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:
RAGtime Demo Days
Lawfare announced an artificial‑intelligence‑powered research platform aimed at federal litigation, debuting it at the RAGtime Demo Days event. The tool automatically pulls court opinions, executive orders, and other national‑security documents into a searchable interface. Editor in Chief Benjamin Wittes presented the demo, highlighting how the system reduces manual document hunting. Lawfare positions the solution as a free‑up for lawyers, scholars, and civil‑society analysts.
AI Agents Use Three Distinct Memory Layers
You use AI agents every day. ChatGPT, Claude, Gemini, Cursor. They remember what you said, who you are, and what you've worked on before. But this memory has three layers, not one. 1️⃣ Working memory Your current chat. Whatever you typed five messages ago....
The AI Talent Problem CIOs Cannot Delegate to HR
Enterprises are racing to adopt AI, but CIOs risk losing top talent because HR systems still focus on traditional pay structures. The most coveted AI engineers now prioritize access to compute, frontier models, and unrestricted experimentation over salary alone. This...
Delos AI Improves Inference with 1,000X Higher Scale
Delos Data unveiled a new cluster architecture that links thousands of GPUs via a high‑bandwidth interconnect, promising up to 1,000× higher inference scale. The design eliminates GPU idle time caused by data‑center networking bottlenecks, and is paired with Mosaic™ software...
Startup Boosts Scale-Up to 1000+ GPUs in a Single Domain
Delos Data unveiled its Nonstop AI platform, a disaggregated server design that can link more than 1,000 GPUs in a single scale‑up domain. By using nine OSFP ports per accelerator and 72 × 200 Gb/s connections per server, the system offers flexible topologies and...
Machine Learning System Design Interview #39 - The Feature Space Trap
In a Netflix senior ML engineer interview, candidates are asked how to handle a model that improved offline AUC by 4% through extensive feature crosses but violates a 20 ms inference latency SLA. The common answer—scaling the inference cluster—is flagged as...
Reviewer Select Unveils Semantic AI Platform for Faster Reviewer Matching
Reviewer Select has launched a new website showcasing its semantic AI platform that accelerates peer‑reviewer matching for journal editors. The system extracts research concepts from manuscripts and ranks reviewer candidates by expertise fit, providing evidence such as DOI links and...
Skan AI Introduces New Intelligence Framework for Enterprise AI Agents
Skan AI unveiled the Agentic Business Context Foundation (ABCF), a technical framework that captures the hidden operational intelligence of enterprise workflows. By modeling human reasoning, exceptions, and informal workarounds, ABCF supplies AI agents with actionable context that traditional documentation and...
Google Pushes Traditional Search Results Further Down the Page
Google announced at I/O that its search homepage will be overhauled with an "intelligent search box" powered by Gemini, pushing large‑language‑model answers ahead of traditional blue‑link results. The redesign encourages conversational follow‑ups and introduces LLM agents that can monitor web...
How Data Science Is Powering the Next Generation of AI Chatbot Solutions
Data science has moved from isolated labs into everyday business tools, reshaping AI chatbots from rigid rule‑based scripts to intent‑aware assistants. Large‑scale NLP models, fine‑tuned on labeled conversation data, now interpret sentiment, detect anomalies, and adapt in real time. Robust...
Outmarket AI Raises $17M to Expand AI Platform Purpose-Built for Insurance
Outmarket AI announced a $17 million Series A round led by Permanent Capital Ventures, bringing its total funding to $21.7 million. The AI platform integrates directly with agency management systems to convert structured and unstructured brokerage data into an intelligence layer for commercial,...
Therapists Turn to AI Note‑Taking Tools, Prompting Privacy and Trust Concerns
Therapists nationwide are rolling out AI‑driven note‑taking platforms that record sessions, transcribe dialogue and draft clinical notes. A patient in Arkansas says the technology left her feeling violated, while vendors argue the tools cut administrative burden and comply with HIPAA....
GPU Autoscaling on Kubernetes with KEDA: Building an External Scaler
KEDA’s default autoscaling ignores GPU metrics, leading to wasted accelerator capacity and higher energy use. To fix this, a custom DaemonSet called keda‑gpu‑scaler reads NVIDIA NVML data on each node and exposes utilization, memory, temperature and power metrics via KEDA’s...
