Dubai AI Festival

Attendees, exhibitors & solutions

Who this event is for

Industry trends

Key trends shaping this market and its wider industry.

Arrival airports

Choose from the following airports, with convenient access to the event and venue.

Hotel tiers

A range of accommodation options near the venue and across key areas.

Local transport

Get around the city with these convenient transport options.

Frequently asked

Glossary

1

Agentic AI

AI systems that plan and execute multi-step tasks with limited human prompting, typically by calling tools, APIs, and internal systems. At an applied AI show in Dubai this is the layer most enterprise buyers want to see demonstrated against real workflows, not slides.

2

Foundation model

A large pre-trained model (text, image, audio, or multimodal) that is adapted to many downstream tasks. Exhibitors usually present either the model itself, the tooling around it, or a vertical application built on top of one.

3

Fine-tuning

Further training of an existing model on domain data so it performs better on a specific task or vocabulary, such as Arabic-language customer service or financial document review. Distinct from training a model from scratch.

4

Retrieval-augmented generation (RAG)

An architecture where a model retrieves relevant documents from a company knowledge base before answering, reducing hallucination and keeping answers grounded in current data. Common buying criterion for enterprise and government deployments.

5

Inference

Running a trained model to produce output. Inference cost, latency, and where it runs (cloud, on-premise, or edge) are the practical questions buyers ask at AI trade events, because they drive the operating budget.

6

Multimodal model

A model that accepts and produces more than one data type, for example text plus images, audio, or video. Relevant for media, surveillance, healthcare imaging, and document-heavy public sector work.

7

AI accelerator / GPU cluster

Specialised silicon and the compute clusters built from it that train and serve large models. Infrastructure providers exhibit racks, cooling, and cluster-management tooling alongside the chips themselves.

8

MLOps

The engineering discipline of deploying, monitoring, versioning, and retraining models in production. Buyers use MLOps platforms to keep AI systems auditable and to avoid drift after go-live.

9

AI governance

Policies, controls, and tooling for model risk, data lineage, bias testing, and audit trails. A core procurement gate for regulated sectors such as banking, insurance, and government in the UAE and wider GCC.

10

Sovereign AI

Building national or institutional AI capability on locally controlled compute, data, and models. It is a recurring theme in Gulf AI programmes and shapes which vendors present local hosting and data-residency options.

11

Hallucination

A confident but incorrect model output. Vendors address it with grounding, retrieval, guardrails, and human review steps, and buyers usually test for it during live demonstrations.

12

Edge AI

Running models on local devices or gateways rather than in a central data centre, used where latency, bandwidth, or data-residency rules rule out cloud inference, for example in ports, retail, and industrial sites.

13

Digital twin

A simulation model of a physical asset, building, or process fed by live data. AI-driven digital twins appear in smart-city, logistics, and energy projects across the UAE.

14

Proof of concept (PoC)

A scoped pilot that proves a model or platform works on a buyer's data before a full contract. Most AI procurement conversations at the show start with agreeing PoC success criteria.