AI is no longer just an emerging technology, it’s becoming part of everyday life, shaping how people learn, shop, and create worldwide.

• Google has unveiled its Agent Payments Protocol (AP2), a secure, open standard that allows AI agents to make payments on behalf of users, whether via cards, stablecoins, or bank transfers-bringing trust and interoperability to AI-driven commerce.

• Anthropic’s Claude adoption trends show how AI usage is accelerating globally, with wealthier economies using it for collaborative augmentation, while emerging markets lean toward direct automation, reshaping how industries from education to business operate.

• Meanwhile, VEED Fabric 1.0 is transforming video production, turning a single image into realistic, studio-grade talking videos up to 60x cheaper and 7x faster—making professional content creation accessible to everyone.

Together, these innovations signal a clear shift: AI is no longer experimental, it’s becoming the infrastructure of how we pay, work, and communicate.

Google’s Framework for AI Payments

Google has introduced the Agent Payments Protocol (AP2), an open and secure framework designed to facilitate trustworthy agent-led payments across platforms. Building on its Agent2Agent and Model Context Protocols, AP2 enables AI agents to transact on behalf of users with strong guarantees in authorization, authenticity, and accountability. It uses cryptographically signed Mandates—verifiable digital contracts—to ensure user instructions are securely followed, supporting both real-time and delegated purchases. The protocol is payment-agnostic, accommodating cards, stablecoins, and bank transfers, and promotes interoperability and scalability across the payments ecosystem. Collaborative development with over 60 partners, including major financial and tech firms, aims to establish AP2 as a universal standard for AI-driven commerce, empowering smarter shopping, personalized offers, coordinated multi-agent tasks, and emerging crypto payment systems, while fostering trust and security throughout the transaction process.

Claude Usage: US to Global Trends

The latest Anthropic Economic Index highlights how AI adoption, particularly via Claude, varies globally and within the US, revealing that usage is closely tied to GDP per capita and economic structure, with wealthier, knowledge-driven countries and states (like Israel, Singapore, and DC) using AI more intensively and for a broader array of tasks beyond coding. Notably, while software engineering dominates everywhere, the fastest-growing sectors are education and sciences, and business users are rapidly pushing toward automating tasks, with 77% of API-based Claude activity being automation, versus a more even split among consumer users. Over the past nine months, trust in AI for directive automation has surged—automation now overtakes augmentation in aggregate—and this shift is especially pronounced in lower-use, lower-income regions where direct work replacement is more common, while advanced markets increasingly use AI as collaborative augmentation.

Now create videos 60x cheaper and 7x faster!

VEED Fabric 1.0 is the world’s first AI talking video model that transforms any single image into a studio-grade, full-motion talking video with realistic lip-sync, emotion, and tone. It enables creators to quickly produce 1-minute videos that are up to 60 times cheaper and 7 times faster than traditional video production. VEED Fabric 1.0 supports a wide range of video types, including social ads, product demos, podcasts, user-generated content (UGC), and educational how-tos, making video creation scalable without the need for cameras, crew, or studios. Users can build branded characters, ship multiple video versions to test performance, and dramatically increase creative output and audience engagement—all within a seamless online platform.

Hand Picked Video

In this video, we’ll look at how ChatGPT and Claude reports compare, covering usage trends, sample sizes, key categories, and why image generation and personal use are rising fast.

Top AI Products from this week

• CodeWords – CodeWords turns plain English into powerful automations in minutes. Build, edit, deploy and fix workflows by chatting with Cody, your AI workflow builder. Connect tools, create complex logic through code, and deploy workflows without setup or configuration.

• Capalyze – Scrape real data from websites in real-time into Univers (our spreadsheet engine with 27.5k stars on GitHub!). Ask questions, visualize with interactive tables & charts, and then slice, dice, and export — all in one place.

• DeepTagger – DeepTagger is a no-code platform that makes your judgment scalable. It uses your annotations as an example to extract information from new documents. Highlight what matters to you once, and let DeepTagger handle therest with precision. API access included.

• Mureka Agent Studio - AI drafts content fast, but often gets facts wrong and sounds off. ContentLift is the fact-checking and quality layer that makes AI drafts accurate, clear and on-brand - verifying sources, fixing claims and refining tone so you can publish with confidence.

• kluster.ai – Over 40% of AI-generated code contains issues. kluster.ai automatically reviews and fixes code in real-time as AI writes it, instantly correcting bugs, preventing security vulnerabilities, and fixing logic errors right in your favorite IDE.

• Gamma 3.0 – Haystack is an AI-native code review platform that makes PRs feel less like a jigsaw puzzle and more like a guided story.

This week in AI

• Meta Ray-Ban HUD Leak - Meta’s upcoming Ray-Ban Display glasses with a monocular HUD and sEMG wristband leak reveals AI, navigation, and gesture controls. Weighing 70g, priced around $800, launching at Meta Connect 2025.

• Holo1.5 AI Model - Holo1.5 models boost UI localization accuracy by 10%, excel in computer use tasks, outperforming top open and closed-source models for native interface software control.

• ChatGPT’s New Orders Feature - OpenAI plans “Orders” in ChatGPT for native checkout and tracking, letting users pay and track purchases within the app. Parental controls are also in development. Launch likely US-first.

• RustGPT: Pure Rust LLM - RustGPT is a transformer-based LLM built entirely in Rust, featuring pre-training, instruction tuning, and interactive chat without external ML frameworks or heavy dependencies.

• HuMo AI Video Gen - HuMo is a multimodal AI for human-centric video generation, blending text, images, and audio to produce high-res, lip-synced, and controllable videos with stable subject preservation.

Paper of The Day

The Cognitive Layered Memory Architecture (COLMA) offers a next-generation AI memory system inspired by human cognitive processes. It integrates distributed storage with cognitive principles to enable dynamic updates, cross-modal associations, and continual learning. Organized into five layers, COLMA supports complex reasoning, recall, and memory integration, simulating biological memory interactions like those between the hippocampus and neocortex. It unifies diverse data formats, offers high interpretability, and excels in scalability and adaptability, aiming to power advanced AI cognition and collaborative multi-agent intelligence.

To read the whole paper 👉️ here.