AI saved my ability to hold complex projects together

I have ADHD, and for most of my life my projects followed the same pattern.

I’d get excited about an idea, go deep for weeks, spin off three more ideas from inside the first one, lose the original thread, then start over somewhere else.

The ideas themselves were not the problem. The problem was continuity.

I could not hold the full shape of what I was building across enough time to actually finish it, refine it, or even explain it clearly to other people.

At one point I spent a week trying to explain one of my projects to around 50 people in my life, and almost nobody could fully understand it. Partly because I could not consistently hold the whole thing in my own head long enough to communicate it.

So I built systems everywhere. Notes. Documents. Voice memos. Half-connected tools.

But everything still felt fragmented.

I started building a persistent memory system around Claude.

Not just a chatbot, but a system that captured decisions, reasoning, project state, and context over time so the AI could hold continuity even when I couldn’t.

The biggest shift was realizing I needed to capture the reasoning behind decisions, not just the tasks themselves.

One moment made that real.

I have an automated aquaponics system in my basement. It's an indoor growing setup that combines water circulation, sensors, lights, pumps, and software automation to grow plants and monitor the environment.

Over time, parts of the system silently drifted. A coding session had rewritten a light controller using the wrong transport protocol. Sensors were returning fake values for probes that were not even wired yet. My network tunnel had expired without me noticing.

The system was still running, but it was drifting further and further away from what I had intended.

When I finally audited it, the persistent memory system held the original decisions I had made months earlier about how each component was supposed to work.

The gap between intent and reality became immediately obvious.

Without that captured reasoning, I would have been troubleshooting symptoms instead of seeing the underlying structural problem.

That changed how I thought about everything.

The system was no longer just helping me remember information. It was helping me preserve continuity of thought.

The biggest change was continuity.

Projects stopped dying every time I context-switched. Decisions stayed made instead of being re-derived from scratch every few weeks. I could pause work and return to it later without losing the thread entirely.

That reduced stress in a way I did not expect.

I stopped trying to hold the full state of multiple projects in my head at all times because I trusted the system to hold what I was not actively thinking about.

My wife noticed it before I fully did. I could actually shut the work part of my brain off sometimes.

Claude does not fix the ADHD.

But it holds the shape of what I’m building when my brain can’t.

Step 1: Pick one project you keep losing the thread on

Step 2: Start capturing decisions, not just tasks

For example: Decided X because Y. Rejected Z because of A.

Step 3: Keep all project notes and reasoning in one place

Step 4: Set up a way for your AI tool to reference past context. This can be as simple as a shared document or as structured as a persistent memory system

Step 5: Ask the AI to help you resume work based on previous decisions and context

Step 6: Use the system consistently, even if it starts messy

Step 7: Build review habits so the system stays current instead of becoming another notebook graveyard