The honest fact about AI use is that almost everyone overestimates their fluency. You reach for Claude or ChatGPT every day, you produce output that looks competent, and you assume the productivity is real. The audit below is the cheapest way to test whether that assumption holds, and to identify the specific next move that would actually move you up the stack.
This is a companion piece to The Three Layers of AI Fluency, which makes the case that real productivity with AI comes from Layer 2 patterns and Layer 3 workflows, not Layer 1 prompts. This piece operationalises that thesis. Twelve questions, scored honestly, then the cheapest move to make for each gap.
How to use this audit
Score yourself honestly. Each question is yes or no. No gradient. The point is not to feel good about your AI fluency. The point is to identify the gaps, pick one, and close it in the next month.
If you score 0 to 3, you are at Layer 1. Most operators are here, regardless of how much they use AI in any given session.
If you score 4 to 7, you are at Layer 2. The patterns are building. The next investment is in workflows.
If you score 8 to 12, you are at Layer 3. The systems are running. The work to do now is iteration.
If you score 11 or 12 on the first pass, you are probably grading on a curve. Run the audit again with a more sceptical second pass. Most operators score lower the second time.
The Layer 1 audit
1. Do you have a personal AI configuration (Claude Project, Custom GPT, Gemini Gem) that you actually use most days?
Yes means you have invested in persistent context once, and that investment compounds across every session.
No means every session starts from zero. You are paying the same setup cost every time you reach for AI. The next move: spend one Saturday morning setting up a single Claude Project or Custom GPT loaded with the context your most frequent work needs (your role, your team, your priorities, your standard output formats). Use it for one week. Iterate.
2. When you find a useful prompt, do you save it somewhere you actually reuse, with edits over time?
Yes means you have started a Layer 2 pattern library. The patterns will get sharper.
No means every prompt is a single use transaction. The next move: open a Notion page, an Obsidian vault, or a markdown file called Prompt Library. Save the next three prompts you find genuinely useful. Reuse them deliberately over the next month. Edit each one at least twice based on what worked.
3. Can you describe at least one of your AI uses in outcome terms (the report that gets produced, the decision that gets accelerated, the hours saved per week) rather than tool terms (I used ChatGPT for that)?
Yes means you have stopped framing AI as a tool and started framing it as a system that produces specific work.
No means your framing is still about the product, not the work. The next move: the next time someone asks what you use AI for, answer with the outcome, not the brand. Practise the language. The framing shifts the thinking, and the thinking shifts the behaviour.
The Layer 2 audit
4. Do you have prompt patterns you have refined across multiple uses, not just collected?
Yes means the patterns are compounding. Each refinement makes the next use sharper.
No means you are still in collection mode, building a library you never edit. The next move: pick one prompt you use repeatedly. Rewrite it. Use the rewritten version five times. Rewrite it again. The refinement is the work, and almost nobody does it.
5. Have you edited any of your AI patterns or workflows in the last two weeks based on what worked and what did not?
Yes means you are iterating. The system is alive.
No means the system is stagnating. Whatever you set up six months ago is what you still have. The next move: this week, edit one pattern. Make it sharper based on the last three times you used it. Set a recurring calendar event to revisit it monthly.
6. Can you articulate why one of your patterns is structured the way it is (which variables it asks for, what order, what tone)?
Yes means the pattern is yours, not a generic template you found online.
No means you are running someone else's pattern without understanding the choices that built it. The next move: pick the pattern you use most. Write a one paragraph note explaining why each part of it is the way it is. If you cannot, that pattern is the next one to rewrite.
7. Do you use more than one model in your week, with a clear sense of which one you reach for in which kind of moment?
Yes means you have hit the limits of any single model and are using the right tool for each job.
No means you are using one model for everything, which means you have not yet pushed any model far enough to feel its weaknesses. The next move: pick a kind of work where your default model has been just okay. Try a different model for the same work for two weeks. Note which one wins, and for what.
8. When a new frontier model launches, do you have a process for evaluating whether to update your patterns?
Yes means you are model agnostic in practice, not just in branding.
