Field notes on Arabic voice AI.
What we learn running voice operations for Saudi and Gulf teams, deployment sequencing, dialect choices, the metrics that matter, and the compliance questions that come up in every review.
Three open Saudi Arabic models, and every number behind them
We trained and published a Saudi speech recogniser, a Najdi chat model and a formal-to-spoken Saudi rewriter, plus the dialogue set underneath them. Word error rate 104% to 39%. Gulf replies 62% to 90%. Total training spend, twelve dollars fifty. Every figure, including the run that failed.
An open Najdi dataset for Saudi enterprise Arabic, because none existed
Open Arabic dialect data is full of daily life and travel and contains nothing that sounds like a refinery shift handover or a SADAD dispute. We built 13,156 gated Najdi dialogues across eight industries and released them under Apache 2.0. What is in it, how it was gated, and where the honest limits are.
What twelve dollars and one GPU buys: training a dialect model
Two full training runs on a single L4, about twelve dollars fifty all in, and one of the two was a failure worth paying for. The method we use for every Voho model: baseline everything, gate rather than assert, evaluate on a corpus you did not write, and read the samples even when the metric says you won.
AI in oil and gas: what survives contact with a Gulf operation
Most AI pilots in upstream and downstream die between the proof of concept and the control room. What separates the ones that reach shift handover: read-only access to the OT layer, evaluation against your own past failures, and agents scoped to one discipline instead of one chatbot for the whole plant.
AI in Qatar: Qatari Arabic, the PDPPL, and who is really on the line
Qatar passed the GCC's first general data privacy law in 2016 and has been building Arabic AI in public since. What that means for anyone deploying voice agents or assistants in Doha: the dialect question, the PDPPL and QFC split, and the fact that most people who call a Qatari service line are not Qatari.
AI voice agents in Kuwait: dialect is the easy part, regulation is not
Kuwait has no single omnibus data protection law, which sounds like fewer obstacles and is usually the opposite. What a bank, telco or ministry in Kuwait City should test before buying an Arabic voice agent: the Kuwaiti register, CITRA and Central Bank rules, and the expatriate majority on the line.
Locally hosted computer use in Saudi Arabia: the agent that clicks should run on your own servers
Computer use agents work by screenshotting your screen and sending it to a model. In Saudi Arabia that makes the screenshot the compliance question, not the feature. What cloud computer use actually transmits, why an ephemeral session is not a residency guarantee, and what on-premise or in-Kingdom computer use has to prove before you buy it.
AI call centres in Saudi Arabia: what actually works on a Saudi line
Why AI call centres that demo well in English fail on real Saudi calls, and the seven things that decide whether one survives: dialect, code-switching, Arabic names and numbers, latency from the Kingdom, telephony, data residency, and who owns quality after launch.
What an AI call centre actually costs in Saudi Arabia
A per-minute price is not a business case. How to model the real cost of an AI call centre in the Kingdom: the four line items vendors leave out, why containment is the denominator, and how to build a number your CFO will accept.
Why we send an engineer to sit in your contact centre
Most AI call centre projects fail after handover, not before it. Why Voho puts forward deployed engineers inside the customer's operation, what they actually do week by week, and when you do not need one.
Six companies worth watching after LEAP 2026
LEAP 2026 announced more than $15bn of AI investment in four days. Six companies whose work stood out on the floor, and what the exhibition revealed about where Saudi software is heading.
Why Saudi Arabia is positioning itself as the world’s third AI power
The United States and China still set the pace. But Saudi Arabia is building the capital, compute, energy, Arabic models and public-sector demand to make a serious top-three case.
Voho and Archibus: an AI chatbot and call centre for service and work requests
Voho puts an Arabic voice agent and chatbot in front of Archibus, so service requests and work requests are raised, classified, tracked and closed without a helpdesk agent re-keying anything.
Voho وArchibus: مساعد محادثة ومركز اتصال ذكي لطلبات الخدمة والعمل
يضع Voho وكيلًا صوتيًا ومساعد محادثة بالعربية أمام Archibus، فتُرفع طلبات الخدمة وطلبات العمل وتُصنَّف وتُتابَع وتُغلق دون أن يعيد موظف الدعم إدخال أي شيء.
Running an Arabic LLM on your own infrastructure in Saudi Arabia
On-premise and air-gapped Arabic LLM deployment for the Kingdom, what offline actually means, how to size the model, which open weights are worth self-hosting, and how to adapt one to Gulf and Najdi dialect plus legal, financial and government vocabulary.
تشغيل نموذج لغوي عربي على بنيتك التحتية داخل السعودية
النشر داخل المنشأة والبيئات المعزولة للنماذج اللغوية العربية في المملكة، ماذا تعني «دون اتصال» فعليًا، وكيف تحدد حجم النموذج، وأي الأوزان المفتوحة تستحق الاستضافة الذاتية، وكيف تكيّفه على اللهجة الخليجية والنجدية والمفردات القانونية والمالية والحكومية.
ALLaM explained: Saudi Arabia's national Arabic LLM, and where it fits
What ALLaM is, how SDAIA built it, where you can actually run it, how it compares with Jais, Falcon Arabic and the frontier models, and how to decide whether it belongs in your Saudi deployment.