AI‑Written Story Wins Commonwealth Short Story Prize, Prompting Fierce Debate
Jamir Nazir's "The Serpent in the Grove" won the 2026 Commonwealth Short Story Prize even after AI‑detection tools flagged it as likely written by ChatGPT. The jury praised the piece while the publisher admitted the truth may never be known,...
Cerebras Shares Surge 68% on Debut, Sparking Fresh IPO Trading Frenzy
Cerebras Systems (CBRS) opened at $350 on May 14, soaring 68% to close at $588 and giving the company a market value of roughly $67 billion. The rally attracted immediate attention from high‑profile investors, including a $32 million purchase by Cathie Wood’s...
IREN Ltd. Buys Dell’s Blackwell Systems for $1.6B, Aims to Lift ARR to $4.4B
Australian data‑center operator IREN Ltd. signed a $1.6 billion agreement with Dell Technologies to acquire Nvidia‑based Blackwell systems. The hardware will support IREN’s five‑year, $3.4 billion AI cloud contract and is expected to boost its ARR from $3.7 billion to $4.4 billion.
Opendoor Co‑Founder Eric Wu Raises $25M to Launch AI Construction Copilot
Serial entrepreneur Eric Wu, co‑founder of Opendoor, announced the launch of NavigateAI with a $25 million seed round led by Elad Gil. The San Francisco startup aims to deliver AI‑driven copilots that guide construction and field workers in real time, promising faster,...
Bravent Rolls Out Autonomous AI Agents for U.S. Enterprise Workflow Automation
Bravent, a Microsoft‑partner consulting firm, announced the U.S. rollout of autonomous AI agents designed to automate enterprise workflows. The launch will be showcased at a Miami‑based event on April 30, featuring Microsoft experts and early‑adopter customers.
University of Houston Develops AI-Powered Radar System to Inspect Concealed Cold-Formed Steel
Engineers at the University of Houston have created an AI‑driven ground‑penetrating radar (GPR) system that can locate and assess concealed cold‑formed steel framing without tearing down walls. The method pairs rapid radar scans with a vision foundation model, InternImage, trained...
SpaceX Targets $1.75 Trillion Valuation as Anthropic and OpenAI Gear Up for 2026 AI IPOs
SpaceX filed an S‑1 seeking a $1.75 trillion valuation, and insiders say Anthropic and OpenAI are preparing their own 2026 IPOs. The three AI powerhouses could dominate market allocations, push valuations higher, and force investors to weigh soaring AI compute costs...
AI Ticketing Platform HelpDesk Boosts Retention with 92% CSAT and 33M Automated Actions
HelpDesk, the AI‑driven ticketing platform from the LiveChat and ChatBot team, processes 33 million automated actions annually, cuts first‑response time to four hours and records a 92% “great” CSAT score. The tool is being hailed as a retention engine that transforms...
Lego, Hip-Hop, And Deepfakes: How Iran Uses AI To Shape Western Opinion
Iran has escalated its influence campaigns by deploying AI‑generated Lego‑style videos, hip‑hop tracks and deepfake battlefield footage aimed at Western audiences. The effort, coordinated by the Basij, IRGC units and the Ministry of Intelligence, uses platforms such as X, TikTok,...
How Internal Communicators Can Embrace AI
Internal communications teams face rising expectations for speed, relevance, and personalization, while resources remain constrained. A new guide from Interact shows how AI can be leveraged beyond simple drafting, improving workflow efficiency, content relevance, and strategic impact. It outlines practical...
AI Is Massively Increasing China’s New Weapon Development Speed: Scientists
Chinese researchers at Chongqing University unveiled ChatBearing, an AI‑driven design agent that integrates large language models with engineering tools and a 4,500‑record bearing database. The system autonomously handles requirement analysis, load calculations, selection, life prediction, strength verification and report generation,...
Why Your AI Strategy Is Failing - Barry O'Reilly (Author, Artificial Organizations)
In this episode, Barry O'Reilly explains why AI initiatives fail when organizations start with tools instead of behavior change, noting that 85% of Gen AI projects and 83% of transformations flop for this reason. He shares personal examples—from using transcription...
U.S. Companies Have an AI Problem. Indian IT Wants to Be the Solution
U.S. firms are struggling to turn AI pilots into profit, with a 95% failure rate for generative‑AI projects. Indian IT giants such as TCS, Infosys, Wipro and Tech Mahindra are pivoting to capture the AI deployment layer, leveraging deep enterprise relationships...
