"Where do you get these fairy tales about AI? Nobody actually uses it like that!"
Here is my answer. Not a feeling. Not a Silicon Valley manifesto. An audit. My data is a breakdown of current hiring needs across 1,131 companies, gathered from public sources and updated daily. The debate over whether companies "actually use" AI does not get settled by vibes. It gets settled where a company posts a job description, lists requirements, and promises to pay money for them.
So let's compare. I have a corpus. The other side has "it seems to me."
Feelings vs. Payroll
The argument "nobody uses AI like that" survives right up until you ask what it is built on. Usually, it is a feed of acquaintances, a couple of disappointing demos, and a general fatigue of "this hype again." That is not data. That is mood.
Mood is worthless and predicts nothing. A company that writes "must be able to build GenAI and drive AI transformation" in a job spec is putting a hiring budget, a salary, and six months of a new hire's work on the line. A job requirement is a bet placed with money. An expert's feeling is a bet placed with nothing.
So I am not arguing. I am showing the invoice.
The 1,131 Companies
To be clear about who we are talking about. This is not a sample of garage startups or subscribers to a single tech blogger.
Top-5 countries by number of companies in the sample:
- UAE — 222
- Serbia, unexpectedly — 168
- Saudi Arabia — 130
- Germany — 94
- Ireland / Qatar — 85 each
Then Kuwait, Bosnia, the Netherlands, France. This is not "Silicon Valley wrote a manifesto." This is a market spanning from the Gulf to Europe that pays salaries and publishes requirements. It is a geography that nobody holds in their head when they talk about "AI hype."
What these companies do — top industries by number of unique companies:
- IT Services & Consulting
- Software Development
- Technology / Information
- Financial Services
- Internet
Consulting, product, finance. Not a garage startup. The ones making money on technology and client funds.
And who they are looking for. The absolute top roles:
- Engineering Manager
- Head of Engineering
- Product Manager / Solution Architect
- Project Director
- Senior Solution Architect / Enterprise Architect
Also in the top 10: AI Architect and CTO. Here is the first blow to the myth that "the market hires AI enthusiasts." It does not. The market looks for people who run engineering, architecture, and product — and demands AI inside that perimeter. Not "play with a model," but "embed AI into what you already manage."
Not "Mentioned AI," but What They Do With It
AI is mentioned at all by ~62% of companies. That number alone demolishes "nobody uses it like that": not "nobody," but nearly two-thirds of the sample. But a mention is a weak argument; a word can be slapped onto anything. So let's break down what stands behind it.
Top-10 applications:
- LLM / GenAI (RAG, agents, orchestration, foundation models) — ~559 cases. Examples: agentic workflows, LLM orchestration, model evaluation, RAG patterns.
- Classical ML / Data Science (models, MLOps, computer vision) — ~386 cases. Examples: AI/ML platforms, model deployment, inference, production ML.
- "Just wrote AI" with no explicit scenario — ~276 cases. The most honest item in the ranking: part of the market still slaps the word on the storefront. This does not cancel the other nine — but I am not going to hide it either.
- AI transformation / strategy / roadmap — ~267 cases. Examples: AI adoption across an org of 70+ engineers; enterprise AI-driven solutions.
- AI governance / risk / compliance / safety — ~174 cases. Examples: AI governance models, risk reviews, lifecycle stages, documentation.
- AI architecture / platform / infrastructure — ~167 cases. Examples: reusable AI architecture patterns, AI platform and integration projects.
- AI for engineering velocity (Copilot, Cursor, coding assistants, SDLC) — ~167 cases. Examples: Copilot / Claude / Cursor in daily workflow — "not familiarity, but demonstrated integration."
- AI product / feature development — ~147 cases. Examples: ML/AI-driven feature development alongside Product.
- AI Delivery / leadership over AI work — ~114 cases. Examples: end-to-end delivery of AI initiatives — from business case to production.
- AI ops / automation / process — ~74 cases. Examples: intelligent automation, document AI.
Note the honesty of the list. Item 3, "just wrote AI," sits right in the ranking; it is not hidden. Yes, part of the market hangs the word on the storefront. But 559 GenAI cases and 386 classical ML cases are not a storefront — that is work. An opponent saying "nobody uses it like that" usually sees item 3 and pretends the first two do not exist.
Governance and Platform Are Not Exotic
The next objection is: "Sure, anyone can wire up a model, but nobody does it seriously." Looking further down the list, that objection fails too.
AI governance / risk / compliance / safety — ~174 cases. AI architecture / platform / infrastructure — ~167 cases. These are not hobby projects. They signal that AI has stopped being something people poke with a stick and has entered the perimeter of accountability: risk reviews, lifecycle stages, documentation, reusable architecture patterns, platform. Governance and platform in this corpus are not a rarity; they are table stakes.
One caveat: AI Delivery as a distinct discipline — leadership over AI work, end-to-end from business case to production — exists in the corpus but is weaker: ~114 cases against 559 for LLM and 386 for ML. The market is already hiring en masse for GenAI and transformation, but it formulates the dedicated role of "the person who drives an AI initiative to a result" less often. This is not a hole in the argument. It is a blank spot that shows exactly where the market moves next.
What They Need It For — Not an HR Chatbot
The business objective matters too. What companies do with AI technically is one thing. The business objective driving it is another. This is where the myth about "toys and hype" dies.
The absolute top business objectives — what they use AI for:
- Strategic AI transformation / not falling behind the market — 26%
- Comply with regulation / reduce legal & model risk (AI Act, governance, responsible AI) — 18%
- Ship an AI product / AI features (revenue through product) — 13%
- Accelerate time-to-market / engineering delivery — 11%
- Better decisions: forecasting / insights / decision support — 10%
Then: cut costs, raise efficiency, protect the business, build other AI capabilities, improve customer experience, logistics, and sales.
Read the first two items again. 26% is strategy and the fear of falling behind. 18% is regulation and risk: AI Act, governance, responsible AI. Nearly half of the market's motivation is not "I want a trendy feature." It is "I am afraid of being left behind" and "I am afraid of getting hit by a regulator." That is not how hype-chasers behave. That is how companies that already count AI as a survival factor and a legal risk simultaneously behave.
Short version: the hiring market is not screaming "build us an HR chatbot." It is screaming "know how to build GenAI and drive AI transformation." Governance and platform are already table stakes. AI Delivery as a distinct discipline exists but is currently weaker than product and LLM.
Who This Market Is Actually Looking For
Together: Top roles: Engineering Manager, Head of Engineering, architects, Project Director, plus AI Architect and CTO. Top applications: GenAI, ML, transformation, governance, platform. Top objectives: do not fall behind, do not break the law, ship product, accelerate delivery.
This is not a description of "one more AI engineer for the staff." This is a description of a person who holds engineering and architecture, can run AI transformation as a program, and does not drop it on governance and risk. Exactly the perimeter the corpus shows as real demand — not the one invented in threads about "AI will replace everyone." And the most interesting part about AI Delivery: the fact that it is currently weaker than LLM and product in the corpus is not its absence; it is a phase. Mass GenAI hiring without a dedicated discipline for driving things to a result does not live long. The blank spot is white, but it fills predictably.
So back to where we started. "Nobody uses AI like that" is a claim about facts. Here are the facts: 1,131 companies, ~62% with AI, 559 GenAI cases, nearly half the motivation in transformation and regulation, governance as table stakes. All of it sits in public hiring requirements, not in my head.
Those saying "nobody uses AI like that" must have broader data than 1,131 public job specs. Show it. We will compare.