Today's AI Pulse
Odyssey lands $310M Series B, valued at $1.45B
Odyssey announced a $310 million Series B round that values the AI lab at $1.45 billion. The round was led by Natural Capital and included backers such as Amazon, AMD Ventures, GV, EQT and IQT. In parallel, Odyssey struck a preferred‑cloud deal with AWS to use Trainium silicon for large‑scale model training.
Also developing:
Morningstar Launches AI Assistant for Flagship Platform
Morningstar has introduced an AI assistant embedded in its Direct Advisory Suite, currently in beta for a limited group of 4,000 advisors out of more than 180,000 users. The voice‑enabled assistant combines multiple AI agents to streamline research, portfolio analysis, meeting preparation, and proposal generation within a single interface. It draws on Morningstar’s extensive data while ensuring client data is never used to train the models and maintains enterprise‑grade security. A broader U.S. rollout is slated for 2026, with Canadian availability to follow later that year.
How to Roll Out an AI Gateway Across Your Organization
Enterprises are increasingly adopting AI gateways to centralize model access, governance, and cost management as AI usage proliferates across teams. The article outlines signs that indicate a fragmented AI stack—such as disparate provider integrations, opaque spending, and inconsistent policies—and presents...
Gartner Warns CMOs Against Using Agency AI Platforms
Gartner warns that up to half of agency‑built AI platforms could become obsolete within three years as open‑source solutions gain dominance. The firm predicts open‑source AI will support more than 75% of business deployments by 2028, offering greater customization and...
Brex Is Now in the ChatGPT App Marketplace So You Can Ask Anything About Your Expense Data
Brex has entered the ChatGPT app marketplace, becoming one of the first fintech solutions available in OpenAI’s ecosystem. ChatGPT Enterprise users can now ask natural‑language questions about their company’s spending, balances, and expense policies directly within the chat interface. The...
Your AI Isn’t Failing. Your Org Just Can’t Absorb It
A Wall Street Journal survey revealed a 38‑point gap between executives, who claim AI saves eight or more hours weekly, and front‑line workers, who report less than two hours of benefit. A National Bureau of Economic Research study of 6,000...
Are We Overlooking Small Language Models? Everything You Need to Know About Efficient SLMs
Researchers are championing small language models (SLMs) as a cost‑effective alternative to massive AI systems, highlighting breakthroughs that deliver near‑state‑of‑the‑art performance with a fraction of the parameters. Efficient fine‑tuning techniques and model distillation enable these compact models to run on...
AI Shifts the Apps Space – Time for a New Apps Leader Board?
The software industry is entering an AI‑first apps era, forcing legacy cloud vendors to adapt or face margin compression and customer churn. Investors are penalizing traditional SaaS firms as AI‑enabled solutions promise lower costs and faster development. Vendors with high...
At Target, the Only Wrong Move Is Opting Out
Target’s senior product leader Kunal Banerjee says AI is an execution accelerator, not a strategic disruptor, and its impact depends on existing planning discipline. He notes that content‑creation teams have realized strong efficiency and cost gains, while design teams see...
Monday: Three Morning Takes
The U.S. Department of War has labeled Anthropic a supply‑chain risk, threatening its AI models from Pentagon contracts unless a deal is struck. DOW CTO Emil Michael signaled openness to renegotiating, suggesting a potential path forward for Anthropic. Meanwhile, new...
Embraer Turns to AI Solution for A-29 Super Tucano CUAS Evolution
Embraer is integrating Valkyrie Aero’s Gunslinger artificial‑intelligence suite into its A‑29 Super Tucano platform to boost counter‑uncrewed aerial system (CUAS) capabilities. The AI-driven system will provide pilots with real‑time tactical decision‑making assistance during drone‑threat engagements. While the Super Tucano already...
Nano Interactive Launches NanoQ Agentic Media Planner Built to Transform Programmatic Planning
Nano Interactive unveiled NanoQ, an agentic AI media planner that automates audience creation and activation. Built on the company’s intent data and the continuously refreshed Intent Library, NanoQ translates natural‑language briefs into custom and off‑the‑shelf segments. In internal tests the...
