How I Use AI (1 Hour Masterclass) | Kata 3

Varun Mayya shows how to use AI as a literal genie by supplying precise context, examples, and specifications so both machines and teams execute correctly.

The Genie Problem

• Both AI and employees fill vague instructions with their own assumptions, producing outcomes right for them but wrong for you (3:03-10:02).
• Writing is active thinking; unclear writing signals unclear thinking and raises coordination costs.

Context Through Examples

• Provide concrete references such as file-naming conventions or golden-brown pancakes so the recipient matches the exact intent (10:04-20:06).
• Experienced hires act as context-repair technicians; shared history lets teams decide on each other's behalf.

AI Research With Human Validation

• AI generates thorough cost models and edge cases; humans run the AI test by calling experts and checking real prices to eliminate hallucinations (20:08-30:09).
• Apply the same loop to business planning, marketing budgets, and hiring specs.

Leadership Clarity & Delegation

• Leaders who report performance poorly create misaligned incentives and lose top performers (30:10-44:24).
• Use AI to draft emails, documents, and feedback in your voice, freeing time for high-leverage decisions while protecting culture.

00:00:00 Introduction

Speaker: Varun Mayya

• Varun frames a future where AI and humans are deeply meshed. He argues that today AI (like GPT-4, Gemini) is a superior idea generator in many cases, but humans remain essential for thinking, judgment, and execution.
• The core claim: leadership clarity and context are decisive for outcomes in teams and organizations; poor clarity can derail careers and coordination costs can cripple execution.
• He teases a concrete framework (the Lightning Way) for scaling clarity, delegation, and execution with AI as a partner rather than a replacement.
• The talk emphasizes that both AI systems and human teams suffer when instructions are vague, and that clarity, examples, references, and precise specifications reduce coordination costs and improve decision-making at scale.

"You cannot have a world where it's just AI or human. I think we're going to live in a very meshed world." (Varun)

"The communication between you and your teams only 30% is in your employees heads. Your clarity as a leader or lack of it can make or break careers." (Varun)

• The intro also sets up a pivot: use AI for research, planning, hiring, incentive design, reporting, and creative direction, but always couple AI with human verification to avoid hallucinations.

Notable idea: AI shines at generating options and first drafts; humans provide the context, values, and final judgment that align with the real-world constraints of your organization.

00:01:10 Session Overview

Speaker: Varun Mayya

• The session promises to move from theory to execution: translate the Lightning Way into practical AI-enabled workflows.
• Goals include improving writing and thinking, researching better, and aligning teams at scale through precise specifications.
• The core problem: misalignment between what leaders intend and what teams actually execute due to vague prompts and missing context.

• Key framing: clarity is a muscle you train with explicit prompts, shared references, and concrete tasks.

00:03:03 The Genie Thought Experiment

Speaker: Varun Mayya

• Varun introduces a thought experiment with Rohit to illustrate how a literal-thinking Genie (KD Genie) would interpret requests. The Genie takes things literally and fills in missing context with the executor's own biases—often with dangerous or unintended consequences.
• The point: both AI systems and human teams behave like genies when instructions are vague. If you want a specific outcome, you must prescribe precise, edge-proven instructions and provide explicit context.

Blockquote (verbatim example): "I am a genie. I have three wishes, but I take things too literally. I fill in context based on what you say, not what you mean. If you say unlimited money, I might route it through illegal channels unless you specify legal means and constraints." (Paraphrase of the Genie thought experiment; timestamp not explicit)

• The demonstration shows how different people interpret the same request differently (e.g., “ice cream” flavor or color) depending on their personal experiences and context, illustrating how work context compounds misalignment in teams.
• The kernel insight: the Genie thought experiment is a usable mental model for teams to diagnose where requests lack context and what gaps a manager must fill with prompts, references, and examples.

00:05:55 AI as a Literal Genie

Speaker: Varun Mayya

• The Genie metaphor evolves into a practical warning: most AI/human employees will do what you ask unless you provide enough context; they will fill in with their own defaults which may be misaligned with your intent.
• Distinguishing between positive vs. negative context: a good Genie will fill in positive, service-oriented context; a KD (literal) Genie will fill in whatever context it believes to be accurate, which can be dangerous if misaligned with real-world constraints (e.g., routing money through offshore accounts or triggering tax liabilities).
• The lesson: to scale clarity, you need to engineer prompts with enough context and guardrails so the AI/human teammates don’t improvise misaligned solutions.

