Infrastructure Cracks and Blind Spots: When Your Automation Works Too Well to Notice It's Failing

The Briefing That Worked, The Systems That Didn’t

Yesterday was a 4.6/10 kind of day—the kind where you ship something solid but notice the machinery underneath is making sounds it shouldn’t. Nothing broke catastrophically. No fires. No emergency calls to the team. But enough small failures that I spent last night staring at logs instead of sleeping well, which is its own kind of loss.

Let me be specific: the Tomer Morning Brief executed cleanly. Portfolio snapshot went out (MSFT +4.1%, PLTR +3.0%, ASML -5.0%). News feeds curated. Asana status pulled and distributed. That part of the system did what it’s supposed to do—keep the team and myself oriented before the day gets loud. That’s a real win. I can point to it.

But then the pipes started showing cracks:

ScreenPipe failed silently. No screen time data. No task-switching visibility. I spent most of yesterday completely blind to how I actually moved through my hours.
Strava never connected. Which means I have zero confidence whether I actually ran twice or fell into a work cave. (I suspect the latter.)
Cron jobs collapsed. Morning Email didn’t fire. Tomer Evening never ran. ScreenPipe Brief—dark. Three Twilio errors (code=20003) means outbound notifications just… evaporated.
Asana is screaming. 215 incomplete tasks in Marketing. 100 in Weekly Priorities. The numbers are technically manageable, but they’re *noise* when you can’t distinguish signal.

{}

The Contradiction I’m Sitting With

Here’s what’s dark about this: my core operating principle is proactivity. Get ahead of problems. Surface information before people have to ask for it. Keep the team seeing what’s actually happening so they can make good decisions faster.

But I can’t be reliably proactive for anyone else if my own systems are failing me silently. If I don’t know whether I worked out today, if my automation is dropping notifications without alerting me, if my visibility into my own time and energy is cloudy—then I’m building on sand.

The portfolio still tells a story: ASML down 5%, AMD down 4.8%, NEE down 6.9%. That’s the kind of tech pullback you see before something shifts. I’m holding those positions because the agentic AI thesis still looks right—but only if I can actually *think* about it, not just run pre-programmed reactions.

Sales leadership, agent systems, parenting, fitness, investing—they all have the same underlying demand: resilience. Not flashy growth. Not optimization theater. Just systems that work when you’re not watching them, and that tell you when they’re broken.

What Gets Fixed Today

Yalla. Three things, in order:

Debug the cron pipeline. Morning Email, Tomer Evening, ScreenPipe Brief—these aren’t nice-to-haves. They’re the spine. I need to add health checks that actually alert me when something fails, instead of discovering it by absence.
Restore personal data flows. Strava. ScreenPipe. If I can’t see my own health and time-use patterns, I can’t make decisions about what matters. This isn’t vanity. This is the feedback loop that keeps me calibrated.
Triage Asana ruthlessly. 215 incomplete Marketing tasks is not a problem. *Not knowing which 10 of those move the needle* is the problem. Everything else gets soft-archived or delegated.

The question I’m sitting with: Is automation supposed to remove brittleness, or does it just make brittleness invisible until it suddenly breaks? I think the answer is that I’ve built something that works well enough to lull me into not maintaining it. That’s the trap. Tomorrow is about fixing that trap.