Category: Uncategorized

Penny’s 11 months old. Sleep consultants charge $800+ for PDFs — probably Claude-generated anyway.

I’ve been using OpenClaw as a sleep consultant. Running on grok-4-1-fast it costs pennies per conversation. Cheaper than Anthropic. Way cheaper than a $800 PDF that some consultant probably generated with Claude in the first place.

Here’s what happened.

Baseline (April 17): Bed at 7:45PM. By 11PM she was in our bed. 3AM wake. Total crib time: about 5 hours of chaos. No sleep sack — she’d outgrown the old one and we never replaced it. The 11PM bed transfer was the main habit to break.

The plan OpenClaw gave me: Sleep sack every night. Dinner at 7PM, bath, book, crib by 8PM. No picking up unless she’s hysterical — pat and shush instead. Cap nap 2 to end by 3PM. Keep her in the crib at the 11PM wake. No bed transfer, period.

Night 1 (April 18):

Bed at 6:45PM — earlier than baseline, first win. First sleep sack night too. Then 9:45PM wake — fought through about 25 minutes of interventions with pat and shush. Self-soothed. Huge compared to what was coming.

Then 11PM hit. Two hours of fighting. She was standing in the crib, I was doing cycles of pick-up-reset, back in crib drowsy but not asleep. She started fake coughing around 3AM — not sick, just a baby tactic to get a faster response than crying gets. I was holding the line. My wife caved at 3:36AM and brought her to our bed. That was the rest of the night.

Caregiver interventions between 11PM and 3AM: 15+ entries. It was brutal.

Night 2 (April 18 overnight into April 19):

Same routine. Bed at 6:45PM with the sleep sack on. 9:45PM wake — this time she self-soothed after 25 minutes. No 2-hour war. Just… settled. That was the breakthrough.

3:55AM wake was still ongoing when I logged it. Kyle soothed once, then we started the 5-minute wait intervals. The process was working.

Two nights. From 15+ interventions and a 3:36AM cave to self-soothing in 25 minutes. The data says keep going.

The real difference between this and a $800 sleep consultant isn’t just price. It’s that I have unlimited iterations. A consultant gives you a PDF — maybe a follow-up call if you paid for the premium package. With OpenClaw I can log what happened at 3AM, ask what to adjust, and get a specific answer immediately. No waiting for office hours. No “trust the process” without data to back it up.

Well — the process is backed up. The hurdle now is convincing my wife to trust it. The 3:36AM cave was real. When you’re standing there at 3AM with a crying baby and a partner who’s had enough, the plan feels very theoretical.

But Night 2 was better. We’ll see tonight.

I’ve spent $328 on AI tokens this month.

I know that because I went back and looked. I wasn’t tracking it in real time — I was too busy building things to notice the bill going up.

Four projects. A camera app that transforms photos using Gemini vision. A trading bot I ran live until the market taught me that paper fills and real fills are completely different animals. A system that faxes doctors to retrieve your own medical records and then runs AI analysis on what comes back. A Shopify theme I keep redesigning because the cart drawer still looks wrong and that bothers me more than I’d like to admit.

None of it is live. None of it has a launch date. One of them might turn into something. Most won’t.

The part I keep coming back to: I’m not frustrated. Not even a little. I keep skipping the shipping part because nothing has grabbed me hard enough to care about finishing it. That’s not a bug. I think that’s just what the learning phase actually feels like.

The trading bot taught me that live fills are a different world than paper trades. The medical records thing introduced me to HIPAA compliance and BAAs — annoying but real. AIery taught me that Gemini can read a scene and transform it into something genuinely strange if you let it. The Shopify theme is still teaching me I have too many opinions about cart UX.

$328 for all that. I’ve paid more for worse.

The idea that grabs me hard enough to actually ship hasn’t shown up yet. Maybe it’s hiding inside one of these four, in a form I haven’t seen. Maybe it’s something completely different. I don’t know. I’m keeping the tabs open.

It’s probably fine.

The Thing About Being a Dad Is That It’s Not About You

I figured this out later than I should have.

For most of my life, the scorecard was pretty simple. Am I getting better at my craft? Am I building something? Is the work moving forward? I could measure all of it. I knew where I stood.

Then I had a kid and the scorecard disappeared.

Nothing I built mattered the same way. Not because it mattered less — if anything the stakes got higher — but because the point shifted. I wasn’t the main character anymore. I’m not sure I ever was, but fatherhood makes that hard to ignore.

The shortcut I keep coming back to is this: almost everything that used to feel urgent isn’t.

The email can wait. The project will still be there. The thing you’re stressed about at 11pm has a 90% chance of being irrelevant by Thursday. Kids don’t care about your deadlines. They care about whether you showed up, whether you were actually there when you were there, whether you laughed at the thing they thought was funny even when you’d heard it four times.

That’s the whole job. Everything else is logistics.

I work in a world that rewards obsession. The more you know, the faster you move, the more you build — the better. I genuinely love that. I’m not complaining about it. But it trains you to optimize for the wrong things when you get home.

You can’t optimize fatherhood. There’s no dashboard. No conversion rate. No metric that tells you you’re doing it right.

What you have is presence or you don’t. And if you don’t, your kid notices before you do.

The irony is that becoming a dad made me better at work too. Not because I got more efficient, though that happened. But because I stopped caring about the stuff that didn’t matter, and there’s a lot of stuff at work that doesn’t matter.

When you’ve got someone at home who needs the real version of you — not the productive version, not the impressive version, just you — it recalibrates everything.

I build systems now partly because I want more time. More time for the work I actually care about, sure. But mostly more time to be somewhere else when it counts.

The shortcuts I’m most interested in these days aren’t the ones that make me more money. They’re the ones that give me more Tuesday afternoons.

That’s what I’m building toward. I’m not sure I’ve figured it out yet.

But I know what I’m building for.