Notable idea (paraphrase): "A good Genie fills in context that helps you achieve your legitimate goals; a KD Genie fills in context with the least risk to itself—the result is often misaligned with your intent." (Varun; no explicit timestamp)

• The episode connects the Genie concept to management: hiring, guiding, and feedback require explicit context so that teams do not improvise onto outcomes that degrade alignment or create liability.

00:10:40 Practical Writing Techniques with AI

Speaker: Varun Mayya

• Writing is framed as a form of thinking. Clarity in writing mirrors clarity in thinking; if your thinking is unclear, your writing will be too.
• The problem with AI mimicking style: it can imitate your syntax but won’t truly think or extrapolate nuanced situations the same way you would.
• Core idea: prompts are a way to transfer your thinking into concrete actions. Context is king; writing is a mechanism to capture that context and translate it into outputs.
• The bottleneck in many projects is missing specification, not missing skill. The antidote is precise checklists, run-on-one-on-ones, and explicit expectations.

Blockquote (verbatim): "Writing is a type of thinking. It's active thinking. The only way you can tell if you've thought clearly is when you've executed and it led to results." (Varun)

• The concept of “example-driven prompts”: provide explicit examples to fix context gaps. If you want a file to be organized in a specific way, show a model folder structure and naming conventions.
• The “prompting” term is introduced as a way to encompass context transfer to AI. The more explicit the prompts (with examples), the better the AI will align with your intended outcome.

00:16:40 Using AI for Business Planning & Research

Speaker: Varun Mayya

• The pancake business thought experiment is revisited to demonstrate how to structure research and plan with AI: start with a set scope, a budget, and a goal for a one-year plan; then let AI surface edge cases and missing questions.
• The process emphasizes deep research questions and follow-up questions that the leader would usually ask. The AI becomes a proactive partner, asking clarifying questions to fill gaps.
• The core practice: treat AI as a co-researcher that surfaces edge cases, but validate numbers with real-world data (e.g., talk to a chef to verify salary ranges, equipment costs, and operating costs).
• The speaker advocates a method: present the AI with precise context (location, budget, constraints), and expect a set of concrete outputs (costs, hiring plan, space requirements, channels).

Notable idea: AI is excellent at producing exhaustive, edge-case-rich outputs, but humans must verify data grounded in reality to avoid hallucinations.

• The importance of a robust follow-up process: if AI returns a draft plan, request follow-up questions from the AI as if it were a colleague prompting you for more information.

00:20:18 AI Tools for Different Tasks

Speaker: Varun Mayya

• The talk enumerates practical applications: from generating ideas to building the plan, to designing incentives and tracking performance.
• The speaker demonstrates that AI can perform complex tasks like incentive design by leveraging knowledge (e.g., from OpenAI or other AI research) and then tailoring them to your context.
• Example: incentive design for a pancake business with a structured KPI framework:

- Total monthly sales target triggers a bonus pool (e.g., 5%). - Food cost percentage threshold determines bonus forfeiture if waste is high. - Clear allocation of bonuses by role (Head Cook 60%, Helper 40%).

• The emphasis is on using AI to design the framework and then validating with real-world inputs (e.g., talking to a chef for salary benchmarks, verifying costs with suppliers).

Notable idea: AI can design incentives that align profitability with behavior, but you must verify real-world numbers and adjust to your local market.

00:23:08 Communication Best Practices

Speaker: Varun Mayya

• The theme centers on reducing miscommunication by providing explicit references and context. When you give a directive, supply references (sample outputs, previous work, templates) so teammates know what success looks like.
• The process of building a shared language and references reduces the need for back-and-forth and speeds up execution.
• Example: use AI to generate reference thumbnails, mood boards, and scene references, then hand those to the team to ensure alignment on style and tone.
• The talk emphasizes that references and prompts create a shared mental model across teams, reducing surprises and misalignment.

Notable idea: writing plus image/video references help unify cross-functional teams around a common standard of output.

• The speaker discusses mood boards and reference-driven creation, noting that AI can generate multiple reference styles for thumbnails, scenes, and edits, enabling faster iteration.
• The broader claim: AI is a strong idea generator; humans should pair AI outputs with references to drive consistent execution.

