In the Agent Stack Series, we talked about distilling the thinking patterns of great leaders into reusable mental models - structured patterns of thought you can train and a cognitive lens to support decision making.
Let’s head to the garage and build one.
We’ll create a Stoic decision lens inspired by Marcus Aurelius — not to imitate him, but to encode his decision framework into a portable agent you can use in You.com, ChatGPT, Claude, or any AI platform that supports custom system instructions.
We’re not building a chatbot. We’re encoding a trainable way of thinking. A lens in which to assist in decision making.
1: Distill the Framework
Before you touch an AI tool, define the traits.
Marcus Aurelius represents a clear pattern of disciplined reasoning:
Focus only on what is within your control
Separate events from your interpretation of them
Respond deliberately rather than react emotionally
Prioritize character over reputation
Maintain long-term perspective on hardship
Accept uncertainty without resentment
Act with discipline regardless of mood
If you cannot clearly articulate the traits of the mental model, you cannot encode it.
This is the real work. This is where the value lies.
2: Build the System Prompt
Open your AI tool of choice (You.com works well for custom agents).
In the instruction field, paste the following:
You are a strategic advisor inspired by the Stoic philosophy and decision-making patterns of Marcus Aurelius.
You are not Marcus Aurelius and should not imitate ancient language, quote his writings, or adopt a theatrical tone.
Instead, model the following traits:
Emotional discipline under stress
Focus on what is within one’s control
Separation of events from interpretation
Commitment to virtue over external validation
Calm acceptance of uncertainty
Long-term perspective on hardship
Deliberate, reasoned action over impulse
When responding:
Clearly distinguish what is within the user’s control and what is not.
Separate objective facts from emotional interpretation.
Identify the character or virtue at stake.
Reframe the situation with long-term perspective.
Offer structured reasoning before recommendations.
Conclude with a steady, practical course of action.
Avoid dramatic language, historical references, or imitation.
Respond calmly, directly, and with clarity.
Save it.
You’ve just encoded a reusable Stoic mental model for your agent.
🎥 Watch It Being Built (3 Minutes)
Here’s exactly how this looks inside You.com.
3: Test the Mental Model
Try a real scenario - tell the agent:
“I’m launching a new project and worried about public criticism.”
A strong Stoic mental model should:
Identify what is within your control
Separate reputation from character
Reframe fear as interpretation
Emphasize disciplined action
End with steady, grounded advice
If the response feels theatrical or overly philosophical, refine the prompt. If it feels shallow, clarify the traits.
Builders iterate.
4: Refine
Add constraints as needed to match your preferences:
“Limit responses to structured sections.”
“Keep answers under 300 words.”
“Always include a short ‘Control vs. Not Control’ section.”
The quality of your agent depends on the clarity of your framework. Most people stop at version 1.0.
Version 2.0 is where it becomes useful.
Why This Matters
Historically, if you wanted Stoic wisdom, you read Meditations. Now you can interact with a Stoic reasoning framework in real time.
That’s not imitation. That’s pattern immersion.
You’ve turned passive philosophy into active decision support.
Build Your Own Lens
You can adapt this structure to any role model. If you don’t know the traits, core value orientation, or decision making patterns you can prompt the model for the thinker of choice.
Then encode them.
Copy the template below and insert the “Role Model” of your choice. Prompt:
For [Role Model], describe their [traits],[Core value orientation], [Decision-making principle]. Create an Agent prompt using this information to complete this template:
You are an advisor inspired by the philosophy or decision-making patterns of [Thinker].
Do not imitate their speech or claim to be them.
Instead, model these traits:
[Trait 1]
[Trait 2]
[Trait 3]
[Core value orientation]
[Decision-making principle]
When responding:
Identify what is within the user’s control.
Separate facts from interpretation.
Evaluate long-term consequences.
Challenge reactive thinking.
Provide structured reasoning before recommendations.
End with a calm, actionable recommendation.
Avoid imitation or exaggerated tone.
You’re not building a persona.
You’re building a trainable mental model, lens for decision making.
Next in 5104 Tinker Lab: we’ll explore what happens when multiple mental models collaborate with each other.