How to Prevent AI-Generated 'Workslop' From Destroying Your Workplace Relationships
The article defines “workslop” as AI‑generated content sent without editing, which erodes trust among coworkers. Research from BetterUp Labs and Stanford Social Media Lab shows 50% of recipients view the sender as less creative and 40% see them as less...
KKR Mulls Exit From Data Centre Cooling Specialist CoolIT
KKR is weighing a multibillion‑dollar sale of CoolIT Systems, a data‑centre liquid‑cooling specialist that it bought for roughly $270 million in 2023. The potential transaction could value the company at over $3 billion, reflecting the rapid shift of CoolIT from gaming‑PC cooling...
Novobanco Partners Feedzai to Modernise AML and Fraud
Novobanco has entered a multi‑year partnership with Feedzai to overhaul its anti‑money‑laundering (AML) and fraud defenses using Feedzai’s AI‑native platform. The initiative consolidates fragmented compliance tools into a single, intelligence‑driven system that merges KYC, AML and fraud functions. Advanced watchlist...
Lenovo and SWM Partner on Next-Generation Robotaxi
Lenovo and South Korean autonomous mobility firm SWM announced a partnership to co‑develop a next‑generation robotaxi. The project centers on SWM’s AP‑700 autonomous driving platform combined with Lenovo Vehicle Computing’s AD1 domain controller, which runs on Nvidia’s Drive AGX Thor...
The Empathy Deficit: The Mistake Most Founders Make with AI
Founders are racing to ship AI products, but a growing "empathy deficit" threatens success. While 52% of executives cite critical thinking as essential, only 17% prioritize empathy—cognitive empathy that models real user behavior. The article argues that neglecting this skill...
The Sustainable Choice for the AI Era
Dr Ayotunde Coker, CEO of Open Access Data Centres, warns that AI adoption will surge across sub‑Saharan Africa, generating unprecedented data‑processing demand. He argues that building on‑premises AI‑focused data centres is prohibitively costly and slow, making vendor‑neutral colocation the sustainable alternative. OADC’s...
Tencent Cloud, Maxon Partner to Add AI to Cinema 4D
At Mobile World Congress 2026, Tencent Cloud announced a partnership with Maxon to embed its generative AI model, HY 3D Global, into Cinema 4D on iPad and desktop. The integration will deliver API‑based high‑quality 3D model generation, enhancing speed, reliability and creative flexibility...
Lockton Re to Deploy Agentic AI to Streamline, Drive Growth
Lockton Re, the reinsurance arm of the global broker, has teamed with digital‑transformation firm Synechron to launch the Agentforce 360 platform, Salesforce’s agentic AI solution. The platform is tightly integrated with Salesforce Financial Services Cloud, giving AI agents access to...
AI Figures Out the Rules of a Mysterious 2,000-Year-Old Board Game From Ancient Rome
Researchers used the Ludii AI system to decode a 2,000‑year‑old Roman stone board discovered in the Netherlands. By combining use‑wear analysis with AI‑driven gameplay simulations, they identified a plausible rule set where a "hunter" with four pieces attempts to trap...
Turning Logistics Data Into a Massive Competitive Advantage
The 2026 Gartner Market Guide highlights AI’s rise in freight audit and payment, shifting the focus from simple invoice checks to comprehensive data audits. Legacy systems struggle with fragmented, unstructured logistics data, creating margin leakage across the $11 trillion spend ecosystem....
19 Large Language Models for Safety or Danger
The article surveys 19 large language models divided into three safety categories—guard‑rail models, looser “uncensored” models, and abliterated models that strip safety layers. It highlights guardrail examples such as Meta’s LlamaGuard, IBM Granite Guardian, and Anthropic Claude, while noting uncensored...
How AI Will Impact Security Careers and What Leaders Should Do About It
Security leaders surveyed in Tines’ Voice of Security 2026 report remain optimistic about AI’s potential, yet 81% report rising workloads and 76% admit to burnout. Teams still spend roughly 44% of their day on manual, repetitive tasks that AI could...
Bad News for Your Burner Account: AI Is Surprisingly Effective at Identifying the People Behind Them
A new ETH Zurich and MATS study shows large language models can de‑anonymize burner accounts with striking accuracy. The AI system identified 68% of users in the test set and achieved a 90% precision rate. Researchers warn that the assumed...
