There is a specific dissonance that sets in when you read productivity advice as a new parent. The advice is not wrong, exactly. It’s just written for a different life: one with long, quiet mornings, a desk you can disappear into for four hours, and no one who will start crying the moment you enter flow state. You have a 10-month-old. You have a full-time job as a Lead AI Researcher. You have a list of home projects that grows faster than it shrinks. The four-hour block does not exist.

For a while I tried to find it anyway. Wake up earlier. Stay up later. Compress. What I discovered is that the problem isn’t the amount of time. It’s the assumption that useful work requires a long continuous strand of it. The field of AI research does not pause while I figure out my schedule. Models drop, papers publish, the stack evolves. Staying current is not optional; it’s the job. The question is not when do I get the big block of focus but how do I build something coherent out of small pieces.

The answer, for me, is engineering. Not motivation, not discipline, not any of the things productivity writing usually credits. A workflow designed for the actual constraints of my actual day. And I should say upfront that none of what follows works without my wife and my mother-in-law. The baseline capacity, the foundation under everything, is human, not logistical. When I have 45 minutes on a train, someone else is with the baby. When I get a focused evening, it’s because someone made that possible. Most productivity writing skips this part because it complicates the story. Acknowledging it isn’t just the polite thing to do; it’s the accurate model. Systems that pretend they run on individual willpower alone are lying about their inputs.

The train ride to the office is 45 minutes. I used to treat it as dead time: catch up on messages, scroll, decompress. At some point I realized I was wasting the most cognitively available window of my day. The shift was in reframing what the commute is for. I don’t use it for execution. I use it for arrangement. By the time I board, the day has already loaded context: overnight Teams/Outlook threads, a paper someone mentioned, a model evaluation result I want to dig into. The commute is where I triage that context, decide what actually matters, and set up the materials I’ll need. Outline the thing I want to write. Queue the paper I’ll read in fragments during the day. Map out the one problem I’m going to push on. Mentally, and sometimes literally, lay out the pieces so that when I next have fifteen free minutes, I don’t have to figure out where to start. I already know. The practical consequence is asymmetric: 45 minutes of arrangement unlocks every subsequent fragment of the day. I arrive at the office with what next already answered, and the day’s first task is the work itself, not the overhead of figuring out what the work is.

A 10-month-old generates 5-minute windows constantly. Watching the baby while she’s absorbed by something. Waiting for the kettle. Sitting on the couch at the end of the night when everyone else is asleep. These windows are not enough for deep work. They are more than enough for consumption, which is a significant portion of staying current in a fast-moving field. Research papers, technical breakdowns on YouTube, a new model’s release notes: none of these require four hours. They require attention and a pre-loaded queue. This is where the phone matters. Not as a distraction device but as a serious reading and note-taking tool. The queue is already arranged from the commute. I’m not deciding what to open; I’m opening the next thing in sequence. A 10-minute paper scan during a nap. A 5-minute model comparison during lunch. A voice note of a half-formed idea while the baby is playing. Individually, each fragment feels too small to count. Accumulated over a week, they add up to something substantial: more reading than I would have done in a hypothetical two-hour block I never actually carved out.

The piece that makes all of this sustainable rather than exhausting is that the learning and the job are not separate tracks. The papers I’m reading on the train are directly relevant to the problems I’m solving at work. The model I explored in 5-minute fragments is the one I’ll be evaluating in a proper meeting by Thursday. The note I dictated while watching the baby surfaces in a design doc three days later. The boundary between staying current and doing the job is largely artificial. This creates a flywheel: the job surfaces what I need to learn next; the learning makes me better at the job; the job gives me the grounding to know which learning is worth the time. Each rotation reinforces the others. Staying up to speed stops feeling like extracurricular obligation and starts feeling like the natural output of being engaged. The flywheel is also what lets me stop when I need to stop. When the 10-month-old needs me, I put the phone down without guilt. The system keeps spinning at low RPM without constant input. I can step away and step back in.

I used to believe that serious work required serious stretches of time, and that anything else was a compromise. What I’ve found is that the constraint, the baby, the commute, the fragmented schedule, forced me to build something more honest: a workflow designed for the life I actually have, not the one where I have four uninterrupted hours every morning. The ship gets built through the neck of the bottle, piece by piece. It still becomes a ship. If you’re in a similar season of life, I’d resist the impulse to wait for the conditions to improve before taking your learning seriously. Find your 45-minute catalyst. Curate the queue. Trust the flywheel.