You’re Not Behind (Yet): How to Learn AI in 19 Minutes...

Phase one establishes foundational habits and tool setups for consistent AI usage. This includes using AI tools like Claude, ChatGPT, Grok, or Gemini as direct replacements for Google searches. Keeping the chosen AI tool open in a pinned browser tab fosters habitual engagement. Utilizing voice input rather than typing significantly increases interaction speed and allows for more expansive queries, transforming the value derived from AI.

Further foundational steps involve downloading mobile AI apps to ensure constant access, enabling AI use as a thinking companion or search tool in various environments. The final foundation is automatically recording all online and in-person meetings (e.g., Zoom, Google Meets). Tools like Grain or Fathom (free) can transcribe these recordings, providing rich data for later AI analysis.

In phase two, AI transitions from a simple search tool to a strategic coach for improved thinking and performance. Instead of asking AI to perform tasks, individuals prompt it to help them think better about their existing work. An example involves an Instagram manager asking AI for high-leverage strategies, common mistakes, and critical questions to ask their manager for optimal goal achievement.

Leveraging transcribed meeting calls, AI can provide insights from conversations. For instance, an AI can analyze a manager-employee coaching session transcript to suggest a two-week skill-improvement curriculum. Business owners can use AI to interview them about business goals, uncovering key levers for planning and growth. The AI functions as a thought partner, asking probing questions that lead to deeper insights, but its advice should be critically evaluated, as it has extensive knowledge but lacks real-world context.

Phase three introduces using AI to actively perform tasks, emphasizing a structured approach to avoid generic outputs. The "10-80-10 rule" is crucial: perform the first 10% of the task (providing context and detailed instructions), have AI handle the middle 80% (drafting content or generating ideas), and then review the final 10% for quality assurance and refinement.

For example, instead of broadly asking AI for 50 Instagram content ideas, provide it with transcripts, top-performing competitor content, and your content strategy document. Instruct the AI to generate specific types of hooks, focusing on counterintuitive takes or pattern interrupts, and avoiding generic advice. This detailed context significantly improves AI's output. The human element of 'taste' is vital here; if AI output triggers an internal 'cringe', it indicates your quality bar is higher, prompting you to provide feedback and iteratively guide the AI, similar to training a junior team member. This iterative process allows users to generate numerous high-quality content ideas by continuously refining AI outputs.

Phase four focuses on building an AI system through prompt engineering and a personalized prompt library, ensuring continuous improvement rather than starting from scratch each time. This involves iteratively refining prompts, akin to perfecting a recipe. For instance, an initial prompt for Instagram hooks evolves by adding constraints like "use pattern interrupts," "avoid generic advice," "under 20 words," and "never use rhetorical questions," based on testing and feedback.

Each refinement creates a new version of the prompt (V1, V2, etc.), which can then be saved using tools like Text Expander for quick deployment via keyboard shortcuts. This process builds a comprehensive prompt library for various tasks (e.g., hook generation, LinkedIn post creation from transcripts, competitor analysis). Experimenting with different AI models (e.g., free vs. paid versions of ChatGPT or Claude) helps identify which models perform best for specific prompt types. This systematic approach ensures that AI output consistently meets increasingly higher quality standards.

Phase five involves integrating AI as an automated infrastructure, where systems are set up to run AI tasks autonomously in the background. This moves beyond manual interaction to automate repetitive workflows. Automation occurs at various levels: Level one uses built-in AI automation within existing tools (e.g., Fire Cut plugin in Premiere Pro for transcript generation). Level two employs simple automation tools like Zapier or Make.com to connect different applications, automating workflows such as transcribing Zoom calls, processing transcripts with ChatGPT, and sending results via Slack.

Level three utilizes more powerful automation platforms like N8N for granular control and sophisticated workflows, though these require greater technical knowledge. An example is automating weekly reports for student coaching calls, combining transcripts, Slack support conversations, and CRM data to summarize student wins, struggles, and future support needs. The most advanced level, level four, involves building custom internal AI applications for specific workplace needs, though for most purposes, mastering tools like Zapier is sufficient.