00:30:15 Team Management & Delegation

Speaker: Varun Mayya

• The focus shifts to leadership clarity, management of incentives, and the need for performance reporting to keep large teams aligned.
• Unreliable narrators are a risk: reports can reflect perceptions rather than actual inputs. The speaker notes real-world trading of storylines about workloads, efficiency, and timelines.
• Key observation: new editors require time to ramp up; a two-editor-per-one-shot-per-day model may be more sustainable for teams scaling quickly.
• The talk emphasizes that without clear reporting, incentives can become misaligned (e.g., overreporting progress, underreporting bottlenecks).
• A recurring motif is to build a culture of transparent, data-driven reporting (e.g., weekly hiring reports) to surface bottlenecks and keep teams on track.

Notable idea: unreliable narrators distort perception of performance; robust, frequent reporting helps align expectations with reality.

• The section underlines a powerful management axiom: the company’s scale and cost of coordination demand high-context feedback and explicit expectations. If you are not clear, you are risking high-performing people leaving for better-aligned teams.

00:37:25 Leadership Clarity & Performance Management

Speaker: Varun Mayya

• This is the core of the Lightning Way: clarity as a leadership superpower. Clarity reduces ambiguity, accelerates decision-making, and aligns teams on strategic and operational goals.
• The speaker asserts a strong link between leader clarity and career outcomes: “Your clarity as a leader or lack of it can make or break careers.”
• He emphasizes the cost of poor delegation and misalignment: high-performing individuals will disengage if they perceive their efforts are wasted or misdirected.
• The “sunshine test” is introduced as a pragmatic rule: if your statements or decisions were printed on a newspaper, would you be ashamed? This is a test of ethical clarity and leadership integrity.
• The talk argues for long-run endurance: teams and individuals do their best work when they are allowed to operate with clear expectations and a sustainable pace over years, not just months.

Notable quote (verbatim): "Your clarity as a leader or lack of it can make or break careers." (Varun; timestamp: 00:40:14)

• The segment also ties together the concept of AI-assisted clarity with human leadership: AI helps surface and structure the required contexts, but the leader must define the guardrails, values, and long-term vision.

00:42:44 Closing Thoughts

Speaker: Varun Mayya

• The closing encourages enjoying the journey and maintaining stamina for long-running initiatives. Pace matters; sustainability is crucial for enduring success.
• Varun likens the organization to backstage access in WWE, emphasizing transparency, performance visibility, and public-facing outputs as part of a larger cultural and business strategy.
• Final exhortation: use AI to scale clarity, delegation, and execution, but maintain responsibility and human judgment. The message is to balance ambition with humane leadership.

Notable closing idea: “We need to apply responsibility with great power, but let’s have fun and do it long-term.”

00:00:00– Resources Mentioned

• AI Tools and Platforms:

- OpenAI (GPT family, AI research platform) - Gemini (Google's AI model) - GPT-5 (referenced as current-era capability; not a released product in 2024–2026 timeframes) - Nano Banana (reference and storyboard/shot reference tool) - V3 (video/creative generation tool; workflow stage)

• Organizational Tools & Channels:

- Slack, WhatsApp, Google Docs (communication and documentation)

• People & Organizations:

- Rohit (participant in the Genie Thought Experiment) - Ranit Magnani (noted for hiring reports) - YAS (the creator’s company) - AOS (the Lightning Way framework often cited in the session) - Andy Grove (the pancake example’s lesson around spec and process)

• Market Channels & Tactics:

- Instagram ads, food delivery apps, local events (marketing channels in the pancake business thought experiment)

• References & Creative Techniques:

- Mood boards, thumbnail references, cinematic references, cross-language film references (for creative direction)

• People and Roles:

- Head cook, editors, designers, channel schedulers, interns, account managers (pancake business example)

00:00:00 Notable Visuals and Diagrams (described)

• Genie Thought Experiment slide: Visual of a genie receiving three wishes with a note that the genie is literal and fixes missing context when interpreting requests. This visual is used to demonstrate how vague prompts lead to misaligned outcomes.
• The “pancake business” thought experiment: A mental model for budgeting, staffing, and operations in a new venture, contrasted with AI-assisted planning.
• A slide comparing AI’s generation of ideas vs. human creativity: the claim that AI offers many ideas, but humans must provide the context and direction to make them actionable.

00:00:00 Key Takeaways from the Session (condensed)

• Clarity and context are the core levers that scale execution with AI.
• AI is a powerful idea generator, but humans must shape direction, constraints, and ethical guardrails.
• Writing is thinking: effective prompts convert thinking into actionable outputs.
• Edge-case research and bottom-up verification (e.g., talking to real-world practitioners) prevent hallucinations.
• Incentives, reporting, and transparent performance management are essential when scaling teams.
• Maintain sustainable pace and long-term work ethic; leadership discipline matters as teams scale.

00:00:00 Resources Mentioned (extracted)

• OpenAI, Gemini, GPT-5, Nano Banana, V3, Slack, WhatsApp, Google Docs, Instagram ads, food delivery apps, local events, Rohit, Ranit Magnani, YAS, AOS, Andy Grove.

00:00:00 Notable Quotes (verbatim)

"You know GPD and Gemini today are better idea generators than [music] human beings. Smart humans have three four good ideas. But after that all their ideas are generic. You cannot have a world where it's just AI or human. I think we're going to live in a very meshed world." (Varun Mayya) 00:00:00

"The communication between you and your teams only 30% is in your employees heads. Your clarity as a leader or lack of it can make or break careers." (Varun Mayya) 00:00:00

"Writing is a type of thinking. It's active thinking. The only way you can tell if you've thought clearly or not is when you've executed it and it has led to results." (Varun Mayya) 00:10:40

"The bottleneck is almost missing spec. Spec is that writing sheet, whatever needs to be done or explained or instruction, not missing skill." (Varun Mayya) 00:10:04

"Your clarity as a leader or lack of it can make or break careers." (Varun Mayya) 00:40:14

"With great power comes great responsibility." (Varun Mayya) 00:40:14

"Enjoy yourself and have fun. People forget that this is like a long sport. Pace matters; sustainability is crucial." (Varun Mayya) 00:42:44

00:00:00 Summary of the Materials and Frameworks

• The central framework is the Lightning Way: using AI to scale clarity, delegation, and execution while maintaining human oversight to avoid hallucinations and misalignment.
• The Genie Thought Experiment is a core mental model to diagnose and address vagueness in requests, prompting teams to provide precise context, examples, and references.
• The session emphasizes that writing and prompts are acts of thinking; the more explicit the prompts (including edge cases and references), the closer the AI output aligns with intent.
• The talk repeatedly returns to the importance of incentives, clear performance indicators, and transparent reporting to manage scaling teams effectively.

00:00:00 Additional Notes on Implementation (takeaways for practitioners)

• Start with a concrete problem and budget.
• Build a shared workspace of references (templates, style guides, prior outputs) to align team output.
• Use AI to surface edge cases and follow-up questions; validate with real-world numbers.
• Design incentives tied to profitability and measurable outputs; ensure they are simple and auditable.
• Maintain a humane leadership approach: care for people, cultivate a positive culture, and avoid overworking non-urgent tasks.

00:00:00 References and Practical References (for recreation)

• The talk references practical exercises: pancake business plan, channel briefs, and writing prompts. It demonstrates how writing and prompting can be used to generate outputs that align with business goals.

00:00:00 Notes on Execution (how to apply this in your team)

• Teach the team to generate prompt templates with concrete examples and expected outputs.
• Use AI for research planning and to surface missing questions; always verify with live data and subject-matter experts.
• Create an incentives framework that ties performance to profitability and quality metrics; use AI to design the framework and human oversight to enforce it.

00:00:00 Glossary (inline definitions)

• AI: Artificial intelligence; computer systems designed to perform tasks that normally require human intelligence.
• Prompting: The act of giving instruction to an AI to generate outputs; prompting is the mechanism to transfer your thinking into action.
• Context: All the background information that influences how a task should be performed; more context leads to outputs that better match intent.
• Edge cases: Unusual or extreme scenarios that are not part of the normal case; planning for edge cases reduces risk.
• Incentives: Reward structures designed to motivate specific behaviors and outcomes.
• Coordination costs: The friction and overhead involved in getting multiple people or teams aligned to complete a task.
• Unreliable narrators: When reports or briefings misrepresent reality due to misperception, biases, or incomplete data.
• Sunshine test: If your actions or statements were printed on a newspaper, would you feel comfortable about it? A test for leadership integrity.
• K​PIs: Key Performance Indicators; metrics used to measure whether a team is achieving its goals.
• V3, Nano Banana: Tools for video creation and reference generation used in the session to illustrate AI-assisted creative workflows.

00:03:03 Topics Covered (tags)

• AI And Leadership
• Clarity And Context
• Genie Thought Experiment
• Incentive Design
• Team Management
• Creative Process With AI

00:00:00 Final Note

• The overall message is that AI is a force multiplier for clarity and execution when paired with disciplined leadership. Treat AI as a partner that scales your ability to communicate expectations, surface edge cases, and align teams, while you, as the leader, maintain the guardrails, human judgment, and sustainable pace. This is how you scale clarity, delegation, and execution across complex systems.