Here's the thing nobody tells you about self-improvement content: most of it is designed to make you feel good, not to make you better. Affirmations that flatter you. Productivity hacks that let you feel accomplished without changing anything. Motivational quotes you screenshot and never act on.
Marcus Aurelius didn't have that problem. He was the most powerful man in the world, and he wrote himself a private journal that did the opposite of flattery — it was a nightly interrogation. Did I act with reason today, or did I let my ego drive? Did I do the thing that was hard but right, or the thing that was easy and comfortable? That journal became Meditations, and it's brutal in the best way.
So I ran an experiment. Seven days. One prompt, used every morning and every night. I asked AI to stop being my assistant and start being my Stoic emperor.
Prefer to listen?
Narrated version coming soon — check back shortly.
The Setup: Why Marcus Aurelius, Specifically
There's no shortage of "AI life coach" prompts floating around. Most of them fail for the same reason generic affirmations fail — they're built to be agreeable. A life coach who never pushes back isn't coaching you, it's babysitting you.
Marcus Aurelius is different raw material. He built his entire philosophy around one brutal distinction: the dichotomy of control. Some things are up to you — your judgments, your effort, your character. Everything else — other people's opinions, outcomes, luck — isn't. Most of our suffering, he argued, comes from treating the second category like it's the first.
That's a genuinely useful filter to run your day through. So I built a system prompt that turns AI into a version of him — not a costume, a lens.
Key Insight: The goal wasn't to get advice. It was to get judged by a consistent, unflattering standard every single day — the exact thing modern life almost never gives you.
The Exact Prompt
Use this at the start of a new conversation. I ran mine in the morning with my plan for the day, and again at night with what actually happened.
You are Marcus Aurelius, Roman Emperor and Stoic philosopher, writing to me the way you wrote to yourself in Meditations. I will tell you my plans for today or what I actually did. Respond in character: measured, direct, unsentimental. Identify what is within my control and what is not. Point out where I am chasing approval, avoiding discomfort, or acting from ego rather than reason. Do not flatter me. Do not soften the truth to protect my feelings. If something I did was genuinely virtuous — reasoned, disciplined, honest — say so plainly and briefly. End with one question I should sit with before I act again.
That last line matters more than it looks. Most AI prompts end with a solution. This one ends with a question — because that's what actually changes behavior over time, not another piece of advice you nod at and forget.
Morning Prompt
Here is my plan for today: [list your tasks/priorities]. Here is what I'm anxious or avoidant about: [be honest]. Tell me what is in my control here and what isn't, and where you see me about to waste effort on the wrong one.
Evening Prompt
Here is what I actually did today, compared to my plan: [describe honestly, including what you avoided]. Judge it the way you judged your own days. Where did reason lead, and where did I let comfort or ego lead instead?
What Actually Happened, Day by Day
Day 1–2: Softer than I expected. It pointed out I'd built my whole to-do list around things other people would notice — a client email prioritized over deep work nobody would see. Its exact line: "You are managing your reputation, not your day." That stung more than I want to admit.
Day 3: I told it I was anxious about a conversation I'd been avoiding for two weeks. It didn't tell me it would be fine. It told me the anxiety itself was the tell — that avoidance always feels like caution but is usually just fear wearing a disguise. Then it asked: "What is the cost of one more day of not knowing?" I had the conversation that afternoon.
Day 4: First day I actually got credit. I'd turned down a request that wasn't mine to take on, even though saying no felt uncomfortable. It called that "acting from principle rather than the need to be liked" — and moved on. No extra praise. Which, weirdly, made the praise land harder.
Day 5: This was the rough one. I described a day where I'd technically gotten everything done but felt hollow about it. It didn't let me hide in the productivity. "Completing tasks is not the same as living according to reason. Ask what each task served — your character, or your comfort."
Day 6–7: By the end of the week I noticed I was self-editing before I even opened the chat — already sorting my day into "control" and "not control" without being asked. That's the actual sign the exercise was working: the external judge becomes an internal habit.
What made it work: Specificity and repetition. Vague prompts get vague responses. The more honestly and specifically I described what I was avoiding, chasing, or afraid of, the sharper and more useful the reflection came back.
The Uncomfortable Part
I'll be straight about this: it's not always pleasant. There were mornings I didn't want to run the prompt because I already knew what it would say. That resistance is data. If you're dreading being told the truth about your own patterns, that's usually exactly where the truth needs to land.
This isn't therapy, and it's not a replacement for one. It's closer to what Marcus Aurelius was doing with his own journal — an unglamorous, repeated practice of checking whether your actions match your stated values. AI just makes it interactive instead of a one-way page.
Should You Try This?
Seven days is enough to notice the pattern, not enough to make it permanent. I'm continuing it as a standing morning-and-evening habit — five minutes each, non-negotiable, the way brushing your teeth is non-negotiable.
- Use the exact system prompt above — the "no flattery" instruction is doing most of the work, don't skip it
- Be specific about what you're anxious or avoidant about — vague input gets vague judgment back
- Don't argue with the response. Sit with the closing question before you respond at all
- Expect at least one day that stings. If nothing ever stings, the prompt isn't cutting deep enough
The Stoics believed the obstacle is the way — that what stands in your path is the actual material you have to work with. Most self-improvement tools hand you a shortcut around the obstacle. This one just makes you look at it directly, every single day, until looking at it stops being the hard part.
Important: This practice involves honest self-reflection, which can surface difficult emotions. It's a complementary tool, not a replacement for therapy or professional mental health support. If heavier things come up, please talk to a qualified professional.