AI Readiness Isn't the Question Anymore
What the shift from assessment to adoption in European hiring says about where AI projects actually stall
Search European job postings today for AI readiness and the results are thin. Search for AI adoption, AI enablement or AI orchestration and the results are the largest single hiring cluster in Europe's data economy: 443 roles in one month, more than any other kind of data or AI position posted.¹ Somewhere in the last two years, the industry stopped asking whether to use AI and started admitting, at scale, that using it isn't going the way anyone planned.
This matters because I've watched a specific pattern repeat across nearly every mid-sized maison I advise. Someone runs a pilot: a chatbot, a forecasting tool, a generative-imagery experiment. The pilot technically works. Eighteen months later, almost nobody in the business actually uses it, and the tool sits quietly renewing its subscription. That gap, between a pilot that works and a business that has changed how it operates, is now visible in hiring data at industrial scale, not only in the handful of companies I see directly.
The vocabulary tells you where the problem moved
Not one posting in my seven-market sample this year asked for an AI readiness assessment, the kind of engagement that used to open almost every AI conversation two or three years back.¹
What companies are hiring for now is adoption, enablement, orchestration and process excellence: roles whose entire job description is getting a tool that already exists to actually change how people work.
UpSlide is hiring a Head of AI Enablement. Vonovia's opening reads Lead Process Excellence & AI Transformation. Raiffeisen is hiring for Business Processes & AI Transformation, not for an AI strategy. None of these are companies still deciding whether AI belongs in the business. They already bought it, and are now hiring someone to make people use it.
Why the pilot working was never the hard part
A pilot succeeding is a narrow, controlled test: does the model produce a usable answer. Adoption is a far wider claim: does a person, under deadline pressure, with an existing habit that already works well enough, choose the new tool over the old way of doing things. Those are different problems with different owners, and most mid-sized organisations resource the first one generously, a vendor contract, a proof of concept, a few weeks of an analyst's time, and the second one barely at all.
MIT's 2025 study of enterprise generative AI found 95% of organisations see zero measurable P&L impact from their pilots, and traced the failure to what its authors called a learning gap: not the model, but the absence of workflow integration, ownership and feedback loops around it.² Europe's 443-posting adoption cluster is the hiring market finally pricing in that second problem, roughly two years after the first one got funded.
If your organisation is weighing another AI pilot, the readiness question worth asking isn't whether the technology can do the job. It almost certainly can. It's who owns the unglamorous work of getting a busy commercial team to actually change a habit, and whether that person has both the authority and the time to do it. Skip that question and you don't get a failed pilot. You get a successful one nobody uses, quietly renewing every month.
If you're wondering why a tool that tested well in the pilot still isn't being used six months later, that's usually where the real conversation starts. I'm always glad to compare notes.
Elisabeth
References
1. Maison Virgilio, proprietary market research: live job-posting analysis across seven European markets, September 2026 (unpublished).
2. MIT NANDA, “The GenAI Divide: State of AI in Business 2025,” as reported by Fortune. https://finance.yahoo.com/news/mit-report-95-generative-ai-105412686.html
