Breaking Boundaries: How DeepL's New Arabic Tool Aims to Lead AI Translation for MENA Businesses | Entrepreneur

In January 2025, Germany-headquartered artificial intelligence (AI)-powered translation service provider DeepL revealed a study showing that 84% of professionals in the UAE and Saudi Arabia have integrated AI translation tools into their workflows. Now, put this statistic in the context of both nations' global economic milestones -which require efficient cross-border business communications and market access- and it immediately holds added value: in 2024, the UAE's total foreign trade stood at US$1.424 trillion (AED5.23 trillion), up 49% compared to its 2021 performance; while the Kingdom's non-oil exports alone in 2024 reached $137.3 billion (SAR515 billion), marking a 13% increase from 2023. As the Chief Revenue Officer at DeepL, David Parry-Jones has been able to closely observe the growing necessity for accurate translation tools. "In this recent study conducted to understand the increasing integration of AI-driven language technologies within organizations, we also found that 46% of professionals in the UAE and KSA reported that AI translation tools have helped them expand their business into new markets," he shares. "Additionally, our findings revealed that AI-driven tools are mostly applied to developing new language skills (UAE: 52%; KSA: 51%), boosting efficiency and saving time (UAE: 50%; KSA: 51%), and managing supplier relationships (UAE: 45%; KSA: 47%)."

These, of course, are just two nations in a region that comprises over 20 Arabic-speaking countries. "Arabic is the fifth most spoken language globally and has long been one of the most requested by our users," Parry-Jones says. "The language is spoken in over 22 countries and plays an important role in global communication. However, its right-to-left script, unique characters, and structural complexity made integration somewhat more challenging. Despite this, we're now pleased to be unveiling document translation, joining our in-app and desktop translation solutions."

Indeed, DeepL's Arabic Document Translation tool -launched officially on April 30, 2025- has been designed to simplify document translation for businesses and professionals that engage with Arabic-speaking markets across MENA. "Launching Arabic within the DeepL platform was a powerful step towards breaking down language barriers and connecting the world!" Parry-Jones adds. "We believe that Language AI is one of the most strategic investments a business can make. Some of the sectors that benefit the most from Language AI are retail, manufacturing and legal, where high quality and accurate translation is vital. In retail, it increases efficiency by developing multilingual marketing assets and customer service tools, translating internal systems, and enabling seamless international expansion. In the legal sector, AI translation services help international law firms overcome language barriers. Additionally, in manufacturing, AI translation allows global manufacturing facilities to ease their supply chain and distribution with accurate translation of customs documents, product descriptions and local regulations."

^{Image source: DeepL}

Now, anyone with the slightest of linguistic interests would know that there are plenty of existing Arabic translation services already available. But the dire inaccuracies provided by these platforms have also been well documented in recent years. When DeepL's Arabic Document Translation Tool was announced in late April, it claimed to outperform GPT-4, Google, and Microsoft in translation quality– something Parry-Jones assures isn't an empty promise. "DeepL outperforms these models because our language model is purpose-built for translation, using proprietary training data collected over seven years," he explains. "Unlike general-purpose models like GPT-4, DeepL is tuned specifically for linguistic accuracy. In blind tests with language experts, DeepL's translations were preferred 1.3x more than Google's, 1.7x more than GPT-4's, and 2.3x more than Microsoft's. The model also requires significantly fewer edits, with Google needing twice as many, and GPT-4 three times more, to reach the same quality. We also rely on the expertise of thousands of hand-picked language specialists who "tutor" the model, resulting in best-in-class translation."

DeepL's decision to integrate the expertise of human translators has been pivotal in ensuring that the AI-powered machine translation platform incorporates the plethora of regional dialects within the Arabic language– a move that has, again, helped in outperforming its competitors. "Our expert team of translators are involved in many stages of our research and development (R&D) process, from building models for a new language to improving existing ones," Parry-Jones continues. "By incorporating a human element in the initial phases of our research, we avoid the risk of our translations sounding robotic and help ensure that our translations pick up cultural nuances. This also applies during the evaluation stage; while synthetic evaluations deliver quick results, the ground truth is to ask those that have invented and mastered the use of language to provide feedback on translations - how accurate they are, how nuanced, how native the language feels. Only we humans can judge on that. To this day, some of our earliest adopters and users are translators. We have recently introduced a new product called Clarify that helps customers clear up ambiguities by suggesting alternatives and asking context-specific questions, ensuring translations capture the right meaning and nuance of what you're trying to say."

But within the scope of business data translation comes the mammoth risk of security lapses or information leaks. "Our Pro customers' data is never stored or accessible to third parties!" Parry-Jones reveals. "In case of data breach, DeepL is legally required to notify users within 72 hours, guaranteeing maximum confidentiality for sensitive content like reports, patents, and customer data. As a company based in Germany, DeepL adheres to the GDPR -one of the strictest data protection and privacy laws worldwide- ensuring data remains protected and compliant. Our advanced encryption and adherences to global regulations, including ISO 27001 and GDPR standards, ensures peace of mind and enables confident multilingual communication across borders."

^{Image source: DeepL}

Offering additional ease to users is the fact that DeepL's Arabic Document Translation Tool prioritizes popular formats such as Microsoft Word, Outlook, and PowerPoint as well as PDFS, while maintaining original layouts and fonts. It also allows users to edit their changes directly before downloading, allowing for increased control across supported file types. "We only launch a new language when we are confident that we can offer a better solution than what's already on the market, and that includes Arabic," Parry-Jones adds.

This faith in the platform's performance has emanated from DeepL's company culture- and if Parry-Jones words are anything to go by, it is all set to carry the platform towards its future goals too. "Since our inception, we have been a research-driven company and will continue to invest heavily in our development as a company within the next few years to create specialized translation and writing solutions to businesses worldwide," he says. "Combining our depth of research with proprietary data accumulated over seven years, we're able to understand unique business needs and address them with our solutions. With the launch of Arabic document translation in April and with any future products we take to market, we look forward to seeing how Middle Eastern businesses are able to use Language AI to expand their business across borders thanks to seamless communication. Working with businesses in the region, we will continue to learn and iterate the offering to make sure it provides them with the tools they need to expand their business across borders."

Related: INSEAD's Associate Professor Vikas Aggarwal on How Ecosystem Partnerships Shape MENA's Future in AI