Toku’s Middle East Expansion Shows Why Enterprise AI Must Adapt to Local Markets

With Talabat live across eight markets, a new UAE entity and an Arabic AI partnership with SESTEK, Toku’s expansion highlights how language, connectivity, data residency and regulation are becoming central to scaling enterprise AI.

Toku Ltd. has expanded its Middle East footprint from two to eight markets since December 2025 and established its first in-country entity in the region, as the Singapore-based AI customer experience company moves to deepen its presence across the Gulf. The company’s new UAE subsidiary will allow it to contract, invoice, hire locally and participate in government and semi-government tenders across the UAE and wider GCC. At the same time, Toku is working with conversational AI provider SESTEK to add native Gulf-dialect speech capabilities and locally hosted deployment options.

The expansion comes as enterprise AI adoption accelerates across the Gulf but remains difficult to scale. A 2025 McKinsey survey of 139 senior executives and board directors across the six GCC countries found that 84% of respondents said their organizations had adopted AI, up from 62% in 2023. Yet only 31% said their organizations had scaled AI or fully deployed it across the organization.

Toku’s Middle East strategy offers a view into what that next stage of deployment may require. Its Talabat deployment is now live across Bahrain, Egypt, Iraq, Jordan, Kuwait, Oman, Qatar and the UAE, supporting more than 8,000 users across in-house and BPO teams. After voice was added alongside chat, Talabat’s customer satisfaction rose to approximately 80%, a measured uplift of more than 20%.

The development also points to a broader shift in enterprise AI. As deployments move beyond pilots and into operational environments, localization increasingly involves more than language. Connectivity, data residency, infrastructure, regulatory requirements and local procurement are becoming part of how AI systems are designed and deployed.

Local Presence Is Becoming Part of AI Expansion

Toku incorporated its wholly owned UAE subsidiary on August 24, 2026. The entity enables the company to contract, invoice, receive payments and hire locally, while also allowing it to participate directly in government and semi-government tenders across the UAE and wider GCC. Toku said it is already participating in several tenders alongside a regional partner. The move reflects an important distinction between entering a market and building an enterprise technology business in that market. For large organizations and government customers, procurement, local contracting, data handling and infrastructure requirements can be as important as the underlying AI capabilities.

Thomas Laboulle, Founder and CEO of Toku, explained that the company’s experience across Asia shaped this approach while conversing with AsiaTechDaily:

“Talabat is now live across all eight of its markets, Bahrain, Egypt, Iraq, Jordan, Kuwait, Oman, Qatar and the UAE, supporting more than 8,000 users across their in-house and BPO teams. Since we introduced voice alongside chat, customer satisfaction (CSAT) has risen to approximately 80 percent, a measured uplift of more than 20 percent. That footprint is broader than the GCC. It spans the wider Middle East, and the first lesson is that Arabic is not one deployment. Those eight markets span four dialect families, Egyptian, Iraqi, Levantine and Gulf, and in much of the Gulf most residents are expatriates, so a single call can move between Arabic, English and a third language. Language is configured market by market, and the system has to keep up mid-sentence.

The second is that the telecom layer is where multi-market deployments fail first, and it never shows in a demo. Numbering, caller identification, recording rules and carrier quality differ in every market. Talabat runs on bring-your-own-carrier connectivity, so we integrated carriers we did not choose under eight regulatory regimes and made them behave as one. Poor audio makes poor transcription, and poor transcription makes a poor agent, human or artificial. The third is about channels. Voice came back because a late order at nine in the evening is a synchronous problem that a queue of chat messages does not solve. Channel strategy should follow the customer’s moment, not the vendor’s cost model, and AI makes that affordable again by bringing the cost of a call down.

