Coffee With: Najeeb Jarrar, Regional Product & Marketing Director, Middle East &Africa at Google. The leaderboard in the global AI race seems to shift everyday and for every use case. Google has firmly re-entered the frontlines of that race with the launch of Gemini 2.5, which has quickly become a top favorite for many, including us here at EnterpriseAM HQ.

We sat down with Google’s Regional Product & Marketing Director Najeeb Jarrar (LinkedIn) to unpack how the tech giant is thinking about user behavior, AI personalization, the future of Google Search and AI Overview, and how Google is balancing innovation with safety in one of its most dynamic markets. Below are edited excerpts from our conversation:

EnterpriseAM: Google I/O 2025 last May delivered plenty for the AI crowd and beyond. What were the headline themes you focused on this year?

Najeeb Jarrar: At I/O this year, we focused on three main themes showing how Google is pushing forward with AI.

The first theme is making our AI models more intelligent. We shipped Gemini 2.5 — our most advanced model yet. We showcased new capabilities like Veo3, the ability to generate videos from prompts that include sound.

Second is agents, or systems that can take action or build something on your behalf. We showed this inside Gemini and also in Chrome with Project Mariner.

The third theme is personalization. Right now, AI is great at handling abstract questions, but we’re focused on making it more useful for individuals differently by understanding their context, history, and interests. This could involve an agent booking a trip, recommending home decorations, or suggesting gifts based on available information.

And the overarching goal is providing AI abilities globally, across languages and use cases, with the right safety, trust, and accuracy. This is exemplified by the AI Overview on Google Search, which uses generative AI to provide direct answers and has been rolled out across the region.

E: In this evolution of AI, have you started to see Gemini or other tools cannibalizing your other products?

NJ: We don’t see this as cannibalization, it’s an evolution. Search has already transformed from ten blue links to images, cards, knowledge panels, and maps. Generative AI is simply the next step in that evolution.

Google has been deeply invested in AI for 25 years, at the forefront of developing algorithms and infrastructure for machine processing, and integrating AI tools into our products and services early on.

E: Do you see this evolution in user behavior?

NJ: Absolutely. AI Overview on Google Search now serves 1.5 bn users every month — it’s the most used AI product today. People find it useful, and it keeps them coming back to learn more and ask better questions.

E: Google is a verb, but Gemini isn't — at least not yet. How do you see the future of your user experience, especially when you face unprecedented competition?

NJ: We’ll keep finding new ways to deliver value. That’s been part of Google’s DNA from the beginning, leading to products like Google Image Search, Google News, and even prayer times in Search — all built to meet user needs.

Same goes for Gemini. Whether you use it as an app, in Gmail or Docs, or in AI Overview — it’s all built to be genuinely helpful.

Honestly, we don’t know what we’ll call it in the future. What matters is obsessing over the user’s needs and the problem we’re trying to solve.

E: Online content about Egypt and the region is limited, even in English. In the past, you had partnered with creators and government bodies to build large online archives, how are you solving this issue today?

NJ: I’d challenge two notions here:

There’s a lot of content in the Middle East in Arabic. It’s not a general content problem, but rather gaps in specific niche verticals. Over the last 15 years, content from the region has exploded — not just on websites, but also in apps, video, audio, and digital books. The Egyptian Knowledge Bank, created by Egypt’s Education Ministry, is a massive source of digital Arabic content.

And the way AI training works is changing. We don’t need massive data dumps anymore. Models now learn faster from smaller sets of filtered, curated content.

We’ve been advancing in this space for over 15 years, and actively worked to enrich Arabic content through partnerships with governments, the private sector, and volunteers, including a collaboration with Arabic Wikipedia for a decade.

E: Let’s talk about safety. With Veo3 and other tools, how do you manage trust and regulation?

NJ: That’s something we take seriously. Over time, we’ve built strong trust and safety mechanisms capable of flagging, taking down, or preventing harmful content.

We incorporate innovation directly into our products — for example, all Video Output (VO) creations include visible watermarks. We’ve also developed SynthID, an invisible watermark technology to identify AI-generated videos.

E: Inside Google, how do you personally use Gemini?

NJ: At Google, there’s a focus on stepping outside comfort zones and learning new skills using Gemini. It’s integrated into many workflows to help with many things, including: Summarizing long documents in Google Docs, creating briefs and visualizations, and crunching data and providing insights.

E: What’s a cool AI tool people may not know about?

NJ: One of the best tools people might not know is NotebookLM — we call it the “Swiss Army knife” of AI. For instance, it allows users to: Upload their own sources, like 100-page pdfs or 30 brainstorming images, have a conversation with that content, or generate an audio podcast from the exchange — like turning a 30-page report into a summary you can listen to on your commute.

Tags: