Introducing the n8n to Tray migration kit

n8n is open source, self-hostable, and fast: drag a few nodes together, wire in an API, and you’ve got something working before lunch. That speed is why so many teams start there, and why a pilot usually works.

That’s exactly the problem. The moment a pilot starts to matter, security wants to know who’s patching the instance, compliance wants proof of controls, and whoever’s on call wants to know what happens when it breaks. Building it was never the hard part. Getting it live, and keeping it running as volume grows, is.

A pilot doesn’t ask much of you. Production does. Here’s what changes once it’s not just yours to run anymore.

You become the platform team

Self-hosting means you own patching, and n8n has had a real run of critical vulnerabilities recently. The controls a compliance review will ask for, RBAC, SSO, audit-log streaming, sit behind n8n’s Enterprise tier, so below that you’re building your own version of all three. And dedicated support is Enterprise-only too, so when something breaks outside business hours, the answer is a community forum thread, not a number to call.

Agents multiply faster than anyone can govern them

n8n’s AI and MCP tooling is genuinely good, which is exactly why this gets out of hand. Any builder can stand up an MCP server and point an agent at a production system, with no central place to see who can call what. Fine with three workflows and one builder. A liability once ten people have done the same thing.

Volume breaks patterns that worked fine in a pilot

Real traffic exposes workarounds a demo never needed: batching large jobs into sub-workflows, building your own polling loop to check on a long-running process. The quiet version is worse. If two branches feed a merge with different numbers of items, n8n combines them by position and stops when the shorter branch runs out. The extra items don’t error. They just don’t make it through.

In a pilot, you’d never notice. At real volume, that’s a data-integrity problem with your name on it.

None of this means n8n was the wrong call. It’s a genuinely good way to build fast. It’s just not built to be the thing you run once a pilot has to survive a security review, a compliance audit, or a real traffic spike.

The licensing looks cheap on paper. The real bill shows up two other ways.

If you self-host, the software is free, but running it isn’t. Patching, monitoring, backups, and logs add up to real hours every week, on top of a genuine infrastructure bill before you’ve spent anything on security or database work.

If you’re on a paid plan, n8n’s 2025 shift to execution-based billing changed what a normal workload costs. A simple automation polling every ten minutes can burn through an entry-level plan’s execution cap in days, and the jump to the tier that includes SSO, audit logs, and separate environments is steep enough that n8n’s own community pushed back hard on it when the change landed.

Free to start rarely means free once it’s running in production, and the tier that finally includes the governance controls you need is usually the same one where the bill jumps hardest.

Moving off an integration platform has always meant months of manual work: reading every workflow, rebuilding the logic node by node, and hoping nothing gets lost in translation. That’s the part the kit removes.

The n8n → Tray Migration Kit, powered by Tray Headless, isn’t something you download and run yourself. Here’s the actual flow: you talk to a Tray team and share your n8n flows with them, and from there the kit does the technical work of turning your n8n workflows into native Tray workflows. Concretely, that means:

1. It reads your n8n export and inventories every node, connector, and expression in each workflow.
2. It maps each connector to its Tray equivalent with a confidence score, so anything ambiguous gets flagged for a human to check rather than silently guessed at.
3. It translates expressions, n8n’s JavaScript-based syntax, into the equivalent Tray logic.
4. It works out build order by mapping dependencies between workflows, so nothing gets built before the thing it relies on, and you don’t hit missing-dependency errors partway through.
5. It recreates the workflow natively on Tray, not as something that runs on top of n8n or points back at your old instance. What comes out the other side is a Tray workflow you own outright.

Every converted workflow still gets reviewed before it goes live. The audit trail and documentation the kit generates automatically is what makes that review fast, not what skips it.

• Becoming your own platform team stops being the default. Tray is SOC 1 and SOC 2 Type 2 audited every year. RBAC, SSO, and audit-log streaming come standard, not gated behind a higher tier, and contractual SLA and enterprise support come with the platform. A reviewer checks against something that already exists, and an outage becomes a support case with a response time attached, not a forum thread.
• Agent sprawl gets the same fix. The platform sets policy on every agent and MCP call and audits it, so there’s one place to see who can call what instead of every builder standing up their own MCP server.
• Volume stops being a workaround problem. One managed runtime for integration, automation, API management, agents, and MCP, with real observability, so high volume doesn’t quietly drop records the way an uneven merge can.

Pricing follows the same logic: sized to what you actually run, with governance included rather than something you pay more to unlock.

You don’t need to talk to anyone to find out where you actually stand. We built a free, self-serve n8n production readiness checklist: 25 checks across security, compliance, agent governance, reliability, and support.

It’s the right first step because it’s the one thing on this page you can do in the next ten minutes, without a sales conversation, without sharing your workflows with anyone. Answer honestly, see exactly which of the three things above are already true for your setup, and use it to decide what to bring to your next security review before someone else brings it up first.