Last week I shared on LinkedIn that I'm tired, and I meant something heavier than a long to-do list. Every day I wrestle with the grief of losing our son, and some days the weight of carrying that while also leading a company and being a dad is just more than I can hold. I put it out there because I'd rather say the true thing than mask up and tell everybody I'm fine.

Thank you to everybody who reached out after that one.
Then on Friday my wife Jordyn and I decided to do an impromptu trip to the mountains and sit at a cabin and stare at trees for two days.
1. The “mountaintop” time always ends the same way..
Here's the thing about going away to get your head right.. eventually you come back to reality.
I've done the version where you unplug completely, and it works for about the length of the drive home, because the whole time you were breathing mountain air the pile kept growing without you. So you get the reset and then you immediately spend it paying down the cost of having been gone. “Great, I feel amazing, but now there's even more work to do.”
That's the loop I've been stuck in, and it isn't really about vacation. It's the urgent stuff constantly superseding the important stuff.. all the things that would make my life easier are exactly the things I never get time to build, because something on fire always wins. So I never build them, and next week something else is on fire.
2. So I hit record
Jordyn and I didn't plan this part.. it just sort of happened. Instead of going up there to escape work, we went up there to talk about it out loud with nothing else competing for the time, and every time we started a real conversation I turned on my AI notetaker.
Seven recordings. A little over seven hours across two days.. the company, the calendar, what's working, what isn't, what I'm avoiding, things like that.
Then Sunday afternoon I sat down for a couple of hours with my AI chief of staff (I call him James) and had him read all of it.
He came back with 4 docs that summarized current state of the company, my current calendar, what an ideal week would look like, and some of the stuff that’s been weighing on me. Then it drafted emails, put things on my calendar, scheduled the recurring stuff, stood up calendar booking links, and did some research I'd been putting off.
So I came back down the mountain and the pile was smaller than when I left.
I want to be careful not to oversell that, because I still have a lot of stuff to do and most of it still needs my judgment BUT it was the first time in a long time that resting and getting things done weren't competing for the same weekend.

3. Here’s what it actually looked like
None of this is a product you can go buy, so here's what's underneath it.
Everything runs on a VM (I’ve shared this in past newsletters). Claude Code writes scripts and those scripts run on schedules. Some of them are watchers.. they wake up, check whether a set of conditions is true, and if it is they fire the next routine, which goes and pulls data or hands something to a model to analyze and then does something with the result. I'm not sitting there orchestrating it.
The two pieces that made this weekend work at all:
A boring Zapier subscription. There's no API for Otter (at least not on the Business plan), but there is a Zapier connection, so every time a transcript gets created it lands in a folder that syncs to the VM and gets consolidated with my other transcripts. Before that I was manually exporting and uploading every single one, which meant I mostly didn't. If a subscription saves me an hour a month it's already paid for itself, and I'd been stubborn about that for way too long.
Custom connectors instead of native ones. If you're using off-the-shelf ChatGPT or Claude you're limited to whatever integrations they've built for you. I wanted Cal.com links that could take a payment, so I got an API key, described what I wanted in plain English, and Claude Code built the plumbing. I didn’t have to ask "does this integrate with that" and just ask "can I get an API key."
The thing I’ve been thinking a lot about the past few days is how the shape of the inputs have changed. It used to be that you had to write a tight, direct prompt to get anything decent out of AI, and that's still true for a lot of it. But once there's a context layer (i.e., it knows me, the company, the tools, the history..) and it has real access to the things I use, then rambling at it for seven hours is a legitimate way to work.
What it doesn't do is decide though.. Jordyn and I still made every hard call ourselves this weekend, we just made them against something organized instead of against the noise in my head. Ask good questions, let AI structure the brain dump, then apply your own judgment to what comes back. You don't outsource the judgment.
Something else I noticed (and I don't think it's small).. rambling alone to AI is worse than rambling at a human who pokes a things and asks good questions while you braindump. Half of what's in those recordings only exists because Jordyn asked "wait, why?" or “What does that mean?” and I had to actually answer. Talking to yourself into a microphone doesn't produce the adversarial question that makes you go.. “huh, actually never mind, that's a bad idea” or “I didn’t think about that.”

4. Where I'm still bad at this
The unglamorous part.. my backlog is hundreds of items deep and I am the bottleneck. I built something that triages things but then it still all stacks up waiting on me to read and approve. I could let AI run on full autopilot, but it would confidently make/build a bunch of stuff I don’t want and break things I care about so I keep the human/judgement gate. That gate is me and I'm slow sometimes.
There's an agent that reviews my inbox every day and drafts the replies I should send. Every time I open those drafts they're not quite right, that's not how I talk, and I have never once sat down and told it so. Which is a 5-minute fix I've not been doing for months.
And the big one I want next is proactive, not reactive. Right now I still have to go ask. What I want is the thing that comes to me and says "you said these things, so I did these things, take a look and hit approve." I've got pieces of that, but nowhere near enough of it.

Putting this into action
If you're going away to reset or simply need to clear your mental browser tabs.. hit record while you're doing it. The talking is the highest-quality input you'll ever hand an AI, and it costs you nothing extra.
Go find the one manual step you keep skipping because it's annoying, and pay the twenty bucks to kill it. Mine was exporting transcripts by hand.
One more thing, and it's a new one
People ask to pick my brain constantly, and I've been doing it for free for years because I like helping. I'm going to start charging for it, partly because this weekend showed me what the thing I'm actually good at is worth.
You bring a problem, I poke at it and ask you the questions you're avoiding, and afterward you get a written brief.. everything you said, organized, plus the open questions and my actual recommendation on what you should do. You don't have to take notes and you don't have to remember any of it.
You're paying for my judgment and my system, and you leave with something you can hold.
If that's not for you, no hard feelings at all, and I hope this issue was useful on its own.
Wrapping up
I don't want to tie a bow on this, because the trip didn't necessarily fix anything. I came home and the grief came home with me, and it'll be here tomorrow too. What was different is that I got 2 days where I wasn't behind the whole time, and I came back to less work instead of more, and that's the first time that's happened in a long time.
The tat on my right arm says "do the next right thing." That's my whole approach to life in a single sentence. Some weeks the next right thing is a hard decision about the company, and some weeks it's just going to the mountains and letting a machine do the typing.
If you're carrying something heavy right now, you're not doing it alone.
Onward & upward 🤘
Drew





