In my last post, The pipeline that runs itself, I described the sales pipeline I built for $30/month. Ten workflows. One server. Automatic lead discovery, outreach, follow-ups, reply monitoring.
That was the architecture. This is the first real test.
Life has been challenging for many of us lately. The world sometimes makes choices for you. I had to step away from my business for two weeks, without a handoff, without an out-of-office reply. When I came back and opened my dashboard, I wasn't sure what to expect.
Twenty-three opens
The first thing I noticed was a prospect who had opened my email 23 times across three separate days.
That's not a bot. That's someone checking whether you followed up.
He'd replied two weeks earlier, before I stepped away. I'd started a conversation. Then silence on my end.
The CRM showed the whole story: initial outreach sent, reply received (classified as "interested" by the AI), then 23 subsequent opens as he waited for me to respond. I hadn't.
The system tracked the signal perfectly. It drafted a follow-up. It flagged it for my approval. But I wasn't there to approve it.
This is the honest truth about semi-autonomous systems: they're only as good as the human in the loop. The automation caught the intent. It prepared the response. But the final step still needed me, and I wasn't there.
When I got back, I stared at that number for a while. Twenty-three opens. I'd built an entire system to catch exactly this kind of signal, and then I wasn't there when it mattered.
I sent a follow-up that afternoon, not sure if the window had closed. He replied within an hour. We're now in conversation about a project.
The system didn't close the deal. But it kept the door open long enough for me to walk back through it.
What the system found while I failed to show up
While I was ignoring that prospect, the pipeline didn't stop. It kept scanning, classifying, and filing.
One morning during my second week away, the lead discovery workflow pulled a funding announcement from an RSS feed. A small AI startup had just closed a pre-seed round, hired their first three engineers, and started posting about operational growing pains on LinkedIn. The system read the article, classified the company as a fit, researched their team, identified the founder, and added everything to the CRM with a draft outreach message ready for my review.
By the time I came back, there were five companies like that sitting in the pipeline. Researched, scored, and waiting. I hadn't found them. I hadn't even known to look for them.
That's the part that surprised me. The 23-open prospect was a failure of the human in the loop. But the five new companies were a success of the system running without one. The discovery, the classification, the research: those don't need judgment calls. They need consistency. And consistency is exactly what automation is good at.
Total cost for two weeks of autonomous operation: about $2 in API calls. All on a $5/month server.
I expected to have to start again from scratch
I thought I'd come back to a gap, to having not made any progress in two weeks on this brand new business: silence, missed signals, dead leads. Instead I came back to a pipeline that had kept working, a CRM that had kept filling, and a prospect who was still waiting for me.
If I had invested a bit more time in automating the last few steps, more would've gotten done. I'm still getting comfortable with having the AI tools do direct replies to humans unsupervised. If I had set it up to do it maybe the 23-open prospect would've already been a booked discovery call or more.
I built the system so far to catch those signals only. I had to do the next steps and I still dropped the ball. But the discovery, the tracking, the drafting, the monitoring: those kept running. And when I came back, I didn't have to reconstruct two weeks of missed activity from memory. It was all there, timestamped, categorized, and waiting.
These last two weeks are nudging me towards trying to drive my automation further, so that when life happens my business can keep going.
That's not efficiency. That's resilience. And for a small business, resilience might be the most valuable thing you can build.
Alex Monegro builds automation systems for startups and SMBs. If your business depends on one person's memory to keep running, that's the problem he solves.
Want to know where you stand?
Take our AI readiness quiz. 10 questions, 2 minutes. You'll get a personalized recommendation for where to start with automation.