Press Release: KPMG Launches Trusted AI Centre of Excellence to Strengthen Singapore’s Position as a Globally Trusted AI Hub
KPMG announced the launch of its Trusted Artificial Intelligence Centre of Excellence in Singapore, backed by the Economic Development Board. The Centre introduces a Trusted AI Assurance service that delivers evidence‑based, sector‑specific assessments of AI governance, security, and regulatory compliance....
Astemo, Hitachi to Build AI Platform for SDVs
Automotive supplier Astemo and technology giant Hitachi have announced a joint effort to build an artificial‑intelligence development platform for software‑defined vehicles, with a target rollout by the end of fiscal 2026. The platform will fuse AI, data, and digital‑twin capabilities...
Timely Takes Podcast: Voice Analysis Technology & Earnings Calls
The Timely Takes podcast features James Palczynski, CEO of DeCue Technologies, discussing AI‑driven voice analysis that reads paralinguistic cues below human perception. The technology can infer stress, emotional state, and even health indicators from earnings‑call audio, a concept first detailed...
Experimenting with Generative AI to Support Delivery Officers at DfE
dxw partnered with the UK Department for Education on a three‑month experiment to evaluate a generative‑AI assistant built in Microsoft Copilot Studio. The prototype, deployed in Teams, aimed to help delivery officers locate answers in extensive policy guidance faster than...
Datacentre Dive: Do AI Datacentre Physics Make On-Premise Unviable?
The surge in AI workloads is forcing datacenters to adopt direct‑to‑chip liquid cooling and 800 V DC power delivery, rendering traditional air‑cooled on‑premise facilities increasingly impractical. Schneider Electric emphasized that GPUs exceeding 700 W cannot be cooled with air, and modern AI racks...
Prompt, Do, Check, Act: The New PDCA
The article re‑imagines the classic PDCA cycle for AI‑augmented knowledge work, swapping “Plan” for a carefully crafted Prompt and letting the large language model execute the “Do” step. Human users retain responsibility for the “Check” and “Act” phases, validating outputs...
Nanjing Agricultural University Launches Green Shield, China’s First Open-Source AI Model for Crop Protection
Nanjing Agricultural University, together with the National Key Laboratory of Agricultural Biosafety and over 30 industry partners, has launched Green Shield, China’s first open‑source large language model dedicated to crop protection. The model was trained on more than 2.5 billion tokens...
AI Chatbots Transform SOPs Into Instant Knowledge Assistants
𝗔𝗜 𝗞𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗔𝘀𝘀𝗶𝘀𝘁𝗮𝗻𝘁𝘀: 𝗧𝘂𝗿𝗻𝗶𝗻𝗴 𝗖𝗼𝗺𝗽𝗮𝗻𝘆 𝗗𝗼𝗰𝘂𝗺𝗲𝗻𝘁𝘀 𝗶𝗻𝘁𝗼 𝗦𝗺𝗮𝗿𝘁 𝗖𝗵𝗮𝘁𝗯𝗼𝘁𝘀 📄⚡ That’s why companies are turning SOPs and manuals into AI-powered knowledge assistants. Why this matters: 🎯 Learning happens at the point of need ⏳ Eliminates endless searching through SharePoint or intranets 🧠 Creates a single,...
Docker Sandboxes and microVMs, Explained
Docker introduced Sandboxes in April 2024, wrapping each container inside a purpose‑built microVM that runs on the host’s native hypervisor. The microVM gives every sandbox its own lightweight kernel and isolated Docker daemon, delivering VM‑level security with seconds‑fast startup. Designed...
The NSA, ‘Mythos’ and the Quiet Emergence of AI Cyber Doctrine
The U.S. government and leading tech firms are rapidly integrating frontier AI models, such as Anthropic's Claude Mythos, into offensive cyber operations, shifting the threat landscape from tool‑centric to autonomous agent‑centric. Mythos demonstrated autonomous discovery and exploitation of thousands of...
What Do Software Developers Do Now?
Software developers are rapidly transitioning from hand‑coding to directing AI agents like Claude Code. The author now relies on AI for analysis, planning, and most of the actual code, reviewing only the output. This shift boosts productivity dramatically but erodes...
Skip Hundreds, Read These 9 AI Essentials
If you want to understand 99% of AI and ML, skip the 1,000+ books on Amazon and grab these 9 👇 #1 "LLM Engineer's Handbook" by @pauliusztin_ & @maximelabonne Iusztin and Labonne offer a masterclass in deploying LLMs. They skip...
