I have a notes app on my phone with over two hundred product ideas in it. Some are a single sentence. Some are a full paragraph. Some have little diagrams I drew at 2am when I couldn't sleep because a concept had lodged itself in my brain and wouldn't leave. Two hundred ideas, most of them written over the past five or six years, almost none of them acted on. Not because they were bad ideas. Because I had no way to build them.
That gap - the distance between having an idea and being able to make it real - is the most frustrating thing about being a non-technical person with a technical imagination. You can see the product. You can describe it. You can sketch the interface, write the copy, plan the user journey. But you can't build it. The last mile is a wall, and on the other side of that wall is everything you want to make.
Claude Code demolished that wall. Not gradually, not with a gentle opening. It demolished it. And in doing so, it didn't just give me a new tool. It changed how I think.
The gap that disappeared
Before Claude Code, the process of turning an idea into a product looked like this: have the idea, get excited about the idea, realise you can't build the idea, try to find someone who can build the idea, fail to find someone, try to learn to code, get frustrated, give up, write the idea in your notes app, move on. Repeat two hundred times over several years.
After Claude Code, the process looks like this: have the idea, open the terminal, describe what you want, build it, ship it. The gap between concept and execution went from months (or never) to hours. That isn't an incremental improvement. That's a category change. It's like going from writing letters to making phone calls. The medium didn't just get faster. The entire nature of communication changed.
And here's what nobody tells you about tools that close the execution gap: they change what you're willing to attempt. When building something takes months of learning and effort, you only attempt the ideas you're most committed to. The stakes are too high for experiments. But when building something takes an afternoon, suddenly you're willing to try anything. The wild ideas, the silly ideas, the "I wonder if" ideas that you would never have invested serious time in - they all become possible. The tool doesn't just execute faster. It changes your relationship with risk.
The hammer problem
There's a famous saying - usually attributed to Abraham Maslow, though nobody is entirely sure - that goes: "If all you have is a hammer, everything looks like a nail." This is the danger of powerful tools. They don't just enable you to build. They shape what you build. And if you aren't careful, you end up building things because you can, not because you should.
I've caught myself doing this. An idea pops into my head and within thirty seconds I'm thinking about how Claude Code would build it. Not whether it is worth building. Not whether anyone needs it. Not whether it serves a genuine purpose. Just how it would work, technically, as a product. The tool creates a gravitational pull toward making things, and that pull can override the more important question of whether a thing should be made at all.
This isn't a problem unique to AI tools. Designers with access to beautiful design software make things that look gorgeous but serve no purpose. Writers with access to publishing platforms publish things that didn't need to be published. Photographers with expensive cameras take photos that didn't need to be taken. Every powerful tool comes with the temptation to use it just because you can. The tool whispers "you could build this" when the better question is "should you build this?"
The antidote, for me, has been taste. Taste is the filter between capability and decision. Claude Code can build anything I describe. But taste tells me which things are worth describing. Taste says "this idea is interesting but not useful." Taste says "this would be impressive but not meaningful." Taste says "this is a solution looking for a problem." Without taste, I would have built fifty products by now, most of them pointless. With taste, I've built eleven, and each one exists for a reason.
The tools I actually use
My daily toolkit is laughably simple. Claude Code for building. A browser for testing. Netlify for hosting. A notes app for ideas. Plausible for analytics. That is it. No Figma, no Notion, no project management software, no version control beyond what Claude handles, no complex deployment pipelines. Five tools.
Each of these tools shapes my output in a specific way. Claude Code shapes what I build - conversational, iterative, fast. The browser shapes how I test - always in real conditions, always on the actual site, never in a preview or staging environment. Netlify shapes how I deploy - instant, simple, no friction between "it works locally" and "it is live." The notes app shapes how I capture ideas - quick, messy, unfiltered. Plausible shapes how I measure - privacy-first, simple dashboards, no overwhelming data.
The simplicity is deliberate. Every tool you add introduces complexity. It introduces another interface to learn, another workflow to maintain, another potential point of friction between you and the thing you're trying to make. I've seen teams with twenty tools in their stack who ship less than I do with five. Not because they're less talented. Because the tools create overhead, and overhead creates drag, and drag slows everything down.
What tools can't replace
There's a narrative in the AI conversation that goes something like: "AI will replace everyone." I find this narrative lazy and wrong. AI tools are extraordinarily powerful, but they replace execution, not judgement. They replace the "how" but not the "what" or the "why." And the "what" and the "why" are where all the value lives.
Claude Code can build a website in minutes. But it can't tell you whether that website should exist. It can't tell you whether the problem it solves is a real problem or an imagined one. It can't tell you whether the design feels right, whether the tone matches the brand, whether the product fills a genuine gap in someone's life. Those are human decisions. Taste decisions. Strategy decisions. The kinds of decisions that require lived experience, cultural awareness, and the ability to connect dots that a machine can't see.
This is why I think the "AI replaces people" narrative misses the point so badly. The people who will thrive aren't the ones who can code the fastest or design the most efficiently - AI will always beat them at speed. The people who will thrive are the ones with taste, with judgement, with the ability to look at a thousand possible things to build and choose the one that actually matters. The tool handles execution. The human handles meaning.
I say this as someone who has benefited enormously from AI tools. Claude Code changed my life in a very literal sense - it turned me from someone who talks about building things into someone who actually builds them. But it didn't change my taste. It didn't change my judgement. It didn't change my ability to spot a good idea or kill a bad one. Those were mine before the tool arrived, and they'll be mine after whatever comes next.
Choose your tools carefully. Not just for what they can do, but for how they'll shape your thinking. Every tool is a lens, and every lens has a bias. The best tool is the one that amplifies your strengths without distorting your judgement. For me, that tool is Claude Code. But the real tool - the one that matters most - is still taste.
