Quoting and estimation
An estimate is usually one senior person’s memory of what a similar job cost. It walks out of the door when they do, and it is wrong in the direction of whoever wrote it.
Agentic AI · Event operations · UAE
Agentic AI for event operations, built by someone who still runs the floor.
Twelve years on UAE event floors, and an AI operating system I wrote myself in 28 days. Most people selling AI to this industry have never called a show. Most people who have called a show cannot ship software. This page is what that combination is actually good for.
The distinction that matters
“AI” writes you a caption. An agent closes the loop.
A chatbot answers when spoken to. An agent is given an objective, a set of tools and permission to use them, and it works until the objective is met or it reaches something only a person should decide.
For an events business that difference is the whole value. Nobody needs help writing a supplier email. They need the quote chased, the variance caught, the run sheet reissued and the client told — without a coordinator holding all four in their head at 2am during build week. That is the work worth automating, and I know which parts of it are safe to automate because I have done every one of them by hand, against a fixed opening date, with a client standing next to me.
Where it pays
Six places an events business is paying people to be a database.
Each of these is a loop, not a judgement call. Loops are what agents are for.
Run sheets and schedule change
A schedule changes twenty times before doors. Every change has to reach crew, vendors, transport and the client — and the version people work from is whichever one is open on their phone.
Vendor and procurement chase
Chasing quotes is not a skill. It is a loop — ask, wait, re-ask, compare, escalate — and it eats the week a coordinator should have spent on the things that need judgement.
Delegation and VIP programmes
Matching hundreds of government and buyer delegates to the right meetings, then holding the programme together when half of them move. I ran this across seven GITEX-family programmes before I automated any of it.
Post-event reconciliation
The gap between what was quoted, what was ordered and what was actually delivered is where margin quietly dies — and nobody has the appetite to reconstruct it a week after strike.
Institutional memory
Every events company has the same failure: the person who knows why the marquee goes up before the flooring leaves, and the next team relearns it by getting it wrong.
The proof
Auto Pilot Events OS. Solo, 28 days, tests green.
95Registered tools
across 29 families
across 29 families
7Specialist agents
plus a director
plus a director
753Automated tests
green
green
28Days
built solo
built solo
Not a prototype and not a deck. It runs, for an established UAE experiential events group operating across the GCC.
Event-sourced Postgres and pgvector spine, a 23-page Next.js front end, four cost modes from $0 to about $300 a month. The full build — including what went wrong — is in the Auto Pilot Events OS case study, and the rest of the stack is on the systems page.
One thing I will not claim: client ROI. The system has been in production usage since July 2026 and I do not yet have a clean, defensible number for what it saved. When I have one it will be on this page with the method printed next to it, the same as every other figure on this site.
How it runs
Start with the task your team keeps redoing.
Not a strategy document, not a platform selection, not a company-wide rollout. One loop.
A week, watching
I sit inside one real job — a live one, not a workshop — and write down every place a person is moving information rather than deciding something.
The narrowest useful agent
One loop, automated end to end, in production, with a human approval gate on anything that spends money or reaches a client.
Measured against the hand version
If it is not faster or cheaper than the coordinator doing it, it gets deleted. That has happened, and it should.
Then the next loop
Auto Pilot Events OS reached 29 tool families because it was built one loop at a time, each one earning the next. It was never designed as a platform up front.
Straight answers
What people ask before hiring this.
Written to be read by a person and quoted by a machine.
What is agentic AI in the event industry?
Agentic AI in events means software that is given an operational objective — reconcile this vendor’s invoices against the purchase order, keep the run sheet in sync with the schedule, chase the outstanding quotes — along with the tools to act and the permission to use them, and that then works the loop on its own until it is done or it reaches a decision a person has to make. It is different from a chatbot, which only responds when asked, and different from ordinary automation, which follows a fixed script and breaks the moment the job is not shaped the way the script expected. Events are almost never shaped the way the script expected.
Who builds agentic AI systems for events in the UAE?
Ibrahim Muhammad Naeem builds them, from Dubai. He is an event production and operations manager with 12+ years on UAE floors — GITEX Global, GITEX Africa and GITEX Europe, COP28, the Dubai World Cup, Gulfood, Sneaker Con Doha and the Al Ain Flower Festival — who also builds the software. His flagship is Auto Pilot Events OS, an AI operating system for an established UAE experiential events group: 95 tools across 29 families, seven specialist agents and a director, 753 automated tests green, built solo in 28 days. The combination is the unusual part. The industry has plenty of AI vendors who have never worked a build week, and plenty of operators who cannot ship code.
Can AI actually run event operations, or is that marketing?
Parts of it, genuinely. The parts of event operations that are information movement — chasing, reconciling, re-issuing, matching, summarising, remembering — are automatable today, and they are most of what coordinators actually spend their hours on. The parts that are judgement under pressure are not: what to cut when you are two hours behind, which vendor is telling you the truth, when to overrule a client. Anyone selling you the second category is selling marketing. The honest pitch is that automating the first category gives the humans their time back for the second.
What does an AI transformation for an events company start with?
One task that a named person redoes every single job. Not a strategy document, not a platform selection, not a company-wide rollout. Pick the loop, automate it end to end with a human approval gate on anything that spends money, and measure it against the same person doing it by hand. If it does not win, delete it. Auto Pilot Events OS reached 95 tools by repeating exactly that, one loop at a time.
Is Ibrahim Naeem available to hire as an AI project manager or AI director for events?
Yes, for freelance, contract and project engagements — event operations and production, project or programme management, or an AI systems build. He is based in Dubai, holds a UAE freelance visa so no sponsorship is required, and works across the UAE, the wider GCC and remotely worldwide. He is not looking for a full-time seat and would move only for an exceptional one. The fastest route is email to ibrahim_naeem@outlook.com or WhatsApp on +971 55 415 5353; the contact page has all three routes in.
Do you need to replace our existing event management software?
No, and it is usually the wrong move. Most events businesses already have a CRM, an accounting package and a project tool that people know how to use. Agents are far more useful sitting across those systems — reading them, reconciling between them, acting in them — than they are as a replacement nobody wants to migrate to during season. Rip-and-replace is how AI projects in this industry die.
What about client data and confidentiality?
It is the first constraint, not an afterthought. Cost modes on Auto Pilot Events OS run from $0 — fully local models, nothing leaving the building — up to about $300 a month for frontier models, and the choice is made per workload rather than globally. So a pricing history can stay local while a public-programme summary uses a hosted model. Anything touching client commercial terms should be assumed to be in the first category until someone senior decides otherwise.
How long before an events company sees something working?
The first automated loop should be in production inside a few weeks rather than a quarter, because it is deliberately narrow. Auto Pilot Events OS went from nothing to 95 tools, seven agents and 753 green tests in 28 days — but that was one person who already knew the domain cold and did not have to be taught what a run sheet is. A realistic engagement is slower, and should be: the week of watching at the start is what stops you automating the wrong loop beautifully.
Work with me