n8n + Claude growth automations, workflow by workflow

most “AI growth automation” content is theory written by someone who has not shipped a workflow. this is not that. this is what n8n and Claude actually automate in a real growth operation, including this site, which schedules and publishes its own content on a cron job with no one pushing a button.

What does automation actually replace in a growth operation

the parts that are tracking, not judgment. reporting a day’s ad spend against results, flagging which creative variants crossed a test threshold, checking whether a scheduled post is due, chasing down which subreddit comment got removed. none of that needs a person watching a dashboard. all of it needs to happen the same way every time, which is exactly what a workflow is good at and a person is not.

What’s an example that’s actually running right now

this site. every blog post can carry a future publish date, and a workflow checks twice a day whether a post’s date has arrived with no commit released since, then pushes the rebuild itself. no one logs in to hit publish. the same pattern runs behind client accounts: a scheduled check against spend data, a Claude call that reads the numbers and drafts a plain-english summary, a push to wherever the founder actually reads updates.

Can automation write your reddit posts or send your outreach for you

no, and this is not a soft no. the SAFE reddit playbook exists because the accounts that get banned are the ones that read like automation, and a workflow sending cold DMs at scale is the fastest way to lose an account, not grow one. what gets automated is the tracking around the human work: which subreddit converted, which account is due for its next post, which comment thread needs a reply today. the comment itself stays a person.

What’s the actual stack

n8n runs the orchestration: schedules, webhooks, branching logic, the glue between tools that were never built to talk to each other. Claude does the reasoning steps inside those workflows, reading a batch of numbers or a thread and producing something a person can act on in seconds instead of an hour. neither tool is doing the strategy. they are doing the parts of the strategy that are repetitive enough to specify once and never think about again.

Where does this break down, what won’t get automated

anything that needs taste. picking which of 150+ monthly ad creatives is actually good, deciding whether a subreddit fits a product, writing the comment that does not read like marketing, deciding when a founder needs to hear bad news directly instead of in a dashboard. automation reports the data behind those calls faster. it does not make the calls. treat any pitch that claims otherwise as a pitch for a worse version of your judgment, not a better one.

Where this shows up in the numbers

the reporting side of this stack is what makes it possible to actually manage $300k a month in ad spend and keep 150+ creatives a month in testing without the tracking itself eating the week. the automation buys back the hours. the decisions still take a person.

---

built and run on client accounts and on this site’s own publishing pipeline.