Today's AI Pulse
OpenAI unveils ChatGPT Images 2.0 with multilingual text and ‘Thinking’ mode
OpenAI rolled out ChatGPT Images 2.0, a next‑generation image model that adds multilingual typography, real‑time web research, and agentic reasoning. The new ‘Thinking’ mode lets paid users generate up to eight coherent images per prompt, including floor plans, maps and UI mock‑ups. The feature expands the model’s ability to work with uploaded content.
Also developing:
Executive Intelligence Podcast - Celonis President Carsten Thoma on Why Enterprise AI Needs Operational Truth Before It Can Deliver
Celonis President Carsten Thoma argues that enterprise AI initiatives are failing because they lack operational context, not because of missing features. He explains Celonis’s creation of transformer‑based models that turn cryptic system event logs into a process language, forming an operational digital twin that offers a system‑agnostic view of how work actually flows. Thoma warns that vendors who lock data and push proprietary agents hinder AI effectiveness across the 25‑30 systems typical in large firms. He also calls for a new social contract to manage AI‑driven workforce changes and stresses Europe’s untapped industrial AI potential.
Certinia Unveils Veda, Its AI Engine for Professional Services Automation
Certinia introduced Veda, an enterprise‑grade AI engine for professional services automation, in its Spring 2026 release. The platform ships with 10 role‑based agents and 64 reusable actions, priced at $30 per user per month plus consumption‑based credits. Early lighthouse customers...
Gujarat HC Sends Notices To Google, Meta, X Over Deepfake
The Gujarat High Court issued notices to Meta, Google, X, Reddit and Scribd, demanding responses to a public‑interest litigation seeking tighter regulation of AI‑generated deepfake content. The bench set a May 8 deadline and ordered the platforms to join the SAHYOG...
AI Can Now Help You Pick Your Starbucks Drink
Starbucks has begun testing a ChatGPT‑based app that recommends drinks based on a user’s mood, weather, outfit or photo. Customers can chat with an @starbucks AI agent, customize a beverage, select a location and then complete checkout in the regular...
Regulatory Chaos Is Coming. AI Agents Are Already Ahead of It
The article highlights a rapidly fragmenting AI regulatory landscape, with 45 U.S. states introducing over 1,500 AI‑related bills and cities adding their own hiring rules. It explains how compliance‑focused AI agents can flag potential violations while leaving final decisions to...
Why Microsoft 365 Users Are Letting Copilot Run Their Meetings & Schedules
Microsoft 365’s Copilot Co‑work embeds generative AI across the suite, automatically crafting meeting agendas, summarizing transcripts, and drafting reports, presentations, and emails. By pulling data from Outlook, Teams, Excel and SharePoint, the feature reduces manual preparation and lets users focus...
We Need to “Right-Size” AI Ambitions in UK, Microsoft VP Tells MPs
Microsoft’s UK vice‑president Hugh Milward testified before Parliament, warning that the country’s AI roadmap lacks clear direction and that the national power grid cannot support a 1‑gigawatt data centre today. He highlighted that without a reliable energy supply, ambitious AI...
Copilot and Agentforce Fall to Form-Based Prompt Injection Tricks
Security researchers at Capsule Security uncovered prompt‑injection flaws in Microsoft Copilot Studio and Salesforce Agentforce that let attackers exfiltrate data via ordinary SharePoint and lead forms. In Copilot, the “ShareLeak” vulnerability (CVE‑2026‑21520) lets a crafted comment field override system prompts...
AI System Funghii Helps Mushroom Growers with Crop Monitoring and Harvest Prediction
Heliovision, a Belgian machine‑vision specialist, launched Funghii, an AI‑driven system that equips mushroom growing cells with four cameras each to monitor growth metrics and predict harvest timing. The platform measures cap diameter, growth rate and mushroom count over the typical...
AI Goes From Chatbot to Taking Control of Your Devices
A new wave of agentic AI tools can act autonomously, handling tasks such as scheduling meetings, drafting documents, and even controlling hardware. Leading platforms like Google Gemini’s agentic mode and Anthropic’s Claude Cowork illustrate the shift from passive chatbots to proactive...