No means you are exposed to model lock in, where switching costs feel large because you have not designed for portability. The next move: pick three of your patterns. Try them in a different model. Note what changes. Decide whether the differences are worth the switch. The skill is the evaluation, not the conclusion.
The Layer 3 audit
9. Do at least three of your AI uses each week run on a trigger (calendar, event, automation) rather than on you remembering to start a session?
Yes means at least part of your AI use is running as a system rather than a habit.
No means every AI use depends on you remembering to use it, which means a busy week erases all of them. The next move: pick the AI use that would matter most if it ran automatically every Sunday or every Monday. Set up the trigger this week, even if the trigger is just a calendar event with the prompt pasted in the description.
10. Does the output of at least one AI workflow land automatically in a document, calendar event, or destination you actually use, without you copy pasting?
Yes means the workflow is integrated into your work, not orbiting around it.
No means the outputs live in chat windows you may or may not return to. The next move: pick one output that matters. Connect it to a destination you cannot avoid (your inbox, your Notion homepage, a recurring document). Use Zapier, Make, n8n, or the native integrations if available. The destination matters more than the model.
11. Has any of your AI work in the last quarter compounded, in the sense that the work is sharper, faster, or cheaper than it was three months ago, without you actively maintaining it?
Yes means you have at least one Layer 3 workflow earning its keep.
No means whatever AI productivity you have is flat. It might be high, but it is not growing. The next move: pick one workflow. Spend a Saturday morning making it 25 percent better. Then leave it alone for six weeks. Re evaluate. That experiment, repeated three times, is what compounding feels like.
12. If you stopped using AI tomorrow, how much of your work would degrade noticeably within a week?
Yes, significantly means AI is genuinely integrated into the work, not running parallel to it.
No, barely means you are still using AI as a sometimes assistant rather than a system. AI that does not change the actual work is not yet compounding for you. The next move: identify the one piece of work where AI could most credibly become irreplaceable. Invest in making that specific piece run through AI for the next month. Not everywhere. One specific piece, all the way.
What to do with your score
The audit is not a grade. It is a map. If you scored Layer 1, you do not need to jump to Layer 3 immediately. You need to make one Layer 2 investment that compounds, then a second, then a third. The same logic holds at Layer 2 transitioning to Layer 3. The work is sequential, and the sequence is not negotiable.
The single move that does the most work, regardless of your score, is to pick the AI use you do most often this month and invest in it specifically. One Claude Project loaded with your context. One Custom GPT configured for one kind of work. One Notion template that lives next to the work it produces. One focused investment beats five scattered experiments, every time.
The honest reframe of the audit, which is what makes it uncomfortable, is that your score is not about how smart you are or how much you know about AI. It is about how much deliberate work you have put into making AI a system in your week, versus how much you have left it as a sequence of clever sessions you remember to start. The difference between the two is the entire game.
Where to go next
For the strategic frame behind this audit, read The Three Layers of AI Fluency.
For Layer 2 pattern libraries to start from, the PromptLeadz Free Vault frameworks are the fastest path in. HIRED for the job search arc. LAUNCH for the first 90 days of any new role. SHAPE for the work of being a manager. POWER for office politics. HARDER for the hard conversations. MONEY for the negotiations. CRITIC for adversarial thinking. Each framework is fifty Layer 2 patterns that adapt to your context and survive the move from chat window to Project to workflow.
For the Layer 3 workflow setups, the Pro Packs include the Claude Project and Custom GPT configurations that turn each framework into a running workflow rather than a library of patterns. The Pro Packs are on the PromptLeadz Pro Collection at $29 each.
The audit is a snapshot. Run it again in 90 days. The score should be at least two points higher. If it is not, the work to do is not more AI use. It is more deliberate AI use, on a narrower set of things, for longer.
PromptLeadz publishes battle tested AI prompt packs for founders, product, sales, marketing, operations, HR, finance, customer success, adversarial thinking, hard conversations, new role launches, job searches, money conversations, office politics, and managers. All prompts are LLM agnostic. Pricing is in USD.
Deixe um comentário: