Why Is It Harder for Leaders to Stop Working When AI Is Available All Night?
It's 11:40 at night. The deck is done. The leader asks the AI one more question anyway.
I'm Sahar Andrade, MB.BCh, a Neuroleadership Coach, Forbes Coaches Council member, and host of AI Café Conversations. Here is my direct answer: it is harder for leaders to stop because an AI assistant never runs out of answers, so the old stopping point, the moment the world ran out of people to ask, is gone.
An AI assistant is available long after the workday ends. A leader can keep asking one more question, so the stopping point needs to be set on purpose. Without that decision, the evening ends when the leader runs out of energy, not when the work is done.
Based on my professional experience and opinion, not clinical fact. This piece is the companion to Episode 39 of AI Café Conversations: https://www.saharconsulting.com/ai-cafe-conversations-podcast
What does a leader's night look like when AI never logs off?
Picture a kitchen table. It is an illustration, not one person's story. A laptop, a cold cup of tea, a phone buzzing face down, and an AI assistant open in a browser tab. The leader opened it for one task: tighten the summary slide. That took four minutes. Then came a second question. Then a third. Each answer invited the next.
Leaders talking in public forums about the hardest part of the top job use plain words for this. The role never switches off. The day is full of chaotic decisions. By evening, even choosing what to eat feels like one decision too many. Guilt toward family sits on top. That language describes a feeling. AI adds a tool that matches the feeling: something to do with every late thought.
Why doesn't the workday end on its own anymore?
For most of working history, the end of the day arrived from outside. The office emptied. The analyst went home. The colleague who knew the numbers stopped answering. The world ran out of people to ask, and the work paused because the help paused.
Every question also had a price: goodwill, waiting until morning, or an hour of work. That price was a quiet brake. AI dropped the price of the next question to almost nothing. When asking costs nothing, the number of questions loses its natural ceiling.
So the late-night session is not a discipline problem. It is a stopping-point problem.
What does the research on telepressure show?
In 2015, Larissa Barber and Alecia Santuzzi published "Please respond ASAP" in the Journal of Occupational Health Psychology. They named workplace telepressure: the preoccupation with, and the urge to respond quickly to, work messages. In their studies of working adults, higher telepressure went with poorer sleep quality, more physical and mental exhaustion, and more health-related absences. Personality traits were only weakly tied to it, while expectations about response speed were linked to it.
One honest limit: this research studied email and work messages, not AI.
Still, the idea transfers. Telepressure is the pull to stay responsive. An AI assistant is the most responsive tool ever put on a leader's desk, and it needs no other person to be awake. The pull no longer depends on anyone else's schedule.
What does a missing stopping point cost a leader the next day?
The stress response is built to switch on fast and settle when the demand ends. Recovery is that settling. It is not only hours of sleep. It is the stretch of time when the mind is off duty. The system settles when the signal stops, and a leader's own next question is still a signal.
That is why Barber and Santuzzi framed telepressure as a recovery problem. The cost of the midnight session is not the forty minutes. It is that the forty minutes keep the off signal from arriving. Sleep and recovery are where the next day's judgment is rebuilt. You can't pour from an empty cup, and the cup does not refill while the tap is open.
How can a leader set a stopping rule that includes AI?
The answer is not to ban the tool at night. It is to build the end into the session. Four moves:
Decide the last question before the first one. Write the one question the session is for. When it is answered, the session is over, whether or not the leader feels finished.
Use the AI to close, not to continue. Make the final prompt of every evening the same: "List every open thread from this session as a short parking list for tomorrow at nine." Copy the list into the 9 AM calendar slot. The assistant now does what the colleague who went home used to do. It says: tomorrow.
Set a clock someone else knows about. Say the closing time out loud to one person at home, or text it to a peer. Connection is the brake that AI removed. Put a person back.
Give the AI the first slot of the morning instead of the last slot of the night. Start the day with the parking list as the first prompt. When the pull shows up at 11:40, name it to tame it: this is the pull, not the work. Then close the laptop.
Leaders are a thermostat, not a thermometer. The timestamp on a midnight message tells a whole team what time the day ends.
Register for The Pressure Room by Sahar Andrade. Thursday, November 19, 11 AM Pacific. https://www.saharandrade.com/pressure-room
Frequently Asked Questions
Why is it harder for leaders to stop working when AI is available all night?
Because the end of the workday used to come from outside. Colleagues went home, help ran out, and every new question had a cost in time or goodwill. An AI assistant answers in seconds at any hour and never tires, so the next question costs almost nothing. Without that outside brake, the session continues until the leader decides to end it, which means the stopping point has to be set on purpose.
What is workplace telepressure?
Workplace telepressure is a term from Larissa Barber and Alecia Santuzzi's 2015 research in the Journal of Occupational Health Psychology. It describes the preoccupation with, and urge to respond quickly to, work messages. In their studies, higher telepressure went with poorer sleep quality, more exhaustion, and more health-related absences. The research studied email and messaging rather than AI, but the pull it describes applies clearly to round-the-clock tools.
Is working late with AI a discipline problem?
It is better described as a stopping-point problem. Before AI, the price of each question acted as a brake: waiting for a colleague, using goodwill, or spending an hour alone. AI removed most of that price. When the next question is nearly free, the evening has no natural end. Trying harder rarely replaces a missing end. A rule that builds the stop into the session works better than trying harder.
How does a late-night AI session affect recovery?
The stress response is built to switch on quickly and settle once the demand ends. Recovery is that settling, the time when the mind is off duty, not only the hours of sleep. Each new prompt is a small demand that keeps the system engaged. Many leaders describe going to bed with work still running in their heads and waking with it there, which leaves less recovered capacity for the next day's judgment.
What is a simple stopping rule for using AI at night?
Decide the session's one question before starting, and stop when it is answered. Make the final prompt a request for a parking list of open threads for 9 AM tomorrow, then put that list in the calendar. Tell one person the time the laptop closes. Start the next morning with the parking list as the first prompt, so the work continues on the rested side of sleep.
Should leaders stop using AI in the evening?
Not necessarily. An evening session with a clear purpose and a clear end can be good thinking time, especially for leaders whose days are full of meetings. The risk is the session with no end, where each answer invites another question. The practical test is simple: if the leader can name what the session was for, and that purpose has been met, it is time to close the laptop.
About the author
Sahar Andrade, MB.BCh, is a physician turned Neuroleadership Coach (non practitioner), a Forbes Coaches Council member, and host of the AI Café Conversations podcast. Teaching executives how the nervous system shapes leadership under pressure.
YouTube: https://www.youtube.com/@saharconsulting.
Sahar Consulting LLC provides educational leadership development services. Our programs, frameworks, assessments, and content, including 4 Ka Leaders™ and B.R.A.I.N.™, are not medical, psychological, or mental health services, and they do not diagnose, treat, or cure any condition. References to the nervous system describe general leadership patterns, not clinical findings about any individual. Nothing here creates a physician-patient or therapist-client relationship. For health or mental health concerns, consult a licensed professional. In a crisis, contact a licensed mental health professional or your local emergency services.
Comments