AI is rapidly shifting from passive generation to active execution, driving personalized shopping, optimizing critical infrastructure, and redefining multimodal intelligence at scale. Here’s what the next wave looks like:

🛍️ Perplexity AI Shopping – Launches personalized conversational shopping that remembers user preferences, curates product cards, and enables seamless PayPal checkout without leaving the interface.

🔎 Tencent HunyuanOCR – Releases lightweight 1B-parameter open-source multimodal OCR model with state-of-the-art document parsing, multilingual recognition, and single-inference accuracy.

🔐 NATO x Google Cloud Deal – Establishes secure, air-gapped sovereign cloud to run classified AI workloads with full data control, enhancing defense resilience and digital sovereignty.

AI isn’t just generating—it’s remembering, interpreting, and protecting with precision.

Perplexity Joins AI Shopping Assistant Race with Personalized Experience

Perplexity has launched a new conversational shopping experience in the US that transforms online shopping by understanding individual users' unique needs and preferences. Unlike traditional search that prioritizes clicks and generic results, Perplexity’s AI assistant remembers past searches, learns users' tastes, and keeps context across queries, enabling personalized product discovery. It focuses on matching products like winter jackets or home goods to users' lifestyles rather than advertiser priorities. The platform eliminates endless scrolling by presenting curated product cards with essential specs and reviews, and integrates seamless checkout via PayPal in a single interface, allowing users to continue shopping without interruption. This approach benefits merchants too, as they retain complete customer visibility and post-purchase control while reaching higher-intent shoppers less likely to abandon carts. Perplexity’s shopping experience aims to make e-commerce personal, intuitive, and efficient, available now on desktop and web, with mobile apps launching soon.

Tencent Unveils HunyuanOCR: Lightweight, Open-Source OCR Model

Tencent has released HunyuanOCR, a native multimodal end-to-end OCR model with just 1 billion parameters, making it lightweight and highly efficient for deployment. Built on the Hunyuan architecture, it achieves state-of-the-art results across various industry benchmarks for document parsing, text detection, recognition, and multilingual translation. HunyuanOCR supports complex document analysis, field extraction from images, chart and formula recognition, video subtitle extraction, and photo translation in 14 languages. Its end-to-end design allows optimal output with a single inference, surpassing traditional cascaded solutions and offering robust performance in diverse scenarios, from invoices and street view to handwriting and games. The model is fully open source, available on GitHub and Hugging Face, and has set new standards for efficiency and usability in the OCR space.​

NATO and Google Cloud Secure AI-Enabled Sovereign Cloud Deal

NATO has entered into a multi-million-dollar agreement with Google Cloud to establish a highly secure, air-gapped sovereign cloud infrastructure for its Joint Analysis, Training and Education Centre (JATEC). This advanced platform, powered by Google Distributed Cloud, allows NATO to operate critical AI and analytics workloads on classified data while maintaining strict data residency, operational control, and digital sovereignty. The technology ensures that NATO’s sensitive information remains within its sovereign territory, disconnected from public networks, thus providing maximum security and autonomy. This partnership signifies a major step toward modernizing NATO’s digital infrastructure, integrating AI and cloud solutions tailored for defense and classified operations, and strengthening alliance resilience. Furthermore, the collaboration reflects a strategic move to develop remote operational capabilities that meet the highest security standards, ensuring long-term security and technological superiority across NATO’s member states.

Hand Picked Video

BGRemover is an AI-powered tool that makes background removal fast and effortless, delivering professional-quality results in seconds with just one click.

For a limited time, enjoy 30% off on every purchase, create stunning content with ease and save big on your next batch of edits.

Top AI Products from this week

• Craft Thanksgiving Release - Craft’s Thanksgiving release is here! Your ideas, tools, and AI now come together in a connected workspace that feels effortless. With a smarter context-aware Assistant, new API & MCP connections, and a smoother flow across Tabs, Whiteboards, and the Code Editor, Craft turns thoughts into action. Plus: our Android beta is here!

• Vibe Coding Academy V2 - Want to switch from Product Manager to Product Engineer? A Product Designer looking to launch production-ready apps? Or hoping to boost your software development workflow as a Dev? Vibe Coding Academy offers evergreen video lessons and reusable prompts to keep you ahead of the curve!

• InterviewFlowAI AI Interviews - InterviewFlowAI automates the entire first-round hiring workflow, resume scoring, candidate applications, and AI-powered phone/Google Meet interviews. Employers can make a job public, share a link for candidates to apply, view resume scores instantly, and send interview links automatically with one click.

• Pulse Platform - Pulse is an AI-powered market intelligence dashboard built for crypto traders who are tired of noise, delays, and scattered info. We built Pulse because the markets overwhelm you with useless data that arrive too late. Pulse does the opposite, it shows only what matters, when it matters.

• URLcut.ai - AI URL shortener with smart slug suggestions (not random codes). Pro tier AI READS your page for perfect suggestions. Add password protection, custom QR codes, and analytics. Community-driven, ships what you ask for.

• InsightTube - InsightTube feels like YouTube — but built for learning. Search any topic, open any channel you follow, and AI instantly highlights what matters. It summarizes ideas, breaks videos into chapters, generates multilingual subtitles, and lets you chat with the content.

This week in AI

• Ilya Sutskever says the age of scaling is over - AI now needs breakthrough research to improve generalization and robustness. He predicts human-level AGI within 5–20 years and stresses aligning AI with human values as powerful systems emerge.

• Fara-7B Microsoft’s Efficient On-Device AI Agent - This ai agent, designed for automating real web tasks directly on user devices. It uses visual perception and executes actions like clicking and typing, outperforming larger models in benchmarks while prioritizing privacy and safety. The model is open-weight and available for experimentation on Foundry and Hugging Face.

• OpenAI Expands Global Data Residency - OpenAI now offers data residency for business customers worldwide, including Asia, Europe, and the US. This allows organizations to store data at rest within their own regions, meeting local compliance and sovereignty requirements for ChatGPT Enterprise, Edu, and API users. Data remains secure and owned by customers, enhancing privacy and trust.

• Anthropic AI Could Double US Productivity - Anthropic’s analysis of 100,000 Claude conversations finds current AI cuts task time by up to 80%, potentially boosting US labor productivity by 1.8% annually—doubling recent growth rates, especially in tech, education, and professional services.

• Aluminium OS Google’s Android AI-Powered PC OS - Google’s Aluminium OS unifies Android and ChromeOS for PCs with AI at its core via Gemini models. Set for 2026 launch, it supports laptops, tablets, mini-PCs, and targets all hardware tiers.

Paper of The Day

This paper introduces "responsible computational foresight," a framework for using AI to support ethical and sustainable policymaking. It emphasizes that AI should augment human judgment rather than replace it, highlighting tools like superforecasting, world simulation, and participatory futures. The framework aims to help policymakers explore multiple future scenarios, make informed decisions, and design resilient, equitable futures, while ensuring transparency, inclusion, and scientific rigor.​

To read the whole paper 👉️ here