My Marketing Team Started at 11 PM. I Was Asleep.
At 11:00 PM last Tuesday, my marketing team got to work.
I had been asleep for about an hour.
By 7:00 AM the next morning, they had identified 10 new leads, caught 4 brand violations before they went out the door, run a full technical SEO audit, written 3 new Notion tasks based on competitor gaps, and generated personalized outreach messages in Dutch, English, and German — each one tailored to a specific company, a specific trigger, a specific person.
I had done none of it.
What Actually Happened That Night
Here's the exact sequence.
11:00 PM — Follow-up Agent. It scanned my lead pipeline and checked which contacts needed follow-up. It also read through the outreach messages queued for the night and found 4 instances where the wrong title was used. Not "Fractional CMO" — something else. It flagged all four, blocked them from going out, and wrote a correction task. Nothing sent, nothing broken.
12:00 AM — SEO Agent. It ran a technical check on the site, then analyzed competitor rankings for "fractional CMO Netherlands." What it found wasn't great: I was barely visible for the searches my actual clients would be using. It logged 3 Notion tasks with specific fixes and keyword opportunities.
1:00 AM — Outreach Agent. It picked up the 4 flagged messages, corrected them, and moved on to generating batch 4 of new leads — 10 companies, B2B SaaS scale-ups with €1M+ ARR, all in the middle of European rollouts. For each one: a personalized message, the right language, a specific trigger (funding round, product launch, market expansion). It also spotted 3 outreach tactics my competitors are already using that I'm not.
7:00 AM — Me. Coffee. Open the briefing. Review what happened overnight.
Why I Built This
I work as a Fractional CMO. The pitch is simple: better marketing than an agency, faster, cheaper, with actual senior ownership of results. I've done this across 200+ companies over 18 years, managing more than €200M in ad spend. I know what good looks like.
But there's a hard ceiling on what one person can do. You can be senior and sharp and fast — and still only have 24 hours in a day.
So I asked myself a question: the systems I build for my clients, the growth infrastructure, the automated pipelines — why am I not using those for myself?
The answer was that nobody had built them for me. So I did.
How the System Works
It's 13 agents running in a pipeline from 11 PM to 7 AM.
Each agent has a role. Each one reads the output of the agents before it and writes handoffs to the agents after it. If the SEO Agent finds a keyword gap, the Content Agent picks it up the next morning. If the Branding Manager catches an error, it blocks the output, writes a correction task, and the Correction Agent propagates the fix across every active file before anything goes out.
The agents communicate through shared handoff files — not a live API, just structured text documents that the next agent reads before starting. Simple, transparent, debuggable.
The tools underneath: Claude Code as the foundation, Notion as the shared board, Google Search Console for SEO data, Gmail for outreach, Cal.com for scheduling. Total monthly cost: a small fraction of a full-time employee.
The First Week — Real Numbers
Here's what the system produced in week one:
40 leads identified — B2B SaaS scale-ups, €1M+ ARR, European expansion phase
Personalized messages per lead per language — NL, EN, DE — with specific hooks per company
Full competitor SEO analysis — keyword gaps, ranking comparisons, technical issues
3 complete blog articles — including this one
12 LinkedIn posts — queued for the next two weeks
Brand violations caught and corrected before anything went out
None of those numbers are rounded up. That's what came out of the first seven nights.
What It Is Not
I want to be precise here, because the hype around AI automation is thick and I have no interest in adding to it.
The system does not make decisions.
Outreach messages are not sent without my approval. Blog posts don't go live without my review. LinkedIn posts don't get published without me seeing them first. The agents draft, flag, organize, analyze, and correct. I decide what goes out.
The system is my back office, not my replacement.
It handles the work that used to happen in the margins — the lead research at midnight, the SEO check I kept putting off, the competitor analysis I never had time for. Now that work happens every night, whether I'm in Valencia having dinner or asleep at 11 PM.
I still do the thinking. The strategy, the positioning, the decision on whether a message is actually good enough to send — that's mine. The agents make sure I'm not starting from scratch every morning.
What It Took to Build
This was not a weekend project.
It took weeks of iteration: defining each agent's role, testing the handoff protocol, catching the edge cases (the wrong title slipping through, a stat that didn't match the brand guidelines, an outreach message that was technically correct but tonally off for a German audience).
The first version had bugs. The Branding Manager caught things the other agents missed. The Correction Agent broke a file format once. I fixed it, documented it, tightened the system.
By week two, the error rate had dropped. By week three, the pipeline was running clean.
Building it took time. Running it takes almost none. And it scales — every new market, every new language, every new outreach batch adds almost zero marginal effort on my side.
That's the point.
If you're a Fractional CMO, a consultant, or a one-person growth team — this is the infrastructure question nobody is asking you yet: not which tools you use, but whether your tools work for you while you sleep.
Mine do.
Bart Knijnenberg is a Fractional CMO with 18 years of experience. He has helped 200+ companies across the Netherlands, Belgium, Germany, and Spain, managing over €200M in ad spend.