Myriad Genetics Launches Prolaris + AI Test for Personalized Prostate Cancer Active Surveillance
Myriad Genetics is introducing the Prolaris + AI Test, the first prostate‑cancer assay that blends its long‑term Prolaris CCR genomic score with an artificial‑intelligence model (AI‑GUR) that predicts near‑term Gleason score upgrades. The AI metric will be available for all Prolaris Biopsy...
AI-Accelerated ERP Transformation Requires Context, Period
Enterprise ERP migrations remain costly, complex, and slow, often consuming multi‑year budgets and large consulting teams. The article argues that speed alone isn’t enough; without operational context, AI‑driven tools merely accelerate technical work without delivering business value. Celonis proposes an...
Darrow, the ‘AI Lab for Legal Risk,’ Launches a Platform to Let Plaintiffs’ Firms Manage Litigation Like an Investment Portfolio
Darrow, the AI‑focused legal‑risk lab, unveiled a new SaaS platform that lets plaintiffs’ firms treat active and potential lawsuits as a portfolio of investments. The tool combines web‑scraped class‑action leads with predictive analytics, risk scoring, and a dashboard that mirrors...
Racism Bias in AI? Hiring Tools Screening Out Black and Asian Job Applicants
A Stanford‑led study of 4 million applications across 156 firms found that AI hiring tools, primarily the Pymetrics platform now owned by Harver, systematically screen out Black and Asian candidates. The analysis identified adverse impact in roughly one in ten roles...
Pitch Deck: How Olyzon Sold Sir Martin Sorrell’s S4S Ventures on Its $10M CTV AI Bet
Paris‑based AI startup Olyzon announced a $10 million Series A round, adding Sir Martin Sorrell’s S4S Ventures to its backers while existing investors Cassius Capital, Ventech and Eurazeo also participated. The company’s AI agents automate audience definition and media‑plan generation for brands...
One in Five Workers Say AI Has Replaced Parts of Their Job
A recent poll of over 2,000 full‑time workers finds that one in five say artificial intelligence has already taken over parts of their role, while 15% report new tasks emerging from AI use. The study reveals that roughly half of...
The AI Marketing Audit Framework: 47 Checks to Run Before You Spend a Penny on AI
Lilach Bullock released a comprehensive AI Marketing Audit Framework that outlines 47 concrete checks across six categories—data foundation, workflow documentation, tool stack, team capability, volume threshold, and commercial fit. The audit scores businesses on a 0‑47 scale, with 35+ indicating...
AI Implementation Roadmap: The Honest 90-Day Path (From Someone Who Ships It)
Lilach Bullock’s latest blog post unveils a pragmatic 90‑day AI implementation roadmap that reliably ships two to three AI‑driven workflows for founder‑led businesses. The six‑step plan compresses discovery to one week, delivers a detailed design, builds, ships, and hands off...
10 Problems AI Can Help Fashion Solve
Business of Fashion’s latest memo outlines ten ways AI can untangle chronic fashion challenges, from accelerating design mockups to revamping product search with large‑language models. The piece highlights case studies showing AI‑driven marketing, the rise of AI‑generated fake product images,...
Support Levels Determine AI's Impact on Med Students' Well‑Being
I'm proud to share my PhD student, Nóra Árvai's, newest paper: Generative AI's Impact on the Mental Health of Medical Students: Scenario Analysis, just published in JMIR. She used scenario analysis to examine potential futures at the crossroads of generative AI...
AI Frontier Narrows to Anthropic‑OpenAI Duopoly, Google Scrambles
We started with five frontier AI labs. Meta is seemingly out of the running. XAI just dissolved into SpaceX. We may be heading toward a duopoly: Anthropic and OpenAI. With Google fighting to stay ahead.
Top GitHub Repos for GPU‑Friendly LLM Fine‑Tuning
Best GitHub repos for fine-tuning LLMs without melting your GPU: 1. Unsloth https://t.co/REbCPgmlK3 2. Axolotl https://t.co/b0Osr4PMx8 3. LLaMA-Factory https://t.co/yKeuyoqR6N 4. PEFT https://t.co/1FMwFISV0J 5. TRL https://t.co/vGIl6ym08G 6. Torchtune https://t.co/Klex5Lj2iC 7. LitGPT https://t.co/CASALLseUl 8. Mergekit https://t.co/pTjshimji2 9. SWIFT https://t.co/2gpvJgaVvt