Early, Not Wrong

Okay, first post. Not what I planned but write what you want to write. Lets build habits together people.

I keep seeing big claims, BIG claims that X is over. The tech ones are the ones that resonate with me the most. Predictions of what the future hold and then obviously people showing how those predictions are wrong. Honestly, getting a bit bored of it and this quote just keeps coming to my mind:

“Insanity is doing the same thing over and over again and expecting different results”. — Often attributed to Einstein but was actually from a novel published in 1983 by Rita Mae Brown.

The “Asylum” includes..

1. In 1960, Herbert Simon said - “Machines will be capable, within twenty years, of doing any work a man can do.”
2. Marvin Minsky joined him 7 years later (1967) claiming that AI problems will be largely solved with a generation.
3. Elon has been claiming that self-driving cars are a solved problem since 2015
4. Zuck’s Metaverse has not become the natural successor to the internet.
5. The various claims around Blockchain (Santander - $20B saving, Smart Contracts replacing Lawyer) are yet to pan out.
6. And finally the various CEOs that have moved the goal posts on software engineers, lawyers etc being replaced by LLMs

To be clear — optimism about what you’re building isn’t insanity. That’s just enthusiasm. The insanity is that we keep acting surprised when the timeline is wrong. Every. Single. Time.

Some of what has been discussed has come to pass though and in each case people continue to work the problems however I think the pattern here is that these things take a lot longer to do than anyone knows. If you treat the data points as trends instead, you just continue to see a pattern of unjustifable confidence about timing even though many of the predictions do eventually come somewhat true.

Coding is dead. Software engineering isn’t.

I prefer what Grady Booch’s is saying which I saw on a podcast the other day. This idea that we’re entering the next golden age and that abstraction is rising to a new level. Giving engineers the freedom to not worry about lower level concerns and they can then focus on more interesting problems. We all like the sounds of that.

Personally, I think that Grady skirted a problem that I see.

Specialist vs Generalist

If you’ve read the about page, you know where I lean on this one

When demand for coding skills exploded, we built an entire industry around filling that demand. We industrialised the creation of specialists and that’s what we needed. The traditional abstraction models weren’t moving fast enough, we were seeing replacement frameworks and new ways of doing things that weren’t shifting the needle enough.

Taking web apps as an example, where most parts of it are just CRUD apps. It’s not hard to see why an LLM enabled tool can replace that. We were already starting to see attempts at this with general no-code platforms like Power Apps or Honeycode alongside all the various flavour of the month frameworks.

Grady’s response to the Dario Amodei claim was something along the lines of ‘coding is dead, not software engineering’. I can believe that. Take web apps, most of it is just CRUD. The same patterns, assembled differently. We were already seeing attempts to replace that with no-code platforms like Power Apps and Honeycode. An LLM-enabled tool doing it isn’t a leap, it’s an inevitability. I find myself writing less and less code by hand, leaning on Claude Code and Kiro more and more to write it for me.

Who doesn’t want that? I certainly didn’t want to specify the CSS by hand for this blog, I did spend a lot of time deciding what I wanted and doing A/B tests. I wrote documentation describing what I wanted and tested the colours. If I can avoid never writing CSS again in my life, I will.

IMO the bigger problem

Where did all the founder led innovators and disruptors go? Why is enshitification seeming to become the norm rather than an exception?

Businesses are becoming stale. Even startups are feeling derivative, just following trends and chasing the same markets using the same playbooks. The ones that should be disrupting are instead focused on shareholders over consumers and treating AI as a strategy rather than a tool.

There are always exceptions, don’t @ me.

That’s blood in the water.

For me, what gets lost in the “will it replace engineers” debate isn’t really about headcount. It’s about what consumers are going to start demanding. Sharper products. More opinionated. Built around a specific problem rather than a sprawling platform trying to be everything to everyone. AI makes it possible to build and iterate at that level of focus without the overhead that used to make it unviable.

Consulting is already seeing the consumer changes. 78% of consulting buyers now expect cost savings from AI. 65% say traditional models no longer deliver value. McKinsey has quietly moved 25% of its global fees to outcome-based contracts because consumers stopped accepting “here’s a scope, what’s the fee?” and started asking “here’s the outcome, what’s it worth?”

The expectation compression is already happening. Consulting is just the first industry where consumers have enough leverage to force the conversation. Software, media, marketing: every category built on packaging expertise as a service is next.

Take Procore for example. Someone with deep construction knowledge and frustration with generic CRM tools built a vertical platform worth over $10 billion. They did it the hard way, years of development, significant investment. The question is what happens when the next person with that same specific knowledge and that same customer obsession doesn’t need either.

The bar to mount a genuine challenge to an incumbent just dropped, significantly. A small team, with a sharp and focused point of view and some genuine customer obsession. With the clarity to know what they’re building and why can do things they’ve never been able to do before - quicker, cheaper and better.

Survivors are going to be organisations treating AI as an amplifier of good judgement, clear thinking and genuine customer obsession. Not a crutch propping up a bloated roadmap. Not a replacement for the hard thinking about what you’re actually building and why.

So organisations quietly reducing their developers aren’t leaning out and getting faster. They’re not tackling the institutional bloat that slows decisions or getting closer to their customers’ problems. And the talent they’re cutting doesn’t just move sideways into another big org. The same AI that made them redundant in one context makes them viable in another. Instead of concentrating at fewer large employers, I think we will see them distribute outward into smaller, more focused teams building things whole rather than owning a slice of something bloated.

They’re throwing chum in the water.

The sharks are coming. A couple of engineers, an opinionated view, real customer obsession, and a Claude subscription.