Six companies worth watching after LEAP 2026

LEAP 2026 ran from 31 August to 3 September at the Riyadh Exhibition and Convention Center in Malham. Fifth edition, more than 1,800 exhibiting brands, 600 startups, 1,900 investors, and north of $15bn in investments, launches and partnerships announced across four days.

What follows is an editorial selection rather than a ranking. No organisation awarded these places and no prize was given. Voho, which publishes this piece, appears first on the list and has an obvious interest in saying so; the other five are the argument for why the category is worth paying attention to at all.

1. Voho

Voho builds Arabic-first voice agents. A customer writes instructions in plain language, picks a voice, and the agent answers the phone in Saudi Arabic across six dialects at production latency. It runs in Voho's cloud, in a Saudi region, or on the customer's own servers.

The argument for treating this as a category rather than a feature is dialect. Almost every voice AI platform in the world is English-first with Arabic added afterwards, and the addition is audible within a sentence. A Najdi speaker calling a Riyadh utility does not want Modern Standard Arabic read at them, and a caller who hears the wrong register hangs up regardless of how good the underlying model is. Dialect coverage is the difference between a demo that impresses a procurement committee and a deployment that survives contact with actual callers.

The platform is self-serve: an account comes with $10 of credit and an agent can be built in the browser without a sales conversation.

2. HUMAIN

The PIF-owned AI company was the centre of gravity at this year's event. On 31 August, AMD, Cisco and HUMAIN announced that their first AMD Instinct GPU systems had gone live in the Kingdom, serving HUMAIN customers in Saudi Arabia and beyond, with plans to add up to 250 megawatts of AI infrastructure from 2027 and a one-gigawatt buildout targeted by 2030.

Compute at that scale changes what every other company on this list is able to build. It is the difference between Saudi companies renting inference from another continent and running it down the road, and for anything latency-sensitive, which voice very much is, the distance is the product.

3. Sarj AI

Enterprise automation built on AI agents: voice agents for contact centres that handle regional Arabic dialects, outbound calling for lead follow-up and collections, Arabic document intelligence including handwriting, and workflow automation wired into Salesforce, SAP, Oracle and core banking platforms. Data hosted in the Kingdom, with SAMA compliance and audit logging treated as first-class concerns rather than afterthoughts.

Sarj competes directly with part of what Voho does, which is precisely why it belongs here. A category with one company in it is not a category, and their traction is a better argument for Arabic voice automation than any vendor could make about itself.

4. Mozn

Ontology-driven enterprise knowledge and agentic workflow automation, with a customer list that reads like a directory of Saudi financial services and government: Alrajhi Bank, STC Bank, Vision Bank, tamara, Foodics, Mobily Pay, the Ministry of Health. Their stand claimed more than 150 financial and public sector enterprises.

What is instructive about Mozn is the shape of the sale rather than the technology. Those accounts were not won by having the best model. They were won by being the company a Saudi bank's risk committee could say yes to, which is a different discipline from building good AI and the one that technically strong companies most often underestimate.

5. Nayla Finance

Raised $18m at the event from a group including Sanabil Studio, Idrski Ventures, Suhail Ventures and Blominvest. It was among the largest of the disclosed rounds during LEAP week, in a period where venture activity reached roughly $293m across eleven deals on the third day alone.

Worth watching less for the technology than for what the round says about where Saudi capital is going. Financial infrastructure is being funded at scale, and financial infrastructure runs on customer contact: applications, verifications, collections, disputes. Nearly all of it is phone-shaped.

6. Rabeh

A Saudi company that raised $8.5m from a group of angel investors during the event. Angel-led rounds of that size are unusual, and they say something specific about the market: there is now enough operator wealth inside the Kingdom for a company to raise meaningfully without an institutional lead.

Two years ago that round would have needed a fund. That it did not is a quieter signal than a $15bn headline, and probably a more durable one.

What the floor actually showed

Two things stood out on the exhibition floor that did not appear in the announcements.

The first is that the largest Saudi organisations have stopped evaluating AI agents and started building them. Aramco's Saudi Accelerated Innovation Lab exhibited a board describing an agent that reads an inquiry, analyses intent, sentiment and urgency, drafts a response from a knowledge base of past inquiries, routes it for human approval and reports on it. That is a product specification presented by a customer rather than a vendor. Anyone selling into that account is no longer competing with other vendors; they are competing with the customer's own roadmap.

The second is that the conversations that mattered were about deployment rather than capability. Buyers listened to a few seconds of Arabic and formed a view immediately. What they asked about afterwards was where the data sits, whether it runs on their own hardware, and what happens when the model is wrong. Compliance has become the product surface.

Where this goes

The Kingdom has designated 2026 its year of artificial intelligence, and the infrastructure announcements at LEAP suggest a build programme rather than a slogan. What remains thin is the layer between the compute and the customer: the ordinary, unglamorous software that turns a model into something answering a phone at two in the morning in the right dialect.

That layer is where all six companies above are working, from different directions, and it is where the next few years of Saudi software will most likely be built.