AI Demand Soars as Traditional Deal Sizes Shrink, Kaseya Reports
Kaseya’s 2026 State of the MSP Report shows AI and automation have become the top client priority, with 48% of managed service providers ranking them above cybersecurity. At the same time, average contract values are compressing—only 41% of MSPs now...
How to Use AI to Organize Teaching Resources
Educators can leverage free AI chatbots such as ChatGPT, Gemini, Copilot, or Claude to streamline the organization of digital teaching resources. By prompting a bot to create consistent file‑naming conventions, map activities to curriculum standards, generate unit‑by‑unit vocabulary tables, and...
Why Are Workers so Worried About AI? Listen to How Business Leaders Talk About It
Workers are increasingly anxious that AI will displace them, a sentiment amplified by recent high‑profile layoffs at companies like Block and Oracle. Business leaders, however, can curb this fear by reframing AI as a tool for capacity‑building rather than cost‑cutting...
Fathom Adds a Bot-Less Meeting Mode in a Bid to Take on Granola
Fathom announced a bot‑less meeting mode that transcribes every call you attend without an AI assistant joining the room. The update adds speaker diarization, video recording options, and a Model Context Protocol (MCP) server for exporting transcript data to other...
Microsoft, Salesforce Patch AI Agent Data Leak Flaws
Security vendor Capsule Security disclosed two high‑severity prompt‑injection vulnerabilities affecting Salesforce’s Agentforce (“PipeLeak”) and Microsoft’s Copilot (“ShareLeak”). The flaws let attackers inject malicious prompts into public‑facing forms, causing unauthorized extraction of CRM leads and SharePoint data, respectively. Both companies have...
American Academy of Nursing Issues Comprehensive AI Position Statement
On Feb. 25, 2026 the American Academy of Nursing (AAN) approved a comprehensive AI position statement that frames artificial intelligence as a collaborative tool to augment, not replace, nursing judgment. Developed by the Academy’s AI Taskforce, the document outlines 13 policy recommendations...
Don’t Manage Every Task Manually — Here’s How You Can Use AI to Outdo Your Competitors in Half the Time
Founders start by handling every task themselves, but as startups grow that hands‑on approach becomes unsustainable. Integrating AI tools such as workflow platforms and language models gives leaders real‑time visibility into projects, contractor load, and launch timelines. The resulting clarity...
Salesforce Launches Headless 360 to Support Agent‑first Enterprise Workflows
Salesforce unveiled Headless 360, a platform that consolidates its Agentforce Vibes coding environment, data, and governance tools into an API‑first layer for building agent‑first enterprise workflows. The offering exposes Salesforce’s core datasets, business logic, and MCP utilities through APIs, CLI...
Waymo's Glitchy Nashville Rollout
Waymo launched its robotaxi service in Nashville this month, marking the company’s latest push into mid‑size U.S. markets. Within days, residents posted videos showing Waymo vehicles driving into active construction zones, damaging fresh asphalt, and blocking intersections, which caused traffic...
Equinix Pushes AI Into Network Layer With Fabric Intelligence
Equinix unveiled Fabric Intelligence, an AI‑driven control layer that automates design, deployment, and management of network connectivity across multi‑cloud, data‑center, and edge environments. The platform uses AI agents, a natural‑language “Super Agent,” and the Model Context Protocol to compress weeks‑long...
Inside PepsiCo’s Supply Chain Command Center with Brad Rogers, Supply Chain Planning Director at PepsiCo
In this episode of Unpacked, Brad Rogers, Supply Chain Planning Director at PepsiCo, walks listeners through the design and operation of PepsiCo’s Northeast beverage command center. He emphasizes the human‑centric, cross‑functional approach that keeps the center proactive rather than reactive,...
The Glasswing Paradox: Why AI’s Greatest Leap Is the Physical Document’s Greatest Comeback
Anthropic unveiled Project Glasswing, featuring the Claude Mythos Preview AI model that can autonomously discover and exploit thousands of high‑severity zero‑day vulnerabilities across major software stacks. The company chose to keep the model private, warning that its capabilities could cause...
Sivers Collaborates with Jabil on Energy-Efficient 1.6T Pluggable Optical Transceiver Module
Sivers Semiconductors and Jabil have partnered to develop a 1.6‑terabit‑per‑second linear receive optical (LRO) transceiver module that leverages Sivers' high‑performance distributed feedback (DFB) lasers. The pluggable module is designed for energy‑efficient optical interconnects, aiming to meet the power‑constrained demands of...
