

Is n8n better than Make?

Depends on the team. Make has a more polished visual builder and credits-based pricing as of August 2025. n8n has code nodes, self-hosting, and stronger AI tooling. If your team writes code, choose n8n. If your team is allergic to code and you want a middle ground between Zapier and n8n, choose Make (n8n.io/vs/make).

Can you build AI agents in n8n?

Yes. n8n ships native AI Agent nodes with LangChain-style orchestration, tool calling, memory, and human-in-the-loop approvals. You can connect to OpenAI, Anthropic, Google, Mistral, NVIDIA, Moonshot Kimi, Alibaba Qwen, and local models via Ollama. New model providers ship in most weekly releases (docs.n8n.io/changelog).

Does n8n support MCP?

Yes, both as an MCP server (so external clients can drive workflows) and as an MCP client (so your agents can call any MCP-compatible tool). One-click connectors exist for Apify, Linear, monday.com, Notion, and PostHog (docs.n8n.io/changelog).

Is n8n secure enough for enterprise use?

For cloud deployments, yes: n8n Cloud is SOC 2 Type 2 audited, SOC 3 reportable, GDPR compliant, hosted in Azure Germany West Central (Frankfurt), and runs annual third-party penetration tests (trust.n8n.io). For self-hosted deployments, security depends on your own setup — n8n provides the building blocks but you own the encryption-at-rest, TLS, and secrets management.

Does n8n work with Microsoft 365?

Yes. n8n shipped the Microsoft Agent 365 Trigger node in May 2026, letting you build agents that show up inside Teams, Outlook, and Word with their own Entra ID identity (blog.n8n.io/deploy-n8n-agents). It also runs natively inside SAP Joule Studio on the SAP Business Technology Platform (n8n.io/sap-partnership).

What are the main downsides of n8n?

The biggest downsides are the steeper learning curve for non-developers, the EUR pricing that adds currency conversion friction, the ops burden of self-hosting, the still-preview AI Assistant on cloud with capped credits, and the fact that forum-only support on the cheaper plans is not enough for some enterprise buyers.