How the IT service sector found the perfect excuse in AI#
Ten years in this industry - across product companies where engineering was treated like a craft, and service companies where it was treated like inventory - has taught me one uncomfortable truth: most IT service organizations never saw developers as developers. We were “resources.” Billable, fungible, disposable resources.
AI didn’t create this mindset. It just gave management a new vocabulary to justify it.
The Resource Economy#
In a resource economy, nobody asks if the work is good. They ask if the work is fast enough and cheap enough. Deliver more in less time has been the unofficial company motto since long before anyone in the C-suite could spell “LLM.” Sprints got shorter, scope got bigger, and the gap between the two was always filled by the same thing: a developer’s personal time, and eventually, their patience for the job itself.
Management, meanwhile, has rarely led - it has followed. Whatever term was trending at the last conference became the theme of the next townhall, regardless of whether it fit the actual problem in front of the team. Agile, then hyper-agile, then “we’re basically a startup now,” then digital transformation, then cloud-native everything - a parade of buzzwords, each one arriving at the direct expense of a developer’s time, respect, and sanity. Client happiness was the only KPI that ever mattered. Developer wellbeing was a line item nobody wanted to own.
AI is simply the newest costume on the same old mannequin.
“Anyone Can Code” - Said No One Who Actually Codes#
The most tiresome line to come out of this AI moment is the confident, faux-progressive declaration from leadership: “Anyone can code now. You don’t even need to be technical.”
It’s a convenient thing to say - if you’re not the one being asked to prove it.
If coding were truly that trivial, HR wouldn’t need developers to build the HR portal. Finance wouldn’t need engineers to fix the billing script. The same leadership team repeating this line in every all-hands isn’t rushing to open a code editor themselves. The claim isn’t a technical observation - it’s a rhetorical tool. Once an organization declares that “anyone can code,” it has quietly given itself permission to stop valuing the people who do. Your work stops being engineering and starts being just another checkbox, because if anyone could’ve done it, then what you delivered was never special to begin with. That’s not a technology shift. That’s a designed insult.
The AI Townhall Problem#
Every town hall now opens with “AI” and closes with “AI.” Every tagline has been rewritten to squeeze the word in somewhere, whether or not it changes anything about how the company actually operates. Cough or sneeze and someone will find a way to relabel it as an “AI-powered initiative.” This isn’t strategy. It’s optics - a company trying to look relevant to clients and investors, using its own engineering workforce as the backdrop for a slide nobody in leadership actually understands.
And the cost of that optics campaign always lands on the same desk: the developer’s.
Harder Growth, Sold as Progress#
Here’s where it stops being annoying and starts being damaging: under the banner of AI, performance bars are quietly being raised while timelines are being cut. The logic, as it’s been explained in more than one review cycle, is simple - “AI makes you faster, so you should be delivering more. And besides, apparently anyone can code now, so the bar for ‘exceptional’ just went up.”
It’s a closed loop that management built for itself: first devalue the skill publicly, then use that devaluation to justify tougher promotion criteria and thinner growth paths. Both moves cost the company nothing. Both moves cost the developer everything.
Elephant in the room#
AI is not here to turn developers into people who are simply good at writing polite English prompts, no matter how badly the AI industry wants to sell that story. And they do want to sell it - every AI company right now has a vested interest in convincing the market that software development as we know it is over, that agents and prompts and loops are the entire future, that SDLC is a relic. It’s a good pitch if you’re trying to be indispensable for the next decade. Companies chasing every sector at once, launching a new AI for this and that product every week, aren’t doing that because the technology demands it - they’re doing it so that when the hype eventually cools, they’re not left irrelevant in a single market they never diversified out of.
My Actual Take on AI#
Here’s where I’ll push back on the doom, because I don’t think AI is the villain here - I think it’s being used as one.
When calculators arrived, they didn’t replace accountants. They became a wingman - something that removed the tedious arithmetic so the accountant could focus on judgment, on advising, on the parts of the job that actually required a human. Nobody declared “anyone can do accounting now” just because a calculator could add numbers.
So no - coding isn’t over, and neither is SDLC. A prompt doesn’t understand your business logic. It doesn’t know your users, your edge cases, your compliance constraints, or the five-year-old architectural decision that’s still quietly load-bearing. Someone who wasn’t a strong developer before AI won’t magically become one now, and someone who genuinely couldn’t code before still can’t, just with better autocomplete. What AI actually demands is more human judgment, not less - a human in the loop who understands why, not just what.
The tragedy is that service-based companies were never built to value that kind of human. They were built to value throughput. And if this AI bet doesn’t pay off the way leadership is promising the clients, the response won’t be introspection - it’ll be a new hiring post, replacing the very team that was told it was replaceable, without so much as a pause.
We were never developers to this system. We were always just resources. AI didn’t change that. It just gave them a more fashionable word for it.
Flush!
- This post is licensed under CC BY-SA 4.0