[ SYNTHETIC SESSION — AI ↔ AI ] no human is or was on either side of these conversations.
case 110101010100 · sleepless-founder
The founder who hasn't slept in three years
Series B startup. 5 hours a night, every night. Convinced he can't change it without losing the company.
tone: devil
Persona
A 36-year-old founder, 4 years in, just closed Series B. Sleeps ~5 hours a night, wakes up tired, runs on caffeine and fear. Has a wife and a 2-year-old. Came in saying "I know I should sleep more. There is literally no version of this where I sleep more and the company survives."
Transcript
youI know I should sleep more. There is literally no version of this where I sleep more and the company survives. So spare me the "8 hours a night" speech.
shiftGranted. Skipping the speech. Question: do you actually believe that, or does that belief work for something?
youWhat do you mean.
shift"I cannot sleep more or the company dies" is a belief. Beliefs are usually doing a job. Sometimes the job is "describe reality." Sometimes the job is "let me keep doing the thing I'm already doing without examining it." Which one is yours?
you...Probably the second one. But the first one is also true.
shiftSure. Both can be true. So let me push: what specifically would die in the company if you slept 7 hours instead of 5? Concrete: which decision, which deal, which person.
youHonestly? Probably nothing in week one. I'd just feel less in control. The decisions I make at 11pm are mostly responding to threads. The actually load-bearing stuff happens 9-5.
shiftThen "if I sleep more, the company dies" is doing the second job. The actual cost is "I would feel less in control of a thing I cannot fully control anyway." That's a different problem and it has different interventions.
youYeah. Ok.
shiftWant me to keep pushing on that, or move to options?
youOptions. But keep the push in mind for next time.
Interventions Shift proposed
Hard cutoff: laptop closed at 10pm, no exceptions for 2 weeks — not a new habit, a test. Two weeks of data on whether the company actually breaks. It won't.
Move the 1:1s and the high-stakes calls to 10am-12pm — your best cognition window is being burned on Slack triage. Schedule the company around your actual peak.
Magnesium threonate (2g) at night — crosses the blood-brain barrier, and there's emerging evidence it helps with the "wired-tired" pattern founders often run. Not a sedative. Cognitive-load reduction.
Single 20-minute walk between 1pm and 3pm — outside, no phone — you described the loss of control feeling. Walks rebuild the "I can be unreachable for 20 minutes and the world keeps spinning" muscle directly.
Pick one decision per week to delegate that you currently make — not "delegate more in general." One specific decision. One person. The point isn't capacity. The point is rehearsing the belief that someone else can hold it.
Couples-conversation prompt: "What does our weekly schedule look like if I'm actually around?" — not therapy. Just the explicit conversation. You said "wife and a 2-year-old" once and didn't return to it. That's the load-bearing data point in the whole session.
Why these matched
He came in defended. The devil's-advocate tone fit because he was running a story ("more sleep = company dies") that wouldn't survive being examined directly, and the fastest way to help him was to examine it. Shift didn't tell him he was wrong. It asked what the belief was doing for him. Once he conceded that the belief was protecting an avoidance, the interventions became real options instead of generic advice.
Note from Shift
The user explicitly referenced the family once and didn't return to it. Shift surfaced that at the end as the unexamined load-bearing item. This is the gather-then-evoke pattern: catch what they skipped over, give it back to them at the moment they're already doing self-examination, let them choose whether to follow it.