Nash David

Building AI systems for ecommerce funnels, user acquisition & efficiency • Exploring productivity & communication craft • Violin learner • By faith

[Nash David @nashpd]

More traffic will not save a broken store.

It will just fill it with more people who leave.

Three friction points kill most online sales before they happen.

[Nash David @nashpd]

@addyosmani Thanks for sharing, Addy. Was beginning to hit the wall with tons of updates. Just when I figure something there are a million new updates. Loops being a very critical one now. Thanks for explaining the universe in a way that I can find information in my personal exploration.

[Nash David @nashpd]

@RyanHoliday And the people you pissed off often become the proof that you were onto something real.

[Nash David @nashpd]

@russellbrunson The moment you stop needing external validation is the moment of redemption. Free from the weight of acknowledgment. Free from the trap of dependency.

[Nash David @nashpd]

@trq212 New to Claude Code, but the harness concept grabbed me early. I use it for life OS tasks, not just code — so naturally I started letting Claude design its own workflow and focused on optimising from there.

[Nash David @nashpd]

@swyx I learnt one more thing from framing as a question. It bring clarity with what I want. Most bad prompts are ineffective because the unclear thinking arrived at the model before it was ready. Like someone asks what would you like to have at a food joint, and i say anything :/

[Nash David @nashpd]

@andrewchen What makes Kurzweil uncomfortable to read now is that he was not guessing. He was extrapolating from a curve that was already running. The forecast felt like science fiction in 1999 because the math was decades ahead of the intuition. Most people still have the intuition problem.

[Nash David @nashpd]

@levelsio I have come to appreciate staying a year on one problem. Most people would have dropped in a week. That is the part nobody talks about when they say "just ship it." Shipping is easy. Staying with it until it works is the whole game. I'm learning this.

[Nash David @nashpd]

The fix is not more ads. It is a 30-minute audit: open your store as a stranger. Read every page with the question "Is this for me?" Fix every place the answer is unclear. Most stores can double conversion without spending a $/₹ more on traffic. Where's your store losing people?

[Nash David @nashpd]

More traffic will not save a broken store.

It will just fill it with more people who leave.

Three friction points kill most online sales before they happen.

[Nash David @nashpd]

#3 Post-click silence. The customer bought. Nothing arrives. No confirmation. No expected delivery. No brand voice in the receipt email. Buyer's remorse fills the silence. Returns go up. Reviews go unwritten.

[Nash David @nashpd]

@naval Never thought something like adobe would fade into a command line on an llm but yes there it is happening. And there are more examples...

[Nash David @nashpd]

@shaneparrish Same logic applies inside businesses. If AI can run your process start to finish, the process was never the value. The judgment around when to run it, how to adapt it, and what to do with the output still is. Most companies are protecting the wrong thing.

[Nash David @nashpd]

@levelsio Building was never the moat. AI just made it obvious. Shipping cost dropped to zero. Getting anyone to care did not. The distribution problem has nowhere left to hide.

[Nash David @nashpd]

@harrydry The analogy works because nobody has built the equivalent of an hourly rate for AI output yet. We know what a developer hour costs. We know what a designer day costs. Token spend has no such anchor. Until operators build that mental model, every bill will feel like a surprise.

[Nash David @nashpd]

@simonw Caps do something counterintuitive. They turn a commodity into an asset. When the spend is unlimited, the tool becomes background noise. When there is a ceiling, every prompt has to earn its place. Uber may have accidentally created more value by constraining it.

[Nash David @nashpd]

That is not a party trick. It is a different category of tool. One most people are still treating like a faster search bar. The gap between operators who have made this shift and those who haven't is getting wider every week. Think systems. Not questions.

[Nash David @nashpd]

Most people using AI tools are treating them like search engines with better answers.

They are not.

[Nash David @nashpd]

Last week I gave Claude Code a task with 12 sequential steps. Told it what I needed at the end. Left it alone. It came back with something I could not have produced in a day working manually.