What 265 Registrations Says About Dubai
Vishal Sachar
Co-Founder & CEO of CLRT
On the last Friday of July we published a Luma page for AI After Hours, a four-part Thursday evening series we are running with TheBlock in Dubai, built for people who use AI every day and have never built with it. No frontier demos, no panel of futurists, a beginners' room by design. By Monday the page had passed two hundred registrations and sat as the most popular August event on Luma Dubai. By the eve of the first session it stood at 265. We had expected to work for every seat. Instead the market told us something before the first session ran, and it is worth reading carefully, because it contradicts the story most organisations in this city are telling themselves about where they are with AI.
Look at what filled, because the shape of the demand matters more than the size. Dubai does not lack AI events. Any given week offers keynotes, summits, panels on the future of everything, and rooms full of people describing what agents will eventually do. The event that outdrew them in August promises none of that. It promises a progression, from chatting with AI to delegating real work to it, then to automation and agents, across four Thursday evenings, for people who would call themselves beginners. Two hundred and sixty-five professionals did not register in a week out of curiosity. Curiosity was satisfied two years ago; everyone has already talked to a model. They registered because daily use has quietly failed to turn into capability, and they know it. A beginners' room filling this fast is not an appetite signal. It is a confession.
Now set that against how organisations read their own adoption. The standard evidence is usage: staff draft emails with AI, summarise documents, rewrite proposals, and the survey comes back saying the workforce has adopted. Leadership files AI under done and moves to the next initiative. But usage is the first stop on a line that runs from chat to delegate to automate to agent, and nearly the entire population is standing at the first stop. A person who chats with a model still does the work; the model just makes the sentences arrive faster. Nothing structural has changed about who carries the task, who checks it, or what the organisation can now do that it could not do before. An organisation whose whole workforce chats with AI has adopted a faster way of writing. It has not adopted AI in any sense that shows up in how work moves.
The funnel we watch from the workshop side runs register, attend, build, ship, and every stage sheds people. Registering is a click. Attending costs an evening. Building costs something different in kind: you must choose a task the tool can actually carry, bound what it may touch, and decide in advance what a wrong output looks like so you can catch one. Shipping, the point where a workflow runs inside a business without its owner rereading every word, costs judgment that nothing in the earlier stages taught. People do not fall out of this funnel because AI is too hard. They fall out because the market only teaches the two ends. Prompting content is free and everywhere, precisely because it is worth nothing. Frontier engineering has its conference circuit. The middle, the passage from user to builder, is where the value concentrates and where the teaching stops.
The encouraging half of the diagnostic is that the gap is teachable, because it is mostly not technical. Moving from chat to delegation does not require a single line of code. It requires a different set of questions: which of my tasks is bounded enough to hand over, how would I verify the result without redoing it, what is the cost of the failure I will not catch. Those are judgment questions, and a room can be walked through them on real work in an evening. One stage of movement is worth more than it looks. An employee who can genuinely delegate a verifiable task to AI is worth more to a business than ten who prompt fluently, because delegation is the first point at which the organisation, rather than the individual, gets faster. That is the stage those registrants are reaching for, whether they would phrase it that way or not.
If you run a company in this city, the room is a mirror. The people who registered are the same population your adoption statistics count as proof that AI is handled: professionals in marketing, finance, HR, sales and consulting who use a model daily. Their registrations say they do not believe their own fluency, and your workforce is not different. The question worth asking is no longer whether your people use AI. It is how many of them can hand real work to it and defend the result, and for most organisations the honest answer is close to none. The same funnel applies one level up: companies also sit at the chat stage, running pilots that draft and summarise while the workflows that would move a P&L stay untouched. The constraint, at both scales, is not appetite and not tooling. It is the untaught middle.
A beginners' room filling this fast is not an appetite signal. It is a confession that daily use has not become capability.
A deeper dive
The reason the middle of the funnel is undersupplied is structural, not accidental. Prompting can be taught to a million people with one video, because chat is generic: the same tricks work on any task, so the content scales and the price falls to zero. Delegation is specific. To move one person from user to builder, someone has to look at their actual work and make calls that do not generalise: which of these tasks is bounded, what would verification even mean here, where is a silent failure expensive, what must this tool never be allowed to touch. That is diagnosis, and diagnosis does not scale like content, so the market routes around it. The internet will teach your staff to prompt for free, forever, and it will never teach them to delegate, because delegation cannot be taught without looking at the work. This is also why the gap is where the value sits. When a capability is both scarce and unteachable at scale, whoever supplies the judgment captures the difference, and the difference here is the entire distance between a workforce that writes faster and a workforce that moves work.
There is a specifically local reading too. Dubai's AI story is usually told from the top down: compute, national strategies, chief AI officers, mandates with dates attached, and that story is real. What a registration page shows is the bottom of the same stack, a large population of working professionals trying to close their own gap, on their own time, on a weekday evening, at an event that promises them work rather than spectacle. That combination, mandate above and appetite below, is rarer than either alone, and it means the scarce layer in this market is neither ambition nor demand but conversion: the machinery that turns daily users into people who can delegate, verify and eventually govern. The organisations that build that layer first will draw on a workforce the statistics claim everyone already has and almost no one actually does. The ones that keep reading usage as adoption will keep running pilots at the chat stage and wondering why nothing in the P&L moves.
Work with CLRT
Teaching a room is one job. Moving an organisation through the same funnel is another, and it is the one CLRT does: finding the tasks in your business that are actually worth delegating, then building the verification and governance that make delegating them safe. If you want to know where your organisation truly sits on the line before deciding what to do about it, the CLRT Ascent diagnostic at ascent.clrtstudio.com is where we start. And if you are in Dubai on a Thursday in August, the room is open.
Vishal Sachar
Vishal Sachar is the Co-Founder and CEO of CLRT, where he helps UAE businesses make sense of applied agentic AI and put it to work. He writes on agentic systems, AI governance, and the economics of automation. Reach him at vishal@clrtstudio.com or on LinkedIn.
