n8n occupies an interesting position in the 2026 automation market. It is the open-source, self-hostable, developer-friendly answer to Zapier and Make. Founded in 2019 in Berlin by Jan Oberhauser, the project has grown to over 60,000 GitHub stars and a commercial product on top with funding from Sequoia (n8n GitHub, retrieved 2026). The Fair-code license, the self-host story, and the developer ergonomics give n8n a loyal community that is unlikely to be swayed by a hosted-only alternative.

I respect n8n. I also think the AI Agent node added in 2024 does not change the underlying shape: n8n is a workflow tool with AI capability, not an agent platform. The distinction matters for the buyer trying to decide between extending an n8n setup and starting fresh on an agent-native product.

What n8n is in 2026

n8n is a workflow automation tool with three deployment surfaces. Self-hosted Community Edition under the Fair-code license. n8n Cloud, the company's hosted product. n8n Enterprise for organisations needing SSO, advanced permissions, and dedicated support (n8n pricing, retrieved 2026).

The workflow surface

Open n8n and you see a canvas. Nodes for HTTP, webhooks, databases, SaaS apps, code blocks. Edges connecting them. The familiar workflow metaphor, executed well. The differentiator is developer ergonomics: the Code node lets you write JavaScript, the extension model lets you ship custom integrations, the self-host story keeps data on your infrastructure.

The Fair-code license

n8n's license is not strictly open source by OSI definition; it restricts commercial hosting of n8n itself. For self-hosting in production, the license is permissive. For SaaS companies wanting to white-label n8n, restrictions apply. This is a feature: it lets the company sustain itself while keeping self-host free for the community.

The n8n AI Agent node: what it actually does

Added through 2024, the n8n AI Agent node embeds an LLM-driven step inside a workflow. The node uses LangChain primitives under the hood and can reason, call tools, and produce structured outputs. The result is a workflow that has one or more "smart" nodes alongside the deterministic ones.

Why this is useful

A workflow can now classify inbound emails, summarise documents, generate responses, and make judgement calls inside what used to be a pure deterministic chain. For many use cases this is a real upgrade over Zapier-style workflows.

Why this is not the same as agent-native

The AI Agent is a node inside the workflow graph. The graph is still the unit of work. The buyer still has to draw the workflow. The agent step does fuzzy work inside a deterministic outer shell, rather than the agent being the loop. For the deeper view see AI agent vs workflow automation and describe outcome, not workflow.

What Gravity does differently

Gravity inverts the relationship: the agent is the unit of work, the integrations are tools the agent consults, the workflow shape is not drawn in advance. There is no canvas to maintain.

The trade-off is real and worth naming. By giving up the canvas, Gravity gives up some auditability and the ability for ops folks to point at a node and say "this is where it failed". The compensating gain is adaptive behaviour and removal of the workflow maintenance tax. For the broader framing see why I bet against workflow platforms.

Self-hosted vs managed: the real cost ledger

Six line items.

Compute

Self-host n8n runs on your VPS or Kubernetes cluster. A small instance is roughly $20-50 per month on basic cloud providers. n8n Cloud Starter is around €20 per month for the equivalent volume (n8n pricing, retrieved 2026).

Storage

Workflow data, run history, credentials. Self-host means your storage. Managed means theirs. Cost is similar at small volume; cheaper at large volume for self-host.

Maintenance

Self-host means you run upgrades, patch security, restore from backups when something breaks. That is engineering time. Managed absorbs it.

Integrations

n8n's integration library is large; self-host means you have access to all of it. Some integrations require API keys you provide; that is bounded work.

Security and audit

Self-host means you own the security posture. For some buyers (regulated industries) this is the reason to choose n8n. For others (SMBs without security teams) this is a liability.

Total ownership

For a small team, the maintenance burden of self-host often exceeds the savings on compute. For a larger team with data-residency requirements, self-host is the right call. The break-even point is usually around team size and risk tolerance, not compute economics. See AI agent cost models.

Capability comparison

Dimensionn8nGravity
Unit of workWorkflow graphDeployed agent
DeploymentSelf-host or CloudHosted (self-host evaluated later)
LicenseFair-code (community), commercialCommercial
Customisation depthCode nodes, custom integrationsAgent definition + tool registry
Adaptive behaviourAI Agent node inside workflowAgent owns the loop
Best buyerDeveloper team with self-host needsFounder / ops lead without infra
Time to first agentHours to a working workflowUnder 60 seconds for deploy step

When n8n is the right choice

Three categories.

Self-host required by data residency

GDPR, HIPAA, government, banking. Some buyers literally cannot use a hosted platform. n8n plus the AI Agent node is the right answer.

Developer team in place

You have engineers who will run the n8n instance, write Code nodes, and extend integrations. The self-host model amortises across multiple workflows.

Stable workflow shape

The work is well-defined, the steps do not vary much, and you want fine-grained control. n8n's visual canvas is the right level of abstraction.

When Gravity is the right choice

Three opposite categories.

No infrastructure team

Founders, small ops teams, anyone who does not want to maintain a self-host. The hosted agent platform removes the infra burden.

Variable task shape

The work has exceptions that a workflow shape cannot capture in advance. The agent re-plans within the bounded outcome. See AI agent failure modes.

Non-developer buyer

The buyer cannot debug a node graph. They want outcomes. Gravity matches that shape. See three startups, three shutdowns for the founder context.

Pricing reality

n8n's pricing is published. Self-host is free under Community Edition. n8n Cloud Starter is €20 per month, Pro €50, and Enterprise on custom pricing (n8n pricing, retrieved 2026). Adding your own infra costs to self-host typically pushes the all-in figure to $40-70 per month for small ops.

Gravity is in pre-launch waitlist as of May 2026. Pricing is per-outcome by design intent and will be published at launch. See AI agent deployment models for the deeper context.

Frequently asked questions

What is n8n and why is it popular with developers?

n8n is an open-source workflow automation tool founded in 2019 in Berlin by Jan Oberhauser. It is Fair-code licensed and self-hostable, which is the reason for its developer-leaning adoption. The visual node-based editor is similar to Zapier and Make, but the self-host option and the developer-friendly extension model differentiate it. The repo has grown to tens of thousands of GitHub stars.

What is the n8n AI Agent node?

The n8n AI Agent node, added through 2024, lets a workflow include an LLM-driven step that can use tools, branch, and reason. It is built around LangChain primitives. Architecturally the AI Agent is one node inside the same workflow graph; it makes the workflow smarter than pure deterministic chains, but the unit of work is still the workflow.

When is n8n the right choice over Gravity?

When data residency requires self-hosting on your infrastructure, when you have a developer team that can run and maintain the n8n instance, when you want to extend with custom code nodes, and when your workflow shape is well-defined and stable. The self-host story is the main reason buyers stay on n8n instead of moving to hosted alternatives.

When is Gravity a better choice than n8n?

When the buyer is not technical, when the work is recurring with variance that the workflow shape cannot capture, when you do not want to run your own infrastructure, and when the cost of maintaining a self-host plus configuring nodes exceeds the cost of buying a finished agent product. The split is the familiar workflow-vs-outcome split, with self-host as the deciding factor for some buyers.

Does Gravity offer self-hosted deployment?

As of May 2026 Gravity is in pre-launch with hosted-only delivery planned at launch. Self-host options will be evaluated for enterprise customers with strict residency requirements after the public launch. For teams that need self-host today, n8n plus the AI Agent node is the practical choice.

Three takeaways before you close this tab

Sources