Skill × Clarity AI Leverage Formula
Increase AI leverage by combining domain skill with precise instructions.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 97%
The Skill × Clarity formula treats AI leverage as the product of two human inputs. Skill is the domain knowledge required to recognize quality, diagnose mistakes, and stop the model from pursuing an unhelpful direction. Clarity is the ability to state the desired outcome and divide it into instructions the AI can follow. If either input is weak, leverage remains limited: expertise without clear communication is difficult for the system to apply, while precise prompts without judgment can efficiently produce the wrong result. Users improve outcomes by staying engaged, reviewing intermediate work, questioning assumptions, and iterating on both their understanding and their instructions. The model positions AI as a multiplier and sparring partner, not an autonomous replacement for original thought.
Origin
Extracted from Marketing Against the Grain
Core principles
- 01AI amplifies existing capability rather than replacing it.
- 02Domain skill helps users recognize and correct bad outputs.
- 03Clarity converts intent into instructions an AI can execute.
- 04Human judgment must remain inside the working process.
How to run it
- 1
Specify the result
State what a successful output must accomplish for a defined user or situation. Include constraints that materially affect quality.
Pro tip Describe success in observable terms rather than asking for something merely impressive or professional.
Watch out A vague target forces the AI to invent important product or content decisions.
- 2
Map the required expertise
Identify the knowledge needed to distinguish a good answer from a plausible but weak one. Fill critical knowledge gaps or involve a qualified reviewer.
Pro tip Write down the failure signals an expert would notice.
Watch out Do not automate decisions whose correctness you cannot assess.
- 3
Translate intent into steps
Break the work into ordered instructions with clear inputs, transformations, and outputs.
Pro tip Provide examples that demonstrate your actual standards and voice.
Watch out A single oversized instruction can conceal errors until the end.
- 4
Review and challenge
Inspect intermediate reasoning or artifacts, question weak choices, and redirect the AI before errors compound.
Pro tip Ask why a choice was made and compare alternatives when the path is uncertain.
Watch out Passive acceptance distances you from the process and reduces learning.
- 5
Diagnose the limiting factor
If results remain weak, determine whether greater domain skill, clearer communication, or both are needed. Improve that factor before adding more automation.
Pro tip Keep examples of successful and unsuccessful outputs to sharpen future judgment.
Watch out Changing tools will not necessarily fix weak skill or unclear intent.
In the wild
A founder asks an AI tool to implement a database-backed quiz. Technical skill helps her notice an unsuitable architecture, while clear step-by-step requirements help the tool correct the implementation without rebuilding everything.
→ The AI produces more useful work because the founder can both direct and evaluate it.
A marketer supplies proven examples, explains the audience and desired response, and asks the AI to draft variants. The marketer rejects generic options and explains why before requesting another round.
→ The AI becomes a useful multiplier of an established content skill rather than a source of undifferentiated posts.
Common mistakes
Expecting leverage without expertise
AI may produce fluent output, but an unskilled user may not recognize strategic, factual, or technical defects.
Giving an ambiguous destination
Even strong domain expertise cannot guide the system if the intended result and constraints remain unstated.
Using AI to avoid the process
Handing off the entire task removes the questioning, correction, and learning that generate better outcomes.
Is it for you?
Best for
It is best for knowledgeable practitioners using AI to accelerate coding, content, analysis, or other judgment-intensive work.
Not ideal for
It is not ideal as a substitute for qualified expertise in areas where the user cannot evaluate correctness or risk.
From the transcript
“the function I think about is like skill times clarity equals the leverage I get out of AI.”
“the people who really succeed use AI to go deeper into the process and learn what's actually happening, understand it, question it, push back, and…”
From the episode
The AI System That Built a 1.4M Audience