Hoda Daou, PhD

Agentic AI represents the next evolution of artificial intelligence, moving from generative systems that create, to intelligent systems that act, reason, and make decisions.   In this latest article for Gulf Business, our CEO Ashish Ipe Koshy shares how this shift is reshaping the future of work, transforming AI from a support function into intelligent teammates capable of driving outcomes autonomously.   At Inception, autonomous agents are embedded in our products to carry out into critical functions like procurement, process automation, productivity and customer experience.   Read the full article here: https://lnkd.in/ddcD9ek7   #Inception #G42 #AgenticAI #FutureOfWork #AbuDhabi #UAE

١٧٠

١ تعليق واحد

HUMAIN, a Saudi Arabian AI firm, has launched Humain Chat, an Arabic-native chatbot built on the Allam large language model, which it claims is the “world’s most advanced Arabic-first AI model” (Bloomberg). Trained on one of the largest Arabic datasets ever assembled, the bot is fluent in Arabic and English, supporting dialects like Egyptian and Lebanese, while embedding sensitivity to “Islamic culture, values, and heritage.” Powered by the ALLAM 34B model, Humain Chat goes beyond translation, aiming to deliver what the company calls “linguistic equality.” It supports real-time web search, Arabic speech input, seamless Arabic-English switching, and collaboration features. Hosted entirely in Saudi Arabia, the platform is designed to comply with the country’s Personal Data Protection Law. “For the first time, people can create, learn, and connect in their own language, culture, and context,” the startup noted in its announcement. However, questions loom. The project originated under the Saudi Data and Artificial Intelligence Authority, raising concerns about compliance with censorship laws. Yet, this issue isn’t unique to the region—American chatbots also carry biases. OpenAI has admitted ChatGPT is “skewed towards Western views,” while Elon Musk’s xAI has reshaped Grok’s ideology in response to political debates, and the Trump administration has pushed for AI models aligned with federal standards.

٣٠

١ تعليق واحد

ALLAM 34B isn’t just another AI. It’s a culturally fluent, Arabic-first AI model, bridging a critical gap in regional language technology while retaining global capabilities. It’s a milestone for Arabic AI and a testament to the Kingdom’s Vision 2030 strategy for tech-driven innovation. The model behind the platform, ALLAM 34B, was trained on more than eight petabytes of data — the largest known Arabic dataset to date. #AI #ArabicAI #ALLAM34B #HUMAIN #MachineLearning #NLP #Vision2030 #LanguageTechnology #Innovation

٤٧

In our latest "My Morning Routine" interview KSA edition, we feature Mohammed Altassan, CEO and Founder of OmniOps, a leading AI infrastructure solutions startup. Altassan, with a rich background spanning energy, data science, and national AI strategy development (including for Saudi Arabia's energy sector), shares insights into the pivotal experience of deploying AI at the Holy Mosque in Makkah, which led OmniOps to specialize in efficient AI model deployment. He discusses the critical importance of deep domain focus, fostering ownership within teams, and maintaining persistence amidst startup challenges. This conversation offers valuable takeaways for leaders in AI, technology, and entrepreneurship. Tap the link to read the full interview.

٥٨

٢ تعليق

The argument for sovereign AI seems to get stronger by the day. It is a function of the tectonic shifts in the geo-political landscape which are laying bare the need for a national AI infrastructure strategy that is not simply continental or even bloc oriented (like the EU or MENA) but at the granularity of the individual nation. But how does that really manifest itself in practice? Well.... 1. It starts with hardware. Which means NVIDIA for most organizations today. Over time AMD and Groq and others are going to have a say. But hardware, while scarce, is available. 2. It requires power. Lots of it. Most countries do this well. Some better than others, but independence increasingly matters there too. AI and energy go hand in hand. 3. It requires expertise. Very specialized expertise. This is an acute problem for many countries, even advanced ones. The last point is solvable in the short term as well as the long term. The short term answer is to license the software to run AI infrastructure - from training to #inference to #fine-tuning. This accelerates the national security objectives and it simultaneously creates the onramp for expertise. The key is to select the right software partner. That partner needs to have done it themselves, at scale on cutting edge #GPUs. The longer term answer is to accelerate learning through these partnerships, such that the nation can run on their own. This private cloud buyer's guide should be on the reading list of every government thinking about these problems. And let's be clear - every government should be thinking about these problems. The telcos in those countries are the natural channel for this strategy. They can build resilient, powerful systems leveraging what they know (networking, operations, security) while leveraging licensable expertise where they require. https://hubs.la/Q03JbG7T0

٢٦

٢ تعليق