The full office hours livestream above covers the architecture, memory system, and cost breakdown in detail.
I got my ADHD diagnosis a couple of years ago.
The thing that crushes me most on any given day isn’t the volume of work. It’s open loops. I do a thing, and the next step requires waiting on someone else or something else to happen.
Those loops don’t just sit quietly in the background. They take over.
I worry I’ll never come back to them.
The stress of the open loop makes it impossible to think about anything else.
It’s partly the object permanence side of ADHD, but it’s worse than that.
It’s not that I forget the loop exists. It’s that I can’t stop thinking about it.
I’ve tried every project management tool, task manager, and knowledge system you can name. They all had the same problem for me: they became a junk drawer I had to keep organized, which is its own open loop.
Nothing stuck because nothing actually worked like my brain works.
Building for my brain, not against it
When I started building with Claude Code, I realized I could stop adapting to tools and start building tools that adapt to me.
The first thing I built was a project management system designed through an interview. I sat down for about 45 minutes and explained the actual problem: I run multiple businesses, the work is interconnected, and if I look at all of it at once I get overwhelmed.
We went back and forth, one question at a time, until we had a spec that made sense for how I actually think.
The result has a “Now” view that shows me exactly one thing per project - the most important next task - unless something is overdue. Not a dashboard of 47 items. One thing.
But the real unlock was energy tagging.
Surfing energy instead of fighting it
Every task in my system gets automatically tagged by type: quick win, deep work, people work, creative. An LLM reads the task and its context and assigns the tag. No effort from me.
This means I can sit down and match my work to my energy.
If I need momentum, I pull up quick wins. If I’m in flow, I find deep work. If I’m in a people mood, there’s a filtered view for that.
I’ve noticed that most productivity systems batch tasks by project or deadline. That never worked for me.
Within any batch of tasks, they require completely different kinds of energy.
Grouping by energy type instead of project was a small change that made the system feel like mine.
I don’t track time. I don’t bill hours.
I’m not optimizing for productivity - I’m optimizing for quality of life. The system is designed for that.
Closing the loops
The open loop problem required something deeper than a task list. It required trust.
So I built background jobs that run all day.
Every 5 minutes, my system syncs Claude Code sessions to a database.
Every 15 minutes, it extracts memories - decisions, patterns, commitments, lessons - and stores them with embeddings for semantic search. When I chat with my assistant, relevant memories get injected automatically based on context.
I don’t have to remember to follow up.
I don’t have to remember where a conversation left off. The system holds it for me.
And that’s the part I never had before. Not a tool that could organize things. A tool I actually trust to hold things so my brain can let go.
The real shift
None of this is technically impressive. A database, some scheduled jobs, an LLM reading transcripts and extracting patterns. The pieces are boring.
What’s not boring is what it feels like to use it.
For the first time, I can put something down and trust that it won’t disappear.
The open loop still exists, but the stress doesn’t.
My brain is free to work on what’s in front of me instead of running a background process on every unresolved thread in my life.
Personal software can feel different because no one understands your problem quite like you do. Not better, not worse - you understand it because you feel it.
The possibility of building software where the feedback loop of “that felt good, do more of that” or “that felt bad, what are the other options” actually shapes the tool - that changes the equation entirely.
I’m not building this because I like building software. I’m building it because open loops were eating my life, and now they’re not.