#350 How to Make Hard Choices in AI with Atay Kozlovski, Researcher at the University of Zurich
In this episode, philosophy researcher Atay Kozlovski discusses the ethical challenges of AI, focusing on how to maintain meaningful human control over increasingly autonomous systems. He highlights common failure modes such as automation bias and algorithmic bias, illustrating them with...
Kutta Launches KED to Deliver AI-Enabled Compute at the Tactical Edge
Kutta Technologies has introduced the Kutta Edge Device (KED), an ultra‑low‑power, rugged compute platform designed for tactical, disconnected environments. Consuming less than 7 watts, the device delivers AI processing, multi‑stream target recognition, and networking while dramatically cutting battery weight. KED integrates...
AI's Future Lies in Embedding, Not Replacing, Middle Management
AI is not replacing the middle layer in business after all. It is pulling it in. If even OpenAI and Anthropic are teaming up with management consultants, that says a lot. The race is no longer just about having the best...
The ‘Bayesian’ Upgrade: Why Google AI’s New Teaching Method Is the Key to LLM Reasoning
Google researchers propose Bayesian Teaching, a method that trains large language models (LLMs) to emulate a Bayesian assistant’s belief‑updating process rather than simply memorizing correct answers. By fine‑tuning on synthetic flight‑booking interactions, the approach forces LLMs to reason under uncertainty,...
Can AI Replace Excel for Vendor Statement Reconciliation?
Finance and AP teams still rely on Excel for vendor statement reconciliation, but growing transaction volumes expose its manual, error‑prone nature. AI‑driven platforms now automate data extraction, transaction matching, and discrepancy detection, handling inconsistent formats and large data sets in...
Podcast: The Industrial Revolution for Law
The Without Limitation podcast released an episode featuring Richard Tromans, the founder of Artificial Lawyer, discussing how the legal‑tech site launched in 2016. Hosted by Matt Pollins, co‑founder of Lupl, the conversation delves into the transformative effects of legal AI,...
NXP’s New i.MX 93W Fuses Edge Compute and Secure Wireless Connectivity to Accelerate Physical AI
NXP Semiconductors unveiled the i.MX 93W applications processor, the first SoC to fuse a dedicated AI neural processing unit with secure tri‑radio Wi‑Fi 6, Bluetooth LE, and Thread/Zigbee connectivity. The chip integrates a dual‑core Cortex‑A55, an Arm Ethos NPU capable of up to...
Google Workspace CLI : Drive, Gmail & Slides Commands for AI Agents
Google has launched a Rust‑based Command Line Interface for Google Workspace that lets developers automate Gmail, Drive, Sheets, Slides and other services with AI agents. The CLI offers over 100 downloadable skills, dynamic command updates that track API changes, and...
Claude Code Automates Testimonial-to-Email Nurture in Minutes
I am officially addicted to Claude Code for marketing Woke up early again Today's mindblowing workflow? Asked Claude to: >> Pull my 5 latest testimonials from Senja >> Turn them into customer-centred emails >> Compose them directly inside CustomerIO 5 minutes later I have a whole new...
GenAI-Based Development Platform - Part 1: Guardrails
The article introduces a GenAI‑based development platform, dubbed Harness, that layers deterministic guardrails around coding agents such as Claude Code. It outlines four protective mechanisms—pre‑commit checklist skill, pre‑commit Git hook, GitHub Actions workflows, and automated pull‑request reviews—to catch errors and...
Exeter Leads Major New Project to Advance AI-Enabled Platform for Early Detection of Hospital Acquired Infections
The NIHR HealthTech Research Centre in Exeter and Sanome have secured a £300,000 Innovate UK SMART grant to advance MEMORI, an AI‑driven clinical decision support platform that predicts hospital‑acquired infections up to seven days before symptoms appear. Early trials show...
AI-Ready Biodata Is America’s Next Strategic Infrastructure
The article argues that AI‑ready biodata is a strategic national infrastructure essential for U.S. leadership in AI‑enabled biotechnology. While competitors, especially China, are building coordinated bio‑data ecosystems, the United States relies on fragmented, underfunded, and insecure repositories. Gaps in data...
