In that famous Vanity Fair interview, Jony Ive shared the most important lesson Steve Jobs taught him: “Focus means saying NO to something that—with every bone in your body—you think is a phenomenal idea, and you wake up thinking about it, but you say NO to it because you’re focusing on something else.”

The Daily Question That Changed Everything

Jobs would ask Ive almost daily: “How many times did you say no today?” The more noes, the better. It wasn’t about being negative—it was about being ruthlessly selective.

I immediately could see how this applied to my scattered approach to tools and technologies. I’ve been jumping between Copilot, ChatGPT, various coding assistants, content creation tools, and productivity apps. Each one seemed like a phenomenal idea at the time.

But here’s what Jobs understood that I didn’t: focus isn’t something you decide on Monday. It’s every minute asking “Why are we talking about this? This is what we’re working on.”

My Claude Code Focus Experiment

Right from day 1 of reading about this philosophy, I made a decision that would have seemed crazy to me just weeks ago. I’m saying no to everything else and betting entirely on Claude Code.

Not just for coding. For everything:

  • Content creation
  • Blog writing
  • Research and analysis
  • Project management
  • Learning new technologies
  • Building my expertise

This blew my mind when I realized what Claude Code actually is. It’s not just another coding assistant—it’s an agentic coding tool that lives in your terminal, understands your entire codebase, and can take real actions across your development workflow.

The Agentic Advantage

What makes Claude Code different from the scattered tool approach I was taking before isn’t just the features—it’s the integration. Direct terminal integration means no switching between applications. Full codebase awareness means it understands context across all files. It’s action-oriented, actually editing files, running tests, creating commits rather than just suggesting changes. Web-enabled research lets it fetch documentation and search for solutions in real-time. These end-to-end workflows go from research to implementation to deployment seamlessly. But the real power isn’t in the individual features. It’s in the singular focus.

The London Expertise Goal

I’ve set myself an ambitious target: become the most knowledgeable person about Claude Code applications in London. Maybe that sounds arrogant, but Jobs would approve—it’s specific, measurable, and requires saying no to countless other opportunities.

Every day I’m exploring new ways to push Claude Code beyond its obvious use cases:

  • Using it for blog post research and writing (like this very post)
  • Project architecture and planning
  • Learning complex codebases
  • Content strategy and creation
  • Even meta-tasks like analyzing my own writing style

The depth you can achieve when you stop context-switching is remarkable. I can already see patterns and possibilities that weren’t visible when I was juggling multiple tools.

Saying No to Phenomenal Ideas

This focus philosophy is hard. Just yesterday I discovered three new AI tools that looked genuinely impressive. My old self would have spent hours exploring each one, setting up accounts, testing features.

Instead, I asked myself Ive’s question: “How many times did you say no today?”

The answer: three times. Three phenomenal ideas that I wake up thinking about but won’t pursue because I’m focusing on something else.

The Compound Effect of Focus

What fascinates me about this experiment isn’t just the productivity gains—it’s the compound learning effect. When you go deep on one tool instead of wide across many, you start seeing connections and possibilities that surface-level users miss.

I’m documenting everything: unusual use cases, workflow optimizations, integration patterns, failure modes. Not just for my own reference, but to build expertise that compounds over time.

Where This Leads

I don’t know how far this focus experiment will take me. Maybe I’ll become that person others turn to when they need to understand what’s really possible with agentic coding tools. Maybe I’ll discover applications that haven’t been explored yet.

What I do know is that Jobs was right about focus. It’s not about finding the perfect tool—it’s about going deeper with the right tool than anyone else is willing to go.

Every minute I’m not switching contexts is a minute I’m building expertise that can’t be easily replicated.

The experiment continues.

This post was researched, written, and published using Claude Code—from the initial YouTube search about Jony Ive to the final Jekyll deployment

Appendix: Claude Code Prompts Used

Here are all the prompts I used with Claude Code to create this blog post, demonstrating the agentic workflow in action:

  1. Initial Research: “this my post idea for today. Can you search YouTube for this? Called the Read tool with the following input: {“file_path”:”/home/vikram/git/vikrampawar.github.io/_drafts/focus.md”}”

  2. YouTube Search: Used WebSearch tool with query: “Steve Jobs focus saying no John Ive narrated YouTube”

  3. Specific Video Search: Follow-up search with: ““Jony Ive” “Steve Jobs” focus saying no YouTube video interview”

  4. Blog Post Request: “Great, can you write a blog post in my style, on why I find this fascinating idea. I’ll now practice this on focussing on Claude Code exclusively to do my work and gain as much knowledge on using it for various things - coding, blogging, writing, creating content etc. Agentic coding and see all the possibilities, see how far this will take me - may become the most knowledgeable on this topic in London one day.”

  5. Style Analysis Reference: “you may find a document where this analysis was already done by you in the past.”

  6. Content Refinement: “rewrite this part as a paragraph [referring to the bullet point list about Claude Code features]”

  7. Meta Documentation: “Add an appendix with a list of all my prompts for this blog post.”

Each prompt built on the previous context, showing how Claude Code maintains conversation state while performing research, analysis, writing, and editing tasks in a single coherent workflow.