Every project management tool I’ve ever used had the same problem.
Not the features - the expectations.
They all expected me to be the kind of person who enjoys curating a Notion database or getting satisfaction from checking a box.
I’m not that person.
I know people who find that relaxing. I’ve tried to be one of them.
But for me, the maintenance work of keeping a system current always outweighed the benefit of having a system at all.
So I’d abandon it, and then I’d have no system, and then open loops would pile up until they became a real problem.
This is an executive function thing. When something is out of my hands - delegated, waiting on someone else - my brain keeps circling back to it.
Not because I’m anxious, but because I don’t trust that I’ll know when the ball gets dropped.
I’ve never had a system that caught that for me.
That’s what I set out to build.
The dream
Somewhere in the back of my mind, I’ve always had this picture: you wake up in the morning and your day is prepared for you.
Not by an app that shows you a list of everything that’s overdue. By something that understands your priorities, looks at your calendar, checks what’s actually urgent versus what can wait, and tells you what to focus on first.
That’s the dream of an assistant. A real one. Someone who absorbs the operational weight so you can show up and do the work.
I built that with Claude Code.
Every weekday at 8:30 AM, a set of agents looks at my calendar, my inbox, my tasks, my projects, my relationships, and my reminders.
It generates a briefing that tells me what matters today, what order to do it in, and what I can safely ignore until tomorrow.
The executive function problem I described above? This solves it directly. I don’t have to figure out my day before I can start my day.
What it actually replaced
I run multiple businesses that overlap in unusual ways. Indy Hall, Stacking the Bricks, the 10,000 Independence Project, personal projects.
Same collaborators showing up across different contexts. Information scattered across email, texts, Discord, meeting notes, and a dozen other places.
Before this, I was stitching it all together manually.
Calendar in one app, tasks in another, relationship notes in my head, event management in a third tool, newsletters through a fourth. None of them talked to each other.
The JFDI system replaced all of that with one environment where everything is connected.
My relationship manager tracks the people in my life - not like a sales CRM, but as a way to notice when a relationship is active, when it’s cooling off, and when it might need attention.
My meeting system pulls context about the people I’m about to sit down with so I can show up fully present instead of drawing on memory.
After the meeting, I brain-dump my notes and it processes them into takeaways, decisions, and follow-up reminders automatically.
None of these pieces are new on their own.
What changed is that they’re wired together.
When I save an article to my knowledge base, it checks my relationships and suggests people I might want to share it with. When I prep for a meeting, it pulls from recent projects, past conversations, and active reminders about that person. Everything feeds everything else.
The part that surprised me
I did not set out to build project management software.
But once I had tasks, relationships, events, and context all in one system, it became obvious that the missing piece was a way to see what to do next. Not a list of everything - a filtered view based on my current energy and the actual priorities.
So I built one. Quick wins when I’m unmotivated. Creative tasks when I’m energized. Deep work when I’m in flow. It lets me approach my work intentionally, using my energy as a tool instead of a constraint.
That reframing - from reactive to proactive - changed how I move through my day more than any single feature.
What I learned about building for yourself
The whole system took about three weeks of actual development using Claude Code.
I hadn’t built software in 15 years. The fact that I could build something this personal and this useful says more about Claude Code than it does about me.
But the real lesson is about SOPs.
Almost everything meaningful in this system is a set of instructions. How to prep for a meeting. How to process a newsletter. How to set up an event. How to handle an email that needs a response.
I found that writing clear instructions for an AI system is the same skill as writing clear instructions for a human assistant - and I’d never been good enough at that to make delegation work.
This system forced me to get specific.
And now, if I ever do hire a human executive assistant, the playbook already exists. The SOPs that run the AI are the same SOPs that could train a person.
The honest version
The system makes mistakes. Dates and times are the most common failure - multiple dates in one request can confuse it, same as they’d confuse a human.
The difference is that when I catch an error, I can write a rule to prevent it from happening again. The system gets permanently better in a way that no tool I’ve used before could.
I had my first thousand-dollar week of Claude Code overages building this.
Worth every penny. Not because the system is flashy, but because problems I’d been complaining about for 20 years are actually solved now. Not worked around. Solved. In exactly the way I want them to be.
Why I’m sharing this
When I look at the AI conversation online, it’s mostly hard lines. Good or bad. Useful or not. Hype or dismissal.
My experience doesn’t fit neatly into any of those.
I’m not building software for a living. I’m a person who runs businesses and needed a better way to stay organized, follow through on commitments, and show up for people consistently.
Claude Code gave me the ability to build exactly that - something no off-the-shelf tool was ever going to provide, because no off-the-shelf tool knows how my work actually works.
This system doesn’t do the work for me.
That’s what a great assistant does. I just didn’t expect to build one myself.