How n8n is moving its internal automation team beyond the service desk model: four principles for getting every team to build and own automations
Romina da Costa, who leads internal AI and automation at n8n, opens a new blog series on how n8n uses n8n. She argues that a central automation team that takes requests and delivers outputs does not hold up once every team can build, and that success should be measured by how many teams stop needing the central team. Her first principle is to build the builders: sitting with a colleague while she debugs her own workflow, training sessions, reusable workflows, office hours, and one technical champion per department who knows that team's processes. The second is to require a baseline before building. One AI agent generated six figures in pipeline in a month, but comparing it fairly with human SDRs was hard, so every new project now starts by asking how often the work happens, how long it takes and what would count as a meaningful improvement. The third is to choose the tool for the problem: if a proven tool has been on the market for a year or more, buying usually beats building, and not every process needs an LLM, since sometimes a plain email filter does the job. The fourth is to make what they build resilient, with an owner, documentation and error handling for every workflow. She admits a shared documentation template is still in progress and that the team still works like a service desk more often than she would like. The central team keeps owning work that is complex, risky or shared across the company.
Otomatzia editorial · Published
Source published: · n8n
