FILED FROM BERLIN — n8n's $60M Series B, announced in late November 2025, is the round that tells the Bulletin the workflow-automation category has caught up to the autonomy-layer thesis. The company's pitch — a self-hostable, open-core workflow platform that ships agentic primitives as first-class citizens — is the cleanest working example of how a pre-autonomy-era category integrates the new substrate without surrendering its existing developer base.

This piece is the working profile.

The company, briefly

n8n was founded in Berlin in 2019 by Jan Oberhauser. The original positioning was as an open-source alternative to Zapier and Make: a workflow-automation platform that the user could self-host, with a fair-code license that protected the project from cloud-vendor enclosure. The product grew steadily through 2022 and 2023. By 2024 it had a real commercial layer, a serious enterprise customer base, and a developer community that was unusual for the workflow-automation category in the depth of its technical engagement.

The 2025 evolution of the product is the part the Bulletin tracks. n8n has spent the last eighteen months shipping agentic primitives into the workflow engine — LLM-powered nodes, multi-agent orchestration patterns, structured output handling, conversation-aware sub-workflows. The product, by late 2025, is recognizably an autonomy-layer platform with a workflow-automation user interface.

The $60M Series B was led by Highland Europe with participation from existing investors. The valuation was not disclosed but is reportedly in the $300M-$500M range. The structural story behind the round is the part that matters more than the dollar figure.

The thesis behind the round

n8n's investors are betting on a specific structural claim: that the workflow-automation surface is the right place for agentic primitives to land at the application layer, and that the user populations who already understand workflow-automation tools will be a faster adoption pipeline for autonomy-layer capability than the new-tool-shaped user populations the venture press tends to focus on.

The bet is defensible on the merits. Workflow-automation tools have, historically, been the entry point for non-engineers who needed to compose automations across systems. The category is well-understood by mid-market IT teams, by operations staff at larger enterprises, by the long tail of independent consultants who build automations for their clients. The user population is, in our reading, in the tens of millions globally. None of those users wake up wanting to learn a new mental model for agentic systems; many of them already have a workflow-automation muscle that the new substrate can plug into.

This is the integration moat n8n is betting on. The autonomy-layer companies that build for new user populations have to teach the mental model. The autonomy-layer companies that build for existing workflow-automation users inherit the mental model and just add the new substrate underneath it. The structural advantage is non-trivial.

What the product actually does

The 2025 n8n product is a workflow-automation engine with agentic primitives as first-class node types. The user composes workflows in the existing node-based UI; the workflow can include LLM nodes for reasoning, multi-step agent nodes for delegated work, structured-output nodes for ensuring the agent's response fits the next downstream node's expectations, and conversation-aware sub-workflows for multi-turn interactions.

The structural decisions are the part to pay attention to. The agentic nodes are first-class, not bolted on. The workflow engine treats agent failures the way it treats any other node failure — with retry, fallback, and dead-letter routing. The agent's state can be inspected, replayed, and audited through the engine's standard observability. The integration with the existing workflow nodes is direct; an agent's output can drive any of the platform's 400-plus integrations without intermediate translation.

This is more interesting than it sounds. The Bulletin has been arguing for a year that the integration substrate is the most under-discussed layer of the autonomy stack. n8n is the working example of why; the integrations are the part of the platform that does the most commercial work, and the platform that already has 400 integrations starts from a structurally different position than the platform that has to build them.

The open-core question

n8n's commercial structure is the most interesting open-core posture the Bulletin has covered. The core engine is fair-code licensed; the user can self-host without restriction. The hosted version is commercial. The enterprise edition adds compliance features, SSO, advanced observability, and the kind of operational capabilities mid-market and enterprise IT teams require.

The structural lesson is that the open-core posture, executed cleanly, is one of the more durable commercial models in the autonomy layer. The open core attracts the developer community. The hosted commercial layer serves the customers who do not want to operate the engine themselves. The enterprise edition captures the high-end commercial value. The model has been criticized in the past as an attempt to have both halves; n8n's execution suggests that the criticism is wrong when the company has the engineering discipline to maintain both halves equally.

What we are watching

Three things specifically.

The first is whether n8n's agentic-primitives adoption curve is steep enough to justify the round's valuation. The investors are betting on a specific commercial inflection; if the inflection arrives slower than the cohort expects, the company is over-valued. If it arrives faster, the round is a bargain.

The second is whether the workflow-automation thesis gets validated by other rounds in the category. Zapier, Make, Pipedream, Tray, and the long tail of workflow-automation tools are all either shipping agentic primitives or considering it. If two or three of them close rounds at comparable scale over the next eighteen months, the workflow-automation thesis is consolidated. If none does, n8n's bet is more isolated than it currently appears.

The third is whether the open-core posture survives the commercial scaling. The base-rate behavior is that open-core companies eventually tighten the core's licensing terms as the commercial pressure increases. We will track n8n's license history with the patience the model deserves.

The Bulletin's editorial position is that n8n is one of the cleaner working examples of an autonomy-layer company that arrived at the category through pre-autonomy-era infrastructure, and that the structural pattern — workflow-automation engine extended into agentic primitives — is the pattern most likely to absorb the long tail of the autonomy-layer addressable market.