How each question will be answered

1 The data

How you brief AI

• Report summaryHow long your messages usually are, in characters median_prompt_length_chars
• Report summaryWhether you ask for a plan before it starts: often, sometimes, rarely or never asks_for_plan_first
• Report, Part 2How I brief you

How you check its work

• Report summaryRoughly how often you check its work against reality, for example “~20%” verification_rate
• Report, Part 2How I check your work

How you push back and keep going

• Report summaryRoughly how often you question or correct it pushback_rate
• Report summaryHow many back-and-forths a task usually takes typical_rounds_per_task

How you keep what you learn

• Report summarySaved instructions it follows: none, basic or extensive persistent_instructions
• Report summaryHow many saved rules came from fixing its mistakes rules_from_corrections

How often you use it

• Your answersHow often you use AI (question 4)

To rule out experience alone

• Your answersHow long you’ve used AI for work (question 4)

2 The analysis

Compare each habit between people with stronger and weaker results, among people with similar tools and similar jobs.

3 What would answer it

• If a habit is consistently more common among people with stronger results, across different tools and jobs, it’s a real candidate.
• If no habit stands out, results may depend more on the tools or the work itself, which is question 2.

1 The data

The person

• Report, Part 2The habits from question 1

The tools

• Your answersWhich AI tools you use (question 2)
• Your answersWhat your AI can do (question 2)
• Your answersWhether your employer provides your AI tool (question 2)
• Report summaryWhat your AI could see: your files and history, your history only, or just this chat access_level
• Report summaryWhether your AI can run code: yes or no env_can_run_code
• Report summaryWhether it can reach real files or data: yes or no env_can_access_real_data
• Report summaryWhether it remembers past chats: yes or no env_has_memory
• Report summaryWhether you use several AIs, or have one review another’s work: yes or no uses_multiple_agents_or_reviews

The work

• Your answersYour role (question 1)
• Your answersYour industry (question 1)
• Your answersOrganization size (question 1)
• Report summaryYour kind of work in a word or two, for example “operations” or “marketing” primary_work_domain
• Report summaryWhether you write software for a living: yes, no or unknown user_is_professional_developer

2 The analysis

Compare people with the same tools but different habits, and people with similar habits but different tools or jobs.

3 What would answer it

• If people with the same tools get very different results, habits matter more.
• If results rise and fall with what the tools can do, or with the kind of work, those matter more. Most likely it’s a mix; the aim is to estimate how much of each.

1 The data

What people build and share

• Report summaryWhether you’ve built anything reusable with AI: yes or no builds_tools
• Report summaryHow many tools, templates or automations you’ve built that get used more than once things_built_count
• Report summaryWhether other people reuse anything you’ve built: yes, no or unknown outputs_used_by_others

What helps

• Your answersTraining, guidance, time and help your workplace gives you (question 3)
• Your answersWhether your employer provides your AI tool (question 2)

What gets in the way

• Report summaryLimits your workplace puts on AI, like blocked tools or no access to real data: none seen, some, severe or unknown workplace_limits
• Report, Part 2What hurts quality or wastes time
• Report summaryWhether it can try things somewhere safe first: yes or no env_has_safe_test_environment

What would break if they stopped

• Report, Part 2Fragility (history version)

2 The analysis

Look for signs an organization could see from the outside, like tools people build and others reuse, and check whether they line up with stronger results. Then look at which workplace limits hold capable people back.

3 What would answer it

• If reusable tools and saved instructions reliably mark the people with the strongest results, organizations have something concrete to look for.
• If common workplace limits keep capable people at lower levels, that’s something organizations can change.