I Built an AI Planning Tool for a Workflow I Don't Actually Have
You build something, test it, it works perfectly — and then you forget about it. Thatwas me with my AI task prioritizer.
I’d built the whole thing in n8n for a course project. It pulled my calendar, found time gaps, asked Claude to suggest realistic tasks for the next day, and sent me a neatly formatted plan at 9 p.m. every night.
The first two emails came through. They looked great. Professional. Useful, even.
And then — nothing. A whole week went by before I realised I wasn’t getting them.
At first, I assumed I’d broken something. Maybe a credential expired or a connection failed. I logged into n8n ready to debug the whole thing.
That’s when I realised, the workflow was fine. The problem was actually my calendar. It was empty.
What I Actually Built
Here’s what the workflow did (on paper, at least):
Pulled tomorrow’s Google Calendar
Identifies gaps between scheduled events
Asked Claude to suggest 3–4 work tasks that fit those windows
Emailed a tidy plan for the next day
Technically, it worked perfectly. And it did exactly what I designed it to do.
But here’s what it didn’t know: my actual to-do list, real life priorities, and what I’m prioritizing this week.
It made educated guesses based on time windows. That’s it.
The Flaw I Didn’t See Coming
I haven’t been scheduling work tasks on my calendar.
I thought I did, I am sure I did at some point in the past or maybe I thought I should.
When I looked honestly at how I work, it was obvious, my calendar has baby naps, toddler play dates, appointments, school pickup. Not “write Substack post draft” or “finish design project.”
Why? Because I can’t guarantee those time blocks will exist.
Some days the baby naps for two hours. Other days, twenty minutes. My “work windows” aren’t predictable enough to schedule in advance.
So instead, I keep a mental list of three focus areas:
A design project I’ve been procrastinating on
AI tool testing and learning (for my course)
Substack content
When the baby sleeps, I glance at the list and pick whatever fits my time and energy. That’s my system.
No calendar. No prioritization tool. Just three buckets and a gut check.
I complicated the plan when all I needed was pen and paper.
What I Actually Learned
The workflow wasn’t wrong — it just solved a problem I don’t have.
If I worked full-time with predictable hours, it would be brilliant. The kind of thing that saves ten minutes of mental math every morning. You get to work, you have a list, you are ready to go.
But I’m not that person right now and bending my days to fit a tool I built would be ridiculous.
And honestly? Even if those emails had kept coming, I’m not sure I’d have wanted them.
When you’re already drowning in digital noise, another “helpful” daily email starts to feel like admin instead of support.
The real problem for me isn’t task prioritization. It’s figuring out how to make meaningful progress in ninety-minute sprints without burning out.
An AI triggered email can’t solve that.
That’s not an automation problem; that’s a life design problem.
The Uncomfortable Question
So what do I actually need?
I’m not sure yet. Maybe nothing.Maybe this chaos works fine for now.
Or maybe the real question isn’t “what should I work on today?” but “how do I move long-term projects forward when I only have 90 unpredictable minutes a day?”
I’m sitting with that one.
What This Taught Me About AI Tools
Building this workflow taught me something that applies to almost every AI experiment I’ve done:
Tools have to solve real problems — not theoretical ones.
“This would be useful for someone else” doesn’t count.
“This seems like it should help” doesn’t count.
“This is technically impressive” definitely doesn’t count.
“Wow, that’s pretty cool.” definitely doesn’t count.
The only question that matters: does it support the work I actually do, the way I actually do it?
If not, it’s not a tool — it’s noise.
This post originally appeared in my Substack newsletter. Subscribe for weekly updates.
