n8n is where a lot of teams build their AI automations: a trigger, a few nodes, a model call, and something useful happens. That is a good place to build, and nothing on this page asks you to move.

What tends to be missing comes later, when someone responsible asks three plain questions: which of our workflows call an AI model, how often do they fail, and what do they cost? In n8n the answers sit in each workflow and each provider’s bill. Copyl puts them on one page.

What Copyl shows from n8n

  • Every workflow, and which ones use AI. Workflows with AI nodes (n8n’s LangChain nodes and the OpenAI node) are listed separately from workflows without them.
  • Runs and failures. Runs per month and failed runs per workflow.
  • Model spend per workflow. Link each n8n model credential to the Anthropic, OpenAI or Gemini key it uses, and spend shows per workflow. When several workflows share one key, Copyl shows the key’s total and says so; it does not invent a split.
  • What is not yet linked. Model credentials without a linked key are listed, so the gaps are visible instead of silently missing.

What Copyl never stores

Node parameters, pinned data, workflow settings and notes, and the content of executions — that is, your prompts and the model output — are never stored. Copyl keeps workflow names and ids, node types and names, credential references, and for each run its id, time and status.

Two ways to connect

Pull — your instance is reachable over https. Enter the instance URL and an n8n API key, which is stored as an AI Service credential in Copyl. Copyl only lists workflows and executions, without execution data, and syncs a month at a time when you choose Sync month. Note that an n8n API key has access to the whole instance unless you use n8n Enterprise API scopes.

Push — your instance stays inside your network. This uses n8n’s log streaming, an n8n Enterprise feature. Copyl gives you a webhook URL and a source token to add as a Webhook destination in n8n. The token is shown once, stored only as a hash, and can be replaced at any time. Copyl holds no n8n credential at all.

Where Agent Hub comes in

Your n8n workflows sit in the same inventory as the rest of your AI spend outside Copyl. Agents you run in Copyl’s Agent Hub go further: versioned profiles, their own permissions, a human approval before anything consequential happens, and a full trail of what each agent did.

To be clear about the boundary: Copyl does not run, import or orchestrate your n8n workflows. They stay in n8n. Copyl gives you the overview and the numbers.

When you don’t need this

If you run one n8n instance with a handful of workflows, one model key and one person who knows all of it, the n8n UI and your provider’s billing page are enough. Copyl earns its place when workflows multiply, keys are shared, more than one team builds, or someone has to answer for AI spend across tools.