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:
Chinese Startup Claims It Developed an AI Collar that Translates Human Words Into Barks and Meows
Chinese AI startup Meng Xiaoyi unveiled PettiChat, a 27.2‑gram smart collar that claims to translate dog barks and cat meows into human language in real time. Powered by Alibaba Cloud’s Qwen model trained on 1.5 million pet audio samples, the device reports 94.6% contextual accuracy for cats and 92.3% for dogs. Priced at 799 yuan (≈ $119), it includes GPS tracking, a companion app, and a one‑hour battery charge that supports up to 1,000 sentence translations. The company says it has secured more than 10,000 pre‑orders.
Anthropic May Keep Supplying Claude to the NSA Despite Being Flagged as a Supply Chain Risk by the Pentagon
Anthropic, flagged by the Pentagon as a supply‑chain risk, will likely continue providing its Claude‑based Mythos model to the NSA after White House Chief of Staff Susie Wiles gave personal approval. The agency lacks enough Nvidia Grace Blackwell chips, making...
The Safety Company Just Rented the Cowboy's GPU Empire. Welcome to Compute Feudalism.
Anthropic signed a lease for 220,000 NVIDIA GPUs housed in a Memphis data center built by xAI and operated by SpaceX, drawing over 300 MW of power. The move was forced by chronic compute shortages that throttled Anthropic’s AI services for...
Machine Learning System Design Interview #36 - The False Positive Blindspot
The post warns that ROC‑AUC can be deceptive on extreme class‑imbalanced data, illustrated by an OpenAI interview where a model with 0.98 ROC‑AUC floods production with false positives. It explains that the massive true‑negative count masks a high false‑positive rate,...
Researchers Let Claude Code Discover AI Scaling Algorithms that Humans Probably Wouldn't Have Designed
Researchers introduced AutoTTS, an offline simulation framework where Claude Code autonomously discovers test‑time scaling algorithms for large language models. By treating width (parallel paths) and depth (path length) as a shared control space, the agent generated a high‑level controller that...
How DeepSeek AI Uses 90% Fewer Tokens to Match Billion-Dollar Models
DeepSeek AI introduces a novel pointing mechanism for visual reasoning that slashes visual token usage by 90% while preserving accuracy. By distilling multiple expert models into a single unified student, it matches the performance of billion‑dollar AI systems with far...
ACAS X Is Already in the Cockpit. The AI-and-ATC Debate Is Three Years Too Late. Part 1.
The FAA has already certified the AI‑augmented Airborne Collision Avoidance System X (ACAS X), which is now operating on commercial aircraft and can issue pilot‑directed maneuvers that override ATC commands. ACAS X’s machine‑learning decision logic was developed at MIT Lincoln Laboratory and ratified...
Deep Generative Molecular Design and Its Value in Modern Drug Discovery (Paper Feb 26)
The February 2026 review maps the rapid rise of deep generative molecular design in drug discovery, arguing that AI is moving from virtual screening toward the creation of new, testable drug hypotheses. It categorises the field into three technical families—graph‑based...
The Finanser’s Week: 18th May – 24th May 2026
The Finanser’s weekly roundup (May 18‑24, 2026) spotlights four blog posts that dissect current banking trends. One piece debunks sensational AI headlines, emphasizing that banks are using artificial intelligence to augment, not replace, core functions. Another explores the evolving, invisible nature of...
Alphabet Bets On AI Glasses And Cheaper Gemini To Deepen Ecosystem
Alphabet unveiled Android XR AI smart glasses at Google I/O 2026, marking its direct entry into the competitive wearables market against Meta and Apple. The launch is backed by new distribution deals with Samsung, Gentle Monster and Warby Parker, widening...
“The System Will Tell Us”: United Airlines Using AI To Help Passengers Make Flight Connections
United Airlines has launched ConnectionSaver, an AI‑driven decision tool that helps gate agents determine whether to hold aircraft for late‑connecting passengers. The system evaluates crew schedules, aircraft routing, ground‑hold limits and arrival windows, and has already facilitated 54,000 extra connections...
The Smallest Reliable Working Set
AI agents suffer from context overload much like software experiences memory leaks. Adding more RAM or expanding a model’s context window often masks the real issue: too much irrelevant information in the working set. Effective agents keep prompts focused, streaming...
Can AI Fix the NHS Workforce Crisis?
In this episode, host James chats with Ash, CEO of Patchwork, about the NHS’s staffing crisis and how AI‑driven, preference‑based rostering could help retain clinicians and cut temporary staffing costs. They explain how traditional fixed‑pattern rotas ignore individual work‑life preferences,...
🔮 Exponential View #575: AI’s Math Breakthrough and Its Creative Limits
OpenAI’s new reasoning model cracked an 80‑year‑old conjecture in discrete geometry by forging an unexpected link to algebraic number theory. The breakthrough mirrors AlphaGo’s surprise moves, highlighting AI’s capacity to traverse isolated scientific domains. In parallel, a Nature‑published multi‑agent system...
AI Set to Disrupt up to 28% of Back-Office and Data Roles in India over the Next Three Years: Report
The SHRM India Skill Intelligence Report warns that AI could disrupