This issue is about a simple gap: the AI is arriving faster than the people to run it. The newest hiring data shows demand for AI-agent skills up more than 280% in a year, yet McKinsey finds only 39% of companies can point to any bottom-line impact from AI. In the Gulf the machine is being switched on regardless. The US just moved the UAE into its top tier for AI chips, and the UAE has put AI into its courts with a human judge still on every final call. The thread through all of it is the one TASC keeps seeing: buying the technology is the easy part. Turning it into value, safely and with the right people, is a workforce job. Here's the news, why it matters, a simple guide to putting your first AI agent to work, and a beginner's crash course on using Claude across a business team.
The clearest sign AI has crossed from experiment to real work is in the job ads: demand for people who can build and run AI agents is up more than 280% in a year.
280% — one-year jump in demand for "agentic AI" skills in US job postings (Stanford HAI AI Index 2026)
If you want to know whether AI has really moved from pilot to production, look at what companies are hiring for. Stanford's 2026 AI Index — using billions of job postings — found demand for "agentic AI" skills up more than 280% in a single year, from 0.06% of US postings in 2024 to 0.23% in 2025. That's roughly 90,000 job ads for people who can build and run agents.
The wider picture is the same. AI skills now appear in 2.5% of all US job postings, up 55% in a year and nearly 300% over a decade. And the fastest-growing asks aren't research skills — they're deployment ones: getting AI to run reliably at scale, wiring it into workflows, keeping it going. Mentions of "chatbot" and "ChatGPT" actually fell. The work has moved past the demo.
For leaders, that's the tell. The bottleneck is shifting from the technology to the people who can operate it. The smart move is to work out which AI skills you'll need over the next year and start building them now — because the whole market is hiring for the same ones.
Why it matters: The scarce resource isn't the model anymore — it's people who can put agents into real workflows and keep them running. Every leader planning an AI rollout is really planning a hiring-and-reskilling problem, whether they've noticed yet or not.
Source: Stanford HAI — The 2026 AI Index Report (Lightcast labour-market chapter)
AI is now used almost everywhere — but most companies still can't point to a single number on the bottom line to show for it.
39% — of companies can attribute any bottom-line (EBIT) impact to AI — most under 5% (McKinsey, The State of AI)
The money going into AI is staggering, and it keeps growing. The largest US cloud companies are on track to spend around $700 billion on AI in 2026, and chipmaker AMD is investing up to $5 billion in Anthropic to lock in future capacity. AI is now used, in some form, almost everywhere.
And yet the value isn't showing up. McKinsey's State of AI found only 39% of organisations can attribute any bottom-line impact to AI at all — and most of those put it below 5%. Just 23% are scaling agents in even one function, and nearly two-thirds haven't begun scaling AI across the business. Adoption is near-universal; payoff is rare.
The reason is not the technology. It's that value comes from redesigning how work is done, governing it properly, and having people who can run it — the parts that don't come in the box. That's the gap this edition's crash course is about, and it's a workforce problem before it's a technology one.
Why it matters: More spend does not mean more value. The gap between money in and results out is a people-and-process gap: value comes from redesigned workflows, governance and skilled staff, not from buying another platform. The winners this cycle are the minority who actually close it.
Source: McKinsey — The State of AI
The compute the Gulf has been waiting for is now flowing — the US moved the UAE into its top trust tier for advanced AI chips.
A:5 — top US trust tier the UAE was moved into, unlocking licence-free Nvidia/AMD AI-chip imports (US Commerce Dept / WSJ)
For two years, the Gulf's AI ambitions ran into one wall: access to the most advanced chips. That wall just came down for the UAE. The US Commerce Department reclassified the country into its highest trust tier — Country Group A:5, alongside India, South Korea and Europe, and up from a group that had included China.
In practice, that means approved UAE players like G42, and the US hyperscalers building data centres there, can now import advanced Nvidia and AMD AI chips without individual export licences. G42 had previously been capped under licence at roughly 35,000 Nvidia GB300-equivalent systems; that ceiling is effectively lifted, accelerating big projects like the Stargate UAE cluster.
The strategic point for employers: compute is about to stop being the constraint in this region. When capacity is abundant, the differentiator moves to people and governance — who can actually design, run and control AI at that scale. That's a workforce question, and it's worth answering before the racks are full.
Why it matters: Compute stops being the Gulf's bottleneck. When the chips flow freely, the thing that decides who wins isn't hardware — it's who has the people and the governance to turn all that capacity into something useful.
What to do: Assume compute-heavy AI is coming to the region fast. Plan the workforce and the governance now — line up the people who can build and run it, rather than waiting until the data centres are full to start thinking about who staffs them.
Source: DataCenterDynamics — US gov't eases AI chip export controls on UAE
The UAE launched what it calls the world's first fully integrated AI-powered judicial platform — and kept a human judge on every final decision.
1st — world's first fully integrated AI judicial platform — judgments stay under human oversight (Gulf News)
The UAE has taken AI somewhere most governments haven't dared: the courtroom. Announced by Sheikh Mansour bin Zayed Al Nahyan, the country launched what it describes as the world's first fully integrated AI-powered judicial platform — AI woven across the whole process, from analysing case files to retrieving legislation and precedents, drafting documents and shortening litigation timelines.
What makes it notable isn't the ambition; it's the guardrail. Officials were clear the system does not replace judges. It's an advanced legal assistant, and every final judgment stays under full human oversight, preserving judicial independence.
That's the pattern worth copying. The most credible AI deployments in the region pair real automation with a named human who owns the decision that matters. If the courts can run AI this way — fast, but with a person accountable — customers and regulators will expect the same of employers. Build that accountability, the audit trail and the Arabic-language quality in from day one, not after a complaint.
Why it matters: Even the courts are deploying agentic AI on a "human-accountable" model. That quietly resets the governance bar for everyone else: if the judiciary keeps a named person responsible for the final call — with an audit trail and Arabic-language rigour — customers and regulators will expect the same standard from your business.
What to do: Copy the model, not just the ambition. For any AI you deploy, keep a named human accountable for consequential decisions, build an audit trail of what the AI did and why, and get Arabic-language quality right from the start.
An AI agent is just software that takes a goal and does the steps for you, checking in when it needs a human. The easiest way to start is to let the AI design the agent for you. Here's how.
Start with one boring task, let the AI design the agent, and keep yourself in the loop. In an afternoon you'll have a working example you can build on — no coding required.
The data is blunt: AI-skill demand is surging and value is scarce, and the difference is people who can run the technology. You can't buy your way out of that — you build, borrow and reskill your way through it. Here's how to start.
The skills gap won't close by buying more technology. Reskill the people you have, own the few roles that matter, and bring in specialists flexibly for the rest — that's how you get ahead of it.
Crash Course. Everyone's talking about AI agents, but most business teams don't know where to actually start. This beginner's crash course is the answer: a plain-English guide to using Claude across a business team, with no coding required. What it is and why it helps, how to set it up on your team's desktops, and two use cases you can build in a day — including a 'brand voice' skill that makes everything Claude writes sound like your company. Built for managers and teams in UAE and KSA who want to move from reading about AI to using it this week.
Read the crash course »
AIQU. Want agents doing real work, not just demos? AIQU is TASC Group's technology division. It designs, builds, integrates and operates agentic AI across the systems you already run — with 100+ pre-built agents and a live pilot in 4 to 8 weeks. Governed, audited and human-in-the-loop, built for regulated GCC enterprises.