AGI

.BBVA has rolled out ChatGPT Enterprise to 11,000 staff, one of the largest AI deployments in banking. A pilot showed workers saved three hours weekly, with 80% daily use and thousands of custom GPTs for internal workflows. The bank now embeds LLMs into risk analysis, software development, and a new “Blue” virtual assistant for customers, while enforcing enterprise‑grade security, role‑based training, and performance monitoring. BBVA expects up to 5% operating‑cost reductions and faster product launches, positioning it as a benchmark for AI adoption in the regulated financial sector.

Microsoft analyzed 37.5 million de‑identified Copilot chats, revealing distinct temporal usage patterns. Early‑morning hours see spikes in religion and philosophy questions, while travel queries peak during commutes. Health topics dominate mobile use around the clock. Programming spikes on weekdays, gaming on weekends, and Valentine’s Day triggers a surge in relationship advice. The study notes a shift from pure search toward personal advice, all while preserving user privacy. These insights guide AI development toward a companion that adapts to users’ practical and reflective needs.

.Microsoft’s new open‑source framework **Promptions** replaces free‑form prompts with dynamic, context‑aware UI controls, turning “prompt engineering” into “prompt selection.” By analyzing intent and conversation history, it offers options for response length, tone, format, etc., cutting the trial‑and‑error cycle. User studies show a 20‑30 % reduction in time‑to‑insight and easier task specification, though some users find the controls opaque. The stateless, scalable design integrates with existing portals and Azure OpenAI, but adoption may require change‑management for power users. Promptions aims to deliver more consistent AI outputs and lower enterprise AI costs.

Perplexity’s analysis of hundreds of millions of agent interactions shows AI “agents” are already boosting enterprise productivity. Adoption is concentrated among high‑value knowledge workers—especially in digital technology, finance, academia, marketing and entrepreneurship—who use agents for cognitive tasks (57% of activity). The dominant use cases are “Productivity & Workflow” (36%) and “Learning & Research” (21%), with agents acting as autonomous thinking partners that gather, synthesize, and act on data in core apps like Google Docs and LinkedIn. Adoption is higher in nations with greater GDP and education. Firms should audit workflow friction, upskill staff to manage AI collaborators, and strengthen security controls as the market expands from $8 bn (2025) to $199 bn (2034).

words.NTT DATA’s research of 2,567 senior executives across 35 countries shows only 15 % are AI leaders. These firms achieve rapid growth by embedding AI into core strategy, focusing on a few high‑impact use cases, and redesigning workflows end‑to‑end. Success relies on substantial infrastructure investment, an “expert‑first” talent model, disciplined change‑management, centralized governance (often via a CAIO), and strategic partnerships. This focused, well‑governed approach creates a self‑reinforcing flywheel that turns early AI wins into sustained profit and competitive advantage.

Accenture and Anthropic have expanded their partnership to industrialise generative AI for enterprises. Their new Business Group blends Anthropic’s Claude models—especially Claude Code, the market‑leading AI coding assistant—with Accenture’s implementation expertise. Accenture will train about 30,000 consultants, enabling junior developers to produce senior‑quality code and accelerating onboarding. The duo offers a product suite that quantifies productivity gains, addresses AI inference costs, and provides industry‑specific compliance solutions for finance, healthcare, and government. Emphasising responsible AI, they will use Innovation Hubs and a Claude Center of Excellence to ensure safe, scalable deployments.

OpenAI unveiled “AI Foundations,” a ChatGPT‑based certification program to bridge the generative‑AI skills gap and certify 10 million Americans by 2030. The course delivers hands‑on learning, a digital “job‑ready” badge, and psychometric validation through Coursera, ETS and Credly. Pilot partners—including Walmart, John Deere, Accenture and the Delaware governor’s office—report up to 30 % higher AI‑augmented productivity, and the initiative feeds an OpenAI Jobs Platform linking certified talent with employers. Parallel teacher and university pilots aim to seed a long‑term, universally recognized AI credential for the AI‑first economy.

OpenRouter’s analysis of over 100 trillion tokens reveals a reality far from the “AI productivity” hype. While open‑source models are chiefly used for role‑play and interactive storytelling, programming queries surged from 11 % to >50 % of all interactions in 2025, with developers feeding massive codebases to LLMs. Chinese‑origin models now account for ~30 % of global usage, and “agentic” inference—multi‑step, tool‑enabled tasks—grew to >50 % of sessions. The “glass‑slipper” effect shows early problem‑solvers gain sticky loyalty, and pricing remains only mildly elastic. These trends reshape AI product roadmaps and market dynamics.

Instacart has embedded a full checkout flow in OpenAI’s ChatGPT using the Agentic Commerce Protocol, letting users browse, cart and pay for groceries without leaving the chat. Real‑time inventory from local stores and Stripe‑powered payments eliminate the traditional “handoff” friction and reduce hallucination risks. Instacart, an early OpenAI Operator preview partner, also leverages ChatGPT Enterprise and Codex for internal automation and aims to serve as a fulfillment layer for major AI platforms. The feature is live on web, with mobile apps forthcoming, highlighting the need for accurate data, scalable payments, compliance and strong partner ecosystems.

.Resemble AI closed a $13 million strategic round—bringing its total funding to $25 million—with investors including Berkeley CalFund, Comcast Ventures, Google’s AI Futures Fund and others. The capital will expand its real‑time deep‑fake detection platform, featuring DETECT‑3B Omni (98 % accuracy across 38+ languages) and Resemble Intelligence, an explainability tool powered by Gemini 3. Deployed in entertainment, telecom and government, the solution addresses rising fraud losses projected at $40 billion by 2027. The company’s 2026 outlook predicts mandatory verification for official communications, identity‑centric security models, heightened regulatory readiness, and rising cyber‑insurance premiums.