The Five-Stage Operating System for Career Development (With AI)

Most growth plans fail the same way: they read like HR paperwork, not an operating system.

You write goals once, then real work shows up, priorities change, and the plan goes stale.

The shift now is practical. You can use ChatGPT as a structured thinking partner to pressure test your assumptions, identify skill gaps, and convert vague ambition into weekly execution.

What this method is, and what it is not

This method helps you think clearly and move faster.

It doesn’t replace your manager, your mentors, or your judgment.

My recommendation: use AI for structure and synthesis, then validate decisions with humans who see your work directly.

Tradeoff: you get speed and breadth, but you must actively manage quality. If you don’t challenge outputs, you'll get polished nonsense.

The setup that actually works

Before you ask for a plan, build a context packet you can paste into ChatGPT.

Use:

• Your current role scope and responsibilities
• Projects you led in the last 6 to 12 months
• Results you can defend with evidence
• Feedback themes you have heard more than once
• Skills you want to build for your next role
• Constraints (time, budget, location, personal bandwidth)

If you already have a personal README, use it.

I also recommend using the GitLab Individual Growth Plan guide as your scaffold so output lands in a format you can run.

For reflection prompts, "Managing Oneself" is still useful because it forces specificity.

The five-stage operating system

1) Calibrate the model to your context

Start with a clear role brief and ask for an interview, not advice.

Use this prompt:

Act as a career coach helping me build a 12-month Individual Growth Plan.
 
Process rules:
- Ask one question at a time.
- Do not give recommendations yet.
- Ask clarifying follow-ups when an answer is vague.
- Separate facts from assumptions.
 
Focus areas:
- strengths I repeatedly use
- weaknesses that constrain outcomes
- goals for 12 months and 3 years
- role constraints and risk tolerance
- what success looks like in measurable terms
 
After each answer:
1. Summarize what you heard in 1 to 2 lines.
2. Note one possible assumption (if any) with confidence (0.0 to 1.0).
3. Ask the next best question.
 
Every 5 questions, output a Markdown snapshot with:
## Current Profile
## Confirmed Evidence
## Open Questions
## Assumptions (with confidence)

Goal of this stage: produce a high-fidelity profile, not a motivational conversation.

2) Extract the pattern, then challenge it

Once the interview is done, force the model to separate facts from assumptions.

Use:

Based on our conversation, produce only Markdown in this exact structure:
 
## What Appears True
- claim
  - evidence from my answers
 
## What Is Still Uncertain
- unknown
- why it matters
- question to resolve it
 
## Assumptions (with confidence)
- assumption
  - confidence: 0.0 to 1.0
  - what would validate or invalidate it
 
## Top 3 Risks If Assumptions Are Wrong
1. risk
   - impact
   - early warning signal
   - mitigation
 
Rules:
- Use only evidence from my inputs.
- If evidence is missing, label it clearly as an assumption.
- Keep it concise and concrete.

This step is where quality usually jumps. You stop treating model output as truth and start treating it as a hypothesis set.

3) Build the plan in execution format

Now ask for a 12-month plan with quarterly milestones and weekly behaviors.

Use:

Turn this into a 12-month Individual Growth Plan.
 
Rules:
- Max 3 priorities.
- Use only evidence from my profile.
- If something is inferred, label it as ASSUMPTION with confidence.
- Make tradeoffs explicit.
 
Output only Markdown in this structure:
 
## Plan Summary
- role target
- time horizon
- success definition in one sentence
 
## Priority 1
### Why now
### Quarterly milestones
### Weekly execution behaviors
### Success metrics
### Dependencies and blockers
### Tradeoffs (what I will stop or deprioritize)
### Risks and mitigations
 
## Priority 2
(same structure)
 
## Priority 3
(same structure)
 
## 30-Day Action Plan
| Action | Owner | Deadline | Metric |
|---|---|---|---|
 
## Assumptions (with confidence)
- assumption
  - confidence: 0.0 to 1.0
  - validation step

Ask for a strict cap of three priorities. More than that usually means you haven’t made a tradeoff.

4) Run a real skills-gap analysis

Do not ask, "What should I learn?" Ask for role-relevant capability deltas.

Use:

Given my target role and current profile:
 
Output only Markdown in this structure:
 
## Capability Map
| Capability | Required Level (1-5) | Current Level (1-5) | Gap | Evidence |
|---|---|---|---|---|
 
## Top 5 Gaps by Impact
For each gap include:
### Gap: [name]
- Why it matters for the target role
- One on-the-job project
- One coaching or feedback mechanism
- One learning asset
- One metric to track progress
- One 30-day test
 
Rules:
- Prioritize role-relevant capabilities over generic skills.
- If current level is uncertain, mark ASSUMPTION with confidence.

This gives you development tied to work, not random content consumption.

5) Install a monthly review loop

A plan without cadence is a document.

A plan with cadence is a system.

Use this monthly check-in prompt:

Review this month against my growth plan.
 
Output only Markdown in this structure:
 
## Wins (with evidence)
 
## Misses and Root Causes
 
## Environment Changes
- what changed
- impact on plan
 
## Continue / Stop / Start (next month)
- Continue:
- Stop:
- Start:
 
## Updated Risk Register
| Risk | Likelihood | Impact | Mitigation | Owner |
|---|---|---|---|---|
 
## Next-Month Experiment
- Hypothesis
- Test design
- Success metric
- Review date
 
## Plan Updates
- what to change in priorities, milestones, or metrics

Keep each monthly review to one page. If it’s longer, it’s too abstract.

Failure modes to watch

These are the common ways this process breaks:

• You paste weak context, then expect strong guidance
• You accept flattering output without challenge
• You confuse activity with skill acquisition
• You set goals with no metric and no review date
• You optimize for inspiration instead of behavior change

If any of those are true, the plan is below the line and needs a reset.

My default stack

If you want a practical default, use this:

• Framework: GitLab IGP guide
• Reflection anchor: "Managing Oneself"
• Model role: interviewer first, advisor second
• Cadence: monthly review, quarterly reset
• Format: 3 priorities, explicit tradeoffs, measurable outcomes

You can add personality instruments if useful, but treat them as signal inputs, not identity definitions. I still agree with the core critique in Personality Tests Are the Astrology of the Office.

What to do this week

If you want to start now:

1. Build your context packet (45 minutes).
2. Run the calibration interview (60 minutes).
3. Produce your first 12-month draft with 3 priorities (30 minutes).
4. Convert it into a 4-week execution plan with metrics (30 minutes).
5. Book your first monthly review on your calendar before you close the doc (5 minutes).

Success metrics for month one:

• You can name your top 3 development priorities in one sentence each.
• Each priority has one observable weekly behavior.
• Each priority has one measurable outcome.
• You completed one review loop and updated the plan based on evidence.