That distinction between writing code and developing software is the core of it. The easy part got easier. The hard part didn't budge. Security, compliance, observability, staging. I recently put together a checklist for vibe coders covering exactly the stuff AI skips. PCI DSS, GDPR, dependency auditing, structured logging. The gap between concept car and production car: https://reading.sh/your-vibe-coded-app-works-now-ship-it-properly-80b98f77ec25
Great insight per usual. There’s another useful analogy. Film making. Barriers to entry have come down. Now we have a lot more video content. More useful automation and tools. Costs have come down. But still takes the right work to get engagement.
MR.TODD sees patterns others don’t. Software Isn’t Dead—You’re Just Building the Wrong Layer—made me know—what’s actually happening versus what everyone incl.news reports say is happening.I may be wrong the Tech cos that are in AI are attempting to sell software making tools to create software for captive use, software that has no value,that has no upgrades and/or is a maintenance orphan as also software that’s not compatible with third parties like bankers, accountants, suppliers,customers etc.It can’t be integrated with other platforms.
MR.TODD’s insights resonate—Why AI makes domain expertise the only moat left—after reading thisTimely post(I was about to plan writing my software) I realised—writing code is not my core competency nor my business—and,on deep thinking I concluded—I’ll never become a billion dollar company on the basis of my software, so better just keep buying it from someone.
MR.TODD article answers in most APT manner with logic,and argues the classic question in operations of build versus buy. I’m convinced after reading MR.TODD valuable article I can tell that AI instead of killing software,will kill cos.who attempt on vertical integration of software. The smart move is usually to own whatever “makes your beer taste better” aka differentiates your product, and buy everything else.
This is the danger of agentic coding. Making software cheap/easy to make does not mean I should use that capability to start making software.I need computers,and,there’re tools that can help me assemble computers—so do I start making computers—ditto with everything I need.
For technical folks, what skillsets would you recommend we level up and invest in?
My general thought is that developers are going to take one of two paths:
1. Extremely technical on something very specific, like the inter working of a language compiler, for situations where LLMs do not have enough data to make good predictions and an expert is needed.
2. More generalists that understand how software gets built and are closer to product managers who use AI to write most of the code.
Wonder if that is something you are seeing as well.
That distinction between writing code and developing software is the core of it. The easy part got easier. The hard part didn't budge. Security, compliance, observability, staging. I recently put together a checklist for vibe coders covering exactly the stuff AI skips. PCI DSS, GDPR, dependency auditing, structured logging. The gap between concept car and production car: https://reading.sh/your-vibe-coded-app-works-now-ship-it-properly-80b98f77ec25
Great insight per usual. There’s another useful analogy. Film making. Barriers to entry have come down. Now we have a lot more video content. More useful automation and tools. Costs have come down. But still takes the right work to get engagement.
MR.TODD sees patterns others don’t. Software Isn’t Dead—You’re Just Building the Wrong Layer—made me know—what’s actually happening versus what everyone incl.news reports say is happening.I may be wrong the Tech cos that are in AI are attempting to sell software making tools to create software for captive use, software that has no value,that has no upgrades and/or is a maintenance orphan as also software that’s not compatible with third parties like bankers, accountants, suppliers,customers etc.It can’t be integrated with other platforms.
MR.TODD’s insights resonate—Why AI makes domain expertise the only moat left—after reading thisTimely post(I was about to plan writing my software) I realised—writing code is not my core competency nor my business—and,on deep thinking I concluded—I’ll never become a billion dollar company on the basis of my software, so better just keep buying it from someone.
MR.TODD article answers in most APT manner with logic,and argues the classic question in operations of build versus buy. I’m convinced after reading MR.TODD valuable article I can tell that AI instead of killing software,will kill cos.who attempt on vertical integration of software. The smart move is usually to own whatever “makes your beer taste better” aka differentiates your product, and buy everything else.
This is the danger of agentic coding. Making software cheap/easy to make does not mean I should use that capability to start making software.I need computers,and,there’re tools that can help me assemble computers—so do I start making computers—ditto with everything I need.
I’m Grateful to MR.TODD for the eye-opener post.
Well said Sheo. I appreciate the comment and glad the article resonated with you and added value.
Great article!
For technical folks, what skillsets would you recommend we level up and invest in?
My general thought is that developers are going to take one of two paths:
1. Extremely technical on something very specific, like the inter working of a language compiler, for situations where LLMs do not have enough data to make good predictions and an expert is needed.
2. More generalists that understand how software gets built and are closer to product managers who use AI to write most of the code.
Wonder if that is something you are seeing as well.