علّام: النموذج اللغوي العربي الوطني السعودي وأين يقع من مشروعك
ما هو نموذج علّام، وكيف بنته سدايا، وأين يمكن تشغيله فعليًا، وكيف يقارَن بجيس وفالكون العربية والنماذج الحدودية، وكيف تقرر ما إذا كان مناسبًا لتشغيلك في السعودية.
Saudi Arabia's National Strategy for Data & AI: what it means for your operations
SDAIA's National Strategy for Data & AI sets the Kingdom's ambition to join the elite league of data-driven economies. Here is what its six pillars and 2030 targets actually require of an organisation deploying AI in Saudi Arabia.
الاستراتيجية الوطنية للبيانات والذكاء الاصطناعي: ماذا تعني لعملياتك
تضع الاستراتيجية الوطنية للبيانات والذكاء الاصطناعي من سدايا طموح المملكة للانضمام إلى نخبة الاقتصادات القائمة على البيانات. إليك ما تتطلبه ركائزها الست ومستهدفات ٢٠٣٠ من أي جهة تنشر الذكاء الاصطناعي في السعودية.
أفضل وكلاء الصوت بالذكاء الاصطناعي في السعودية: دليل ٢٠٢٦
كيف تقيّم وكلاء الصوت بالذكاء الاصطناعي للسوق السعودي، تغطية اللهجات، زمن الاستجابة داخل المملكة، نظام حماية البيانات الشخصية، الاتصالات، والأسئلة التي تفصل العرض التجريبي عن التشغيل الفعلي.
Best voice AI agents for Saudi Arabia: a 2026 buyer's guide
How to evaluate voice AI agents for the Saudi market: dialect coverage, latency from the Kingdom, PDPL and data residency, telephony, and the questions that separate a demo from a deployment.
أفضل تحويل الكلام إلى نص بالعربية للهجات السعودية
إطار مقارنة صادق لتحويل الكلام إلى نص بالعربية في السعودية، دقة اللهجة والتبديل بين اللغتين، صوت الهاتف، القاموس المخصص للأسماء السعودية، زمن البث، وكيف تبني معيارك على مكالماتك أنت.
Best Arabic speech-to-text for Saudi dialects: how to choose
An honest comparison framework for Arabic STT in Saudi Arabia: dialect and code-switching accuracy, telephony audio, custom vocabulary for Saudi names, streaming latency, and how to benchmark on your own calls.
أفضل تحويل النص إلى كلام بالعربية للسوق السعودي
ما الذي يفصل صوتًا عربيًا يبدو محليًا في الرياض عن صوت يشبه نشرة الأخبار، التشكيل، الأرقام والتواريخ، نبرة اللهجة، زمن البث، وموافقة استنساخ الصوت.
Best Arabic text-to-speech for Saudi Arabia: choosing a voice
What separates an Arabic TTS voice that sounds native in Riyadh from one that sounds like a newsreader: diacritization, numbers and dates, dialect prosody, streaming latency, and voice cloning consent.
أفضل نماذج اللغة الكبيرة للعربية والتشغيل في السعودية
النماذج الحدودية، والنماذج الإقليمية العربية أولًا، والأوزان المفتوحة، كيف تختار النموذج خلف تشغيل صوتي أو نصي سعودي على أساس اللهجة والكمون واستدعاء الأدوات والاستضافة والتكلفة.
Best LLMs for Arabic and Saudi deployments
Frontier models, Arabic-first regional models, and open weights: how to choose the model behind a Saudi voice or chat deployment on dialect, latency, tool calling, hosting and cost.
What an enterprise Arabic voice rollout actually looks like
The sequence large organisations follow when they move voice operations onto AI, from first scoped intent to full production traffic.
كيف تبني مركز اتصالات ذكي بالعربية
دليل عملي لإنشاء تجربة صوتية عربية تضيف الثقة وتزيد التحويلات في السعودية.
Five metrics that tell you whether your voice agent is working
Containment, escalation quality, time to resolution, repeat contact rate, and post-call sentiment, and how each one misleads you on its own.
أفضل لهجات عربية للرد الآلي في السعودية
اختر نبرة الصوت الأمثل بين الخليجية، النجدية، والفصحى لجذب العملاء المحليين.
Voice AI and Saudi PDPL: what compliance teams ask for
The questions that come up in every Saudi enterprise security review, and how to have the answers ready before the review starts.
SEO صوتي: جذب عملاء السعودية عبر البحث المحلي
اعرف كيف يمكن لمحتواك العربي وصوتك الذكي أن يظهر في نتائج البحث الصحيحة.
Start your AI transformation today.
Sign up and build your first agent in the browser, or book a call if you would rather someone walked you through it. Most people do not need the call.
Start
$10 of credit, free
Granted when you sign up, about 140 minutes of live calls. No card to begin.
Then
$0.07 a connected minute
Pay as you go. No seats, no platform fee, no minimum commitment.
When you need it
Enterprise terms
Saudi data residency, an uptime SLA and on-premise deployment, on an agreement.