3 Simple Tips to Unlock Claude's GENIUS Mode 🤯
The video presents three techniques for getting more out of Claude AI: prompting it to ask clarifying questions, using it as an adversarial sparring partner to challenge assumptions, and converting successful conversations into reusable skills. The creator argues these shifts move users from passive question-answering to active collaborative thinking.
Summary
The video opens with the creator claiming that most people use Claude like a search engine — typing a question and accepting the first answer — which leaves the majority of the model's capability unused. To demonstrate a better approach, the creator walks through a live example of researching and validating a business idea: an app that scrapes public sources to surface local small business leads.
Tip one involves appending the phrase 'ask me clarifying questions until you're 95% confident you can complete the task successfully' to any prompt. In the demo, Claude responds by asking who the primary customer is, what the intended business model is, what differentiates the app from competitors like Apollo or ZoomInfo, and what format the output should take. The creator argues that this simple addition forces Claude to surface ambiguities and gather context before producing a response, resulting in a more targeted and personalized output rather than a generic answer.
Tip two addresses the tendency to accept AI output uncritically. The creator recommends following up any AI response with the prompt 'be my sparring partner, identify my blind spots, risks and assumptions.' In the demo, Claude pushes back on two core assumptions: first, that finding local businesses is the hard part (when in reality the bottleneck is converting them), and second, that large platforms like Apollo lack local data (when they have the resources to build it). The creator frames this adversarial mode as the mechanism that makes collaboration with AI genuinely intellectually valuable, rather than flattering but hollow.
Tip three focuses on efficiency and repeatability. After a productive conversation, the creator instructs Claude to 'create a reusable skill based on this chat.' Claude analyzes the session and generates two skill files — one for business idea validation with built-in clarifying questions, and one for the sparring partner framework — which are saved as a plugin. In a new conversation, the creator demonstrates that these skills can be triggered either explicitly with a slash command or implicitly by typing natural language like 'challenge my thinking,' with Claude intelligently selecting the relevant skill. The creator concludes that combining these three tips places users well ahead of the typical Claude user.
Key Insights
- The creator argues that appending 'ask me clarifying questions until you're 95% confident you can complete the task successfully' to a prompt causes Claude to surface vague or missing details before generating a response, producing more tailored output than a direct question would.
- When asked to act as a sparring partner, Claude challenged the assumption that finding local businesses is the hard part of the proposed app, arguing instead that the real bottleneck is knowing what to say to convert those businesses into paying customers.
- Claude also challenged the assumption that large competitors like Apollo lack localized business data, pointing out that well-funded platforms with large teams could replicate public-source scraping just as easily as a new entrant.
- The creator demonstrates that Claude can analyze a completed conversation and automatically generate structured skill files capturing the workflow — such as a validation research process and an adversarial sparring framework — which can then be saved and reused across future sessions without re-explaining preferences.
- When applying the sparring partner skill to a new business idea around Claude AI training, Claude identified that the biggest competitive threat is Anthropic itself, which gives away AI education for free, and suggested the winning positioning is owning an 'applied, live, done-with-you' training lane.
Topics
Full transcript available for MurmurCast members
Sign Up to Access