None of this was learned in the Gulf. The playbook was built in Southeast Asia, tested across 15 countries in Latin America and applied again with Glovo in Europe; the Gulf is where we proved it travels, taking our Middle East footprint from two markets to eight in about seven months. The point of a playbook is that each new market is a configuration not a rebuild. For GCC customers (Bahrain, Kuwait, Oman, Qatar, Saudi Arabia and the UAE) specifically, where compliance and data residency expectations tend to be more stringent, it means mapping each market’s regulatory, connectivity and language requirements before any implementation work, and designing the hand-off between AI and human agents before the first call. It also means contracting locally, through our UAE entity.”

The Talabat deployment is significant because it provides a practical test of this model across different regulatory, infrastructure and linguistic environments. Toku’s contact center is integrated with Talabat’s CRM environments and manages inbound and outbound service across Bahrain, Egypt, Iraq, Jordan, Kuwait, Oman, Qatar and the UAE. The company says adding voice alongside chat lifted CSAT to approximately 80%, representing a measured increase of more than 20%.

Language is emerging as another critical layer of enterprise AI deployment. Toku’s June 2026 memorandum of understanding with SESTEK is intended to integrate SESTEK’s Agentic CX Suite into Toku’s offering. The partnership brings native Gulf Arabic speech capabilities, including support for dialect variation and multilingual code-switching, while Toku provides locally hosted deployment options.

This matters because voice AI performance is closely connected to the environment in which conversations occur. In the GCC, customers may move between Arabic and English or other languages during the same interaction, while dialects vary considerably across markets.

The significance of Toku’s Middle East strategy extends beyond customer service. For enterprise AI providers, data residency increasingly affects where models run, where customer interactions are processed and stored, and how systems connect to telecommunications and enterprise infrastructure. Toku’s own announcement positions locally hosted, sovereign and hybrid infrastructure as increasingly relevant to government and enterprise customers in the region.

That creates a different model of localization. Instead of adapting a globally standardized AI product after entering a market, providers increasingly have to design deployments around local requirements from the beginning. The same logic is visible in Toku’s partnership strategy. Rather than attempting to build every language capability internally, the company is integrating a specialist Arabic-language provider behind its own orchestration layer. This approach reflects a broader direction for enterprise AI: global platforms may provide the orchestration, infrastructure and workflow layer, while local specialists contribute language models, domain knowledge, connectivity or compliance capabilities.

The Enterprise AI Market Is Moving From Availability to Fit

Toku’s Middle East expansion does not by itself establish that localized infrastructure or language capabilities will determine every enterprise AI purchase. But its deployment across eight markets illustrates the growing number of variables that providers have to manage when moving from an AI demonstration to a production system. The company’s experience also highlights why adoption statistics alone can obscure the harder part of enterprise AI. The GCC is reporting high levels of AI adoption, but the smaller share of organizations that have fully scaled AI suggests that implementation, integration and operational fit remain significant challenges.

For providers operating across fragmented markets, therefore, the competitive question is increasingly not whether an AI system can perform a task. It is whether that system can perform the task under the language, infrastructure, regulatory and data requirements of the market where the customer actually operates.

Toku’s Middle East strategy is built around that premise: local entities, local partners, localized language capabilities and deployment flexibility are becoming part of the product itself. As enterprise AI moves deeper into regulated and operationally critical environments, that distinction could become increasingly important.

Quick Takeaways

• Toku has expanded from two to eight Middle East markets since December 2025, with Talabat now live across all eight markets.
• Its new UAE subsidiary strengthens local market access, enabling Toku to contract, invoice, hire and participate in government and semi-government tenders across the GCC.
• Talabat provides early commercial validation, with Toku supporting more than 8,000 users and CSAT reaching approximately 80% after voice was added alongside chat.
• Arabic AI is becoming more specialized, with Toku partnering with SESTEK for Gulf-dialect speech and multilingual code-switching capabilities.
• Data residency and deployment flexibility are moving into the core AI proposition, particularly for government, banking and telecom customers. (
• The broader GCC market shows a scaling gap: AI adoption reached 84% in 2025, but only 31% of surveyed organizations reported scaling or fully deploying AI across operations. (Toku)