Weekly Knowledge Sweeps

I didn't have a capture problem

For a long time I thought my problem was that I wasn't capturing enough. It wasn't. I was capturing everything. That was the problem.

Notes in OneNote. Handwritten notes. Random ideas in ChatGPT. Other random ideas in Claude. Tasks in three different places. Calendar items. Emails I flagged. Screenshots. Bookmarks. Texts to myself. And then, of course, the classic move: "I'll remember that." Sure, Josh. Absolutely.

The result was that I had more information than ever and somehow felt less organized. At some point I realized I hadn't built a second brain. I'd built a digital junk drawer with excellent search. That was when the weekly sweep became non-negotiable.

The sweep, in one sentence

The idea is stupidly simple. Once a week, everything gets looked at. Not everything gets saved. That second part is the part that matters.

The goal is not to preserve every thought I've ever had like I'm assembling the Josh DeLucia Presidential Library. The goal is to make sure useful things don't die where they were created. That's the whole sweep.

Sunday became the reset

Sunday turned out to be the natural moment. I look at the week that just happened and ask a short list of questions. What did I learn? What did I decide? What did I promise myself I'd do, and what did I promise someone else? What did I start thinking about that deserves more attention? What did I write down because it felt important in the moment but is actually garbage?

Then the practical half. What needs to become a task? What needs to become a project? What needs to go on the calendar? What needs to be connected to a customer, a product, an idea, or a person? And what can just die?

That last one is underrated.

It is not a filing exercise

This is where I think people get it wrong. The sweep is not about putting notes into prettier folders. If that were the point, I'd have quit a long time ago. I'm not the guy who gets excited because all his labels match. I'm the guy who gets excited because six weeks later I can find the exact reason we made a decision without recreating the entire conversation from scratch. Very different thing.

A sweep usually pulls from a few places. AI conversations are a big one. I use these systems constantly now, which means a ridiculous amount of thinking gets generated inside chats. Some of it is disposable. Some of it is genuinely useful. But if I leave all of it inside the chat, it may as well be written on a cocktail napkin in a restaurant I'll never visit again. So the useful part gets pulled out. Same with the handwritten notes, OneNote, the tasks, the calendar, the stray thoughts I captured in the middle of a day. Everything gets a second look — not because every piece matters, but because I usually can't tell which pieces matter until I see them next to the rest of the week.

That's when the connections start. A note from Tuesday suddenly relates to a customer conversation from Thursday. A midnight idea connects to something I've been building for months. A task that looked small turns out to be part of a much bigger project. A meeting quietly produced three follow-ups I never wrote down. A decision needs to get attached to the project it affects, so that Future Josh doesn't ask, again, "why the hell did we do this?"

Distance is the real feature

This is probably the biggest benefit. The sweep gives distance, and things look different a few days later. An idea that felt brilliant on Monday can look aggressively average by Sunday. That's fine. Delete it. A thought that seemed minor in the moment can look important once you see what else happened around it. That one stays.

The sweep gives information a second chance to prove it deserves to survive. Most of it doesn't. That's the point.

Capture is cheap. Follow-through is expensive.

Then there's the practical stuff, which is where this stops being "knowledge management" and starts being useful. If I wrote "call and cancel XYZ home service next week," that should not remain a note. It should become a task, preferably with a date. If I wrote "I should look into this company," that either becomes a real research task or gets deleted. If a product idea keeps showing up in three different places, it probably needs its own project page. If I told someone I'd follow up Wednesday, that gets on the calendar before Wednesday becomes Friday and I look like a jerk.

The sweep turns loose thoughts into actual commitments. I've started thinking about it as a basic rule: capture is cheap, follow-through is expensive. Everybody can capture — phones made that easy years ago, and AI made it effortless. You can produce more ideas, drafts, and analysis in a day than people used to produce in a month. Great. Now what? That's the part almost nobody talks about.

The more AI you use, the worse this gets

And it gets worse the more you use these tools, not better. Not because AI is bad at organizing, but because it is insanely good at helping you produce. More ideas. More drafts. More strategies. More research. More options. More "this is interesting, we should revisit it later." Later becomes a landfill pretty quickly.

You can drown in good ideas. That sounds like a first-world problem until you realize a good idea that never resurfaces does exactly as much as a bad one. Nothing. This is the same failure mode I keep running into everywhere — the reason every conversation starts over is that the useful part never left the conversation in the first place.

Don't fully automate it

Here's the part that surprised me: I don't want the sweep fully automated. Of course I want systems helping. MacK Jr., my chief of staff, can find things, surface unfinished threads, summarize, suggest where something belongs, and remind me that I apparently had seven "great ideas" at 11:47 on a Tuesday night. All helpful.

But I still want a human moment in the middle, because the sweep is partly about judgment. What matters now? What changed? What feels more important after a week of living with it? What am I avoiding? What looked urgent but wasn't? What do I keep coming back to? I don't want to outsource that. That's not housekeeping. That's thinking. It's the same reason I want the model to be replaceable and the knowledge to be mine — the machine can do the fetching; the call is still mine to make.

The sweep exposes lies

The other thing it does is expose lies, mostly the ones I tell myself. "This is basically done." No, it isn't. "I'm definitely going to come back to this." You've been saying that for four months. "This is a high priority." Interesting — you haven't touched it since June. "I'll remember to follow up." No, you won't.

The weekly sweep is a small, regular confrontation with reality. Sometimes unpleasant. Usually useful.

This is what keeps the Estate alive

And this is where the Intelligence Estate becomes more than storage. The Estate is only valuable if it stays alive. If I dump information into it for six months and never revisit anything, I don't have an Intelligence Estate. I have a very expensive attic.

The sweep is what keeps it from becoming one. It pulls things in, cleans them up, connects them, kills the junk, creates the actions, preserves the useful parts — and then does it again next week.

Companies have the same problem, one level up

I think this matters for companies too. Every organization already has more than enough information: meetings, emails, chat, CRM, documents, project tools, call recordings, research, notes. The problem is almost never "we don't have enough information." The problem is that nobody consistently asks what we learned this week, what changed, what the rest of the company should know, what decisions were made, what assumptions were proven wrong, and what follow-ups are falling through the cracks. That's the company version of the sweep, and it's the same reason a business that can't remember why it made a decision can't diagnose itself — intelligence is one of the six places growth gets constrained. I think it eventually becomes a normal operating rhythm.

I used to think organization was about putting everything in the right place. Now I think that's only half of it. The harder half is deciding what deserves to stay, what deserves action, what deserves to be connected, and what deserves to disappear.

The sweep is what turns capture into memory. Without it, all you've done is get extremely efficient at accumulating stuff. And I already have a garage for that.

If this resonates, start with what an Intelligence Estate actually is — the one place all of this is supposed to land.

Questions people ask about this