How This Open-Source AI Agent Actually Learns From Its Own Mistakes : Hermes Agent
Nous Research’s Hermes Agent is an open‑source AI assistant that continuously improves through its Generalized Action and Prompt Adaptation (GAPA) system. Every 15 tool calls the agent refines its prompts, creating reusable skills without altering the underlying model. This self‑evolving...
Delhi-NCR Leads ChatGPT Adoption in India, Top 10 Cities Account for 50% Users: OpenAI Study
OpenAI’s first‑ever capability‑gap report reveals that Delhi‑NCR leads ChatGPT adoption in India, with the top ten cities accounting for half of all users despite representing only 10% of the population. India’s AI usage is three times more city‑centric than in...
Gupshup Launches Superagent
Gupshup unveiled Superagent, an autonomous AI agent that designs, launches, and optimizes customer conversations across major messaging and voice channels. The platform handles everything from campaign setup and journey orchestration to transaction processing and performance monitoring, all via simple prompts....
Automated My 60‑Minute Study Workflow with AI
I finally automated the study workflow I’ve been doing manually for months. My "realistic handwritten notes" workflow used to take 3 apps and 60 minutes. Ribbi AI handles everything with 3 prompts and the output looks identicale. Here's the Skill I built so...
Accountability Demands Human Oversight for AI‑Driven ERP
With AI agents in ERP, who's accountable for errors in financial or supply chain decisions? Strong governance and human oversight are crucial. It's not about who owns the AI, but ensuring someone owns the outcome with trust but verify. #AIGovernance...
Executive Viewpoint: How AI Is Changing the Role of the Insurance Broker
Artificial intelligence is reshaping insurance broking from a transactional, renewal‑driven model to a continuous, data‑rich advisory practice. Large brokers are already using AI for benchmarking, coverage design and portfolio analysis, delivering faster, more consistent risk assessments and stronger client retention....
AI-Run Store Opens, Then Forgets Its Employees
An AI agent opened a store in San Francisco. Then it forgot the staff https://t.co/itybMA0Cpb via @FastCompany
Design System, Not Tools, Drives Real AI ROI
AI tools aren’t the edge. System design is. Chatbots = interface Automation = leverage Knowledge = memory Data + workflows = ROI Tools are replaceable. Capabilities aren’t. https://t.co/0YisCQc4t9
I Tested DeepSeek Vs. ChatGPT: Which Is Better in 2026?
The article pits open‑source DeepSeek against OpenAI’s ChatGPT, testing both on ten real‑world tasks ranging from summarization to coding. DeepSeek shines in technical reasoning, multilingual translation, and data analysis while remaining completely free, whereas ChatGPT delivers higher polish, broader integrations,...
Starbucks' ChatGPT App Suggests
So Starbucks has a new app on ChatGPT that will help you pick your drinks based on mood, weather, etc. So I started messing with it this morning and now I think the robot wants to call the authorities… https://t.co/KTdElSn26i https://t.co/IZTjtQrS0c
Guide AI‑Anxious Seniors Toward Future‑Proof Careers
Redoing this poll: Your high school senior is worried about their career prospects because of AI. What path do you advise:
Citi's Mandated Transformation Nears Its End, Bring on the AI Spending
Citigroup is scaling back its broad technology transformation program and trimming staff after reporting that 90% of its projects have hit or are close to their targets. The bank will redirect those resources toward a methodical, large‑scale rollout of artificial...
Google's New Desktop App Makes AI Mode Default
AI Mode is the default there -> Google released a Windows desktop app yesterday. It defaults to AI Mode. Yes, we are already at a point where AI Mode is the default (for some search products). :) https://t.co/IdgaUpIN5M
Do Data and AI Talent Needs Conflict with a Workforce Seeking Stability?
The Analytics Institute and SAS released the Data Salaries & Job Sentiment Analysis 2026, revealing that 64% of organisations plan to expand their data teams while 70% of data and AI professionals say they will not change employers this year. The report...
TOBY Secures US FDA Breakthrough Device Designation for Urine-Based Multi-Cancer Test
TOBY has received FDA Breakthrough Device Designation for its urine‑based Multi‑Cancer Early Detection (MCED) test. The non‑invasive platform analyzes volatile organic compounds in a single urine sample using spectroscopy and machine‑learning algorithms to identify multiple cancer types. The designation positions...
FREE AI Bootcamp - 🎓 Google AI Boost Bootcamp
Google has launched a free AI Boost Bootcamp that teaches users how to leverage its Gemini model across core Workspace apps. The self‑paced program consists of seven concise lessons that can be completed in under 50 minutes, culminating in a...
AI Is Coming for Your Tasks, Not Your Job. Here's What to Do About It.
AI is reshaping work by targeting specific tasks rather than entire job titles. The authors, Aneesh Raman and Ryan Roslansky, propose a three‑bucket framework—tasks AI can do alone, tasks to perform with AI, and tasks that remain uniquely human—to help...
The Most Important Number
The post examines how advanced large language models could act as high‑capacity variety attenuators, amplifiers, translators and transducers for general management, potentially flattening hierarchies and allowing tiny firms to tackle large‑scale tasks. It questions whether this vision is realistic, noting...
TDX 2026 Roundup: Agentforce Edition
At Salesforce TDX 2026 the company unveiled a suite of tools to accelerate enterprise agent development. It open‑sourced Agent Script, a strongly‑typed language that lets developers declaratively mix LLM reasoning with deterministic logic. The new Agentforce Labs Quickstart removes provisioning friction, while...
Vodafone Rolls Out New Scam Fighting AI Tool
Vodafone has integrated an AI‑driven Scam Call Protection feature into its Secure Net Mobile service, automatically analyzing inbound calls and flagging suspicious numbers with on‑screen alerts. The tool operates in the background, requiring no extra apps, and joins existing safeguards...
Civil Society in Crisis Times: New Geographies of Governance in an Era of AI
The paper by Hardill, Milnes, Mills, and Jones examines how artificial intelligence reshapes governance across the UK, focusing on civil‑society organisations that advise citizens in Wales and England. It maps emerging geographies of digital transformation, highlighting the interplay between AI,...
Meta Expands Broadcom Partnership to Co-Develop Custom AI Silicon
Meta is expanding its collaboration with Broadcom to co‑develop multiple generations of its Training and Inference Accelerator (MTIA) custom AI chips. The partnership will cover chip design, advanced packaging, and Ethernet‑based networking, with an initial commitment of more than 1 GW...
AGIBOT and Longcheer Technology Achieve World’s First Embodied AI Deployment in Consumer Electronics Precision Manufacturing Mass-Production Line
AGIBOT announced that its G2 embodied‑AI robots are now operating on Longcheer Technology’s tablet production lines, marking the world’s first large‑scale deployment of such systems in consumer‑electronics manufacturing. The robots handle precision loading, unloading and testing tasks at up to...
Anthropic's IPO Play, Novo Nordisk X OpenAI, and the First Brain Sensor Goes Human
The episode covers a rapid round‑up of AI‑related headlines: Apple is testing four frame designs for smart glasses that will focus on camera, call, and AI assistant functions rather than AR; Vercel’s CEO announced a surge to a $340 million ARR...
How the Gemma 4 Vision Agent’s “Agentic Loop” Solves Complex Visual Reasoning
The Gemma 4 Vision Agent combines the Gemma 4 Vision Language Model with the 300‑million‑parameter Falcon Perception Model, creating a multimodal system that can handle object detection, segmentation, and real‑time tracking. Its novel "agentic loop" iteratively refines predictions, delivering higher...
Second AI System Deployed for Asylum Caseworkers to Be Deployed This Month as Ministers Vow ‘Decision-Makers Cannot Use the Tool...
The UK Home Office will roll out a second AI‑driven tool, Asylum Case Summarisation (ACS), to all asylum caseworkers this month. ACS analyses interview transcripts and produces concise summaries to aid decision‑makers, joining the already‑deployed Asylum Policy Search (APS) chat‑based...
The Sequence AI of the Week #843: The AI We Built But Can't Release: A Practical View Into the Claude...
Anthropic released a system card for its unreleased Claude Mythos preview, revealing an AI that outperforms current public models but cannot be launched due to safety and governance concerns. The document details the model’s “cyber‑leap” capabilities, reckless competence, and internal...