GenieAI Unveils Eidetic Intelligence – Patent-Pending AI Redefining Legal Accuracy Amid Industry Disruption
GenieAI launched Eidetic Intelligence, a patent‑pending AI architecture built specifically for legal work. In internal tests the system achieved 90% accuracy on simulated risk assessments, outpacing all other large language model providers. The platform layers deterministic state‑machine workflows, quality‑gated validators,...
China’s “Palantir Equivalent” Xunce Forecasts Strong Revenue Growth Amid AI Data Surge
Shenzhen‑based Xunce Technology, dubbed China’s Palantir equivalent, forecasted 2025 revenue of RMB 1.283 billion, a 103% year‑on‑year increase. The company expects its adjusted net loss to narrow to RMB 55 million as it expands AI data‑agent infrastructure into robotics and aviation. Revenue surged in...
When Legal AI Sounds Right But Fails Across Borders
Legal AI can generate polished, English‑centric answers that appear credible but often miss jurisdiction‑specific nuances, especially in multilingual or cross‑border contexts. The underlying foundation models lack the structured, comparative legal knowledge needed to recognize non‑equivalence, leading to subtly incorrect advice....
AI and the New Blueprint of Terrorism
AI is lowering the barrier for small, non‑state groups to conduct targeted violence by pairing advanced models with affordable robotics, sensors, and energy tech. Open‑source and open‑weight models, while less powerful than proprietary systems, can run locally on modest hardware...
Xpeng Reframes Autonomous Driving as AI Deployment in the Physical World
Chinese EV maker Xpeng unveiled its second‑generation vision‑language‑action (VLA) model, announcing a strategic shift to bypass Level 3 autonomous driving and aim directly for Level 4 capabilities. The company merged its cockpit and autonomous‑driving divisions into a single AI platform, positioning the...
AI Walks Into an Arbitration: What Could Go Wrong?
Artificial intelligence is already being deployed in arbitration for document review, evidence organization, and drafting, offering speed and cost savings. Yet the rapid adoption outpaces regulatory guidance, with South African bodies issuing only soft guidelines and international rules lagging behind....
Synthetic Video Now Realistic; Verification Becomes Critical
Four years ago, AI-generated video was a fever dream. Movements were glitchy. Faces were unstable. Physics? Entirely optional. Fast forward to today: The transformation is striking. Lighting behaves naturally. Motion follows the laws of physics. In short clips, the line between synthetic...
AI Progress Is Collaborative, Not a US‑China Race
I think highly of Vinod but am skeptical of this very common AI viewpoint. First, it's proving difficult for anyone to really break out of the AI pack. AI models are much more similar than they are different. There will...
Nihilism Finds Validation in AI's Truth‑Irrelevant Output
There's a "curious affinity between... nihilism and AI enthusiasm: If truth is irrelevant anyway, then LLMs – which essentially produce plausible statements without reference to truth (critics call it bullshit) – are not a threat but confirmation" https://t.co/6vkPk0n6NP
Clear, Step‑by‑step Integration Guide Beats Hype
Let's say you are an agent builder and want to integrate a promising new vendor you found. What would you be happiest to see in the docs (not based on twitter hype; you personally for your situation right now):
AI Is the Railway, Not Just the Train
Many people think of AI as the train. In reality, AI is the railway. The real shift is the infrastructure being built today: compute, energy, talent, and digital stack. Countries that invest in this infrastructure will shape the future of AI. My conversation with...
Robot Achieves True First-Person Vision
Seeing Through Its Own Eyes: Building a #Robot with True First-Person Vision via @WevolverApp #AI #MachineLearning #Robotics #ArtificialIntelligence #Innovation #Technology https://t.co/uk88U8QE6Q
NiCE Secures CX Orchestration Lead with $955M AI Deal
The $955M @NiCEltd - @Cognigy deal isn’t just another tuck-in; it’s a move to own the CX orchestration layer for the autonomous enterprise. NiCE is putting rivals on notice by integrating native #Agentic AI. 🚀 Read our take: https://t.co/n8v7iUkNoL #CIO...
AI Advances Threaten Online Anonymity, Linking Accounts to Identities
A new study suggests AI could make online anonymity harder to maintain. Researchers found that the same technology powering tools like ChatGPT can help link supposedly anonymous social media accounts to real identities. As AI gets better at pattern recognition, the privacy...