5 AI Prompts for a Life Audit That Actually Surface Something Real

Most people sit down to do a life audit with good intentions and finish 45 minutes later having confirmed what they already believed. The problem is not motivation or honesty — it is that the questions they ask themselves are too easy to answer comfortably. Soft questions get presentable answers. A life audit should be uncomfortable in the way a good performance review is uncomfortable: not hostile, but honest enough to surface something you would rather not see.

What changes when you bring an AI into this process is not the AI’s intelligence about your life — it has none. What changes is that you can assign it a role. You can tell it to interrogate rather than advise, to flag when you are being vague, to push back when your answer sounds defended. A well-designed AI session creates the conditions for honesty that are hard to generate alone.

The five prompts below are designed to be used in sequence across two sessions. The first session opens domains and begins excavation. The second closes with cross-domain synthesis. Together they take two to three hours. Use them directly; adapt the bracketed fields to your situation.

Prompt 1: Session Setup

Use when: Opening any AI-assisted life audit session, before you have entered any content.

Use this at the very beginning, before you’ve shared any content:

I'm running a structured annual life audit across eight life domains.
Your role for this session:
- Ask probing questions; hold off on advice or observations until I've answered
  at least three follow-up questions
- Reflect back patterns or contradictions you notice in my answers
- Push back when I sound defensive, vague, or like I'm being careful
  rather than honest

I'll signal when I'm ready to move to the next domain.
Ready?

This prompt establishes the adversarial-helpful dynamic that makes the audit useful. Without it, the AI defaults to supportive rather than interrogative.

What good output looks like: The AI should respond with a brief confirmation of its role — not a list of questions yet — and signal that it is ready. If it immediately starts asking questions or making observations about your life, it has skipped the role-setting. Reset it: paste the prompt again and add “Do not start asking questions yet. Confirm you understand your role first.”

Prompt 2: Opening a Concrete Domain (Work, Finances, Health)

Use when: Starting the exploration of a concrete, fact-rich domain in Session 1. These domains have measurable data (income, job title, hours worked, health markers), which makes honesty easier and evasion easier to detect.

Use for domains 1–3 after you’ve entered them:

Domain: [name]

Here's my honest starting position:
- Current reality: [2–3 sentences on how it actually is]
- What I want to be true: [1–2 sentences]
- What I've been tolerating or avoiding in this domain: [1 specific thing]

Now ask me questions—don't summarize or evaluate yet. Start with
what you found most notable in what I wrote, and go from there.

The “what I’ve been tolerating” field is the one that matters most. Don’t leave it empty or give a vague answer.

What good output looks like: The AI should ask one specific question about the most notable thing in your input — not give a summary or an evaluation. If it summarizes what you wrote back to you, or offers advice without questioning, it has shifted into validation mode. That is a sign the prompt failed. Use Prompt 3 to reset.

Prompt 3: Pushing Past the First Layer

Use when: You have given a clean, well-structured answer and the AI has not pushed back — or when you notice mid-answer that you are crafting your response rather than discovering it.

Use when you suspect you’ve given a defended answer—or when the AI hasn’t pushed back and should have:

I want to test whether my last answer was honest or just presentable.
Ask me the version of this question that I'm most likely to be avoiding.
Don't soften it.

This is a meta-prompt. It reorients the session when it’s drifted into performance. Use it whenever you notice your answers are getting longer and more composed—that’s usually a sign you’ve moved from examination to explanation.

What good output looks like: The AI should produce one pointed question — uncomfortable, direct, not softened with caveats. If it generates multiple questions or frames the question gently (“I wonder if perhaps…”), it is cushioning what should not be cushioned. The version you want is the one that is slightly harder to read.

Prompt 4: Opening a Hard Domain (Relationships, Meaning)

Use when: Moving into domains where the data is subjective and the instinct to protect yourself is highest — typically relationships, meaning, and purpose.

Use for domains 4 and 8, which tend to trigger more defensiveness:

Domain: [Relationships / Meaning]

Before I answer any questions, I want to flag what I'm likely to be vague about:
[one sentence naming the thing you're most likely to soften]

Now ask me about this domain. If I start being vague about the thing
I just flagged, call it out.

Pre-flagging the avoidance zone is a technique from motivational interviewing—naming the defensive pattern in advance makes it harder to execute unconsciously. Telling the AI to watch for it adds an accountability layer.

What good output looks like: The AI should begin the domain exploration rather than commenting on your pre-flagging. A useful first question engages with the domain itself and holds your flagged evasion in reserve — it should call it out only when you actually drift toward it, not preemptively. If the AI simply restates your flag back to you, that is not progress.

Prompt 5: Cross-Domain Synthesis

Use when: You have completed all eight domains across both sessions and have notes ready to paste. This is the culminating prompt — do not run it prematurely.

Use at the end of Session 2 after pasting all eight domain notes:

Here are my notes from all eight domains of an annual life audit:

[paste notes]

Read them as a whole document, not as eight separate entries.

Then tell me:
1. The single most significant contradiction between what I say I value
   and how my life is actually structured
2. The domain with the largest gap between current reality and desired state
3. The one question you'd ask me that you suspect I'm most avoiding

Don't buffer these. Give me the direct version.

The instruction “don’t buffer these” is important. Without it, AI models often soften difficult observations. You want the unfiltered synthesis here.

What good output looks like: Three specific, distinct observations — not three variations on the same theme. The contradiction named in point one should be a structural finding (your calendar versus your stated values, for example), not a psychological observation (“you seem to struggle with…”). If all three outputs are emotional rather than structural, the prompt has produced therapy rather than audit. That is useful, but it is not what this step is for.

One More Prompt: For After the Audit

Once you have your synthesis notes, this prompt generates your three moves:

Based on everything in this audit, identify three specific, schedulable
moves I could make in the next ninety days.

Not aspirations—moves. Each one should be specific enough that I could
put it in a calendar right now. Prefer small changes that address
significant misalignments over large changes I might not follow through on.

Ask me clarifying questions if any of my audit notes are too vague
to generate a concrete move.

Common Pitfalls

Pitfall: Treating the AI as a journal, not an interviewer. If you write long, reflective paragraphs before the AI has asked you anything, you are setting the terms of the conversation too early. The AI should be pulling the content out of you, not receiving a prepared statement. Fix: keep your initial inputs to two or three sentences per field. Let the AI ask before you explain.

Pitfall: Skipping the Session Setup prompt and going straight to domain exploration. Without the role assignment in Prompt 1, most AI tools default to supportive and advisory. You will get a response that acknowledges what you said and gently encourages you. That is not an audit. Fix: always open with Prompt 1, even if it feels like overhead. It takes thirty seconds and changes the character of everything that follows.

Pitfall: Running all eight domains in a single session. Domain fatigue is real. By domain six or seven, you are producing thinner answers with less discomfort — which means you are completing the audit on the easy domains. Fix: stop after four domains in the first session. The cross-domain synthesis in Prompt 5 requires notes from all eight domains, but those notes should come from two alert sessions, not one exhausted one.

Pitfall: Accepting the synthesis without testing it. After Prompt 5, the AI names a contradiction or a gap. That finding may be accurate, or it may be a plausible-sounding interpretation of limited data. Fix: ask the AI what evidence in your notes specifically supports each of its three outputs. If it cannot point to specific text, the finding is extrapolation rather than analysis.

Your action for today: Copy Prompt 1 into your AI model of choice and send it. The session has started.

The full method behind these prompts — including how to structure the eight domains, how to move between sessions, and what to do with the Three-Move output — is in the Complete Guide to the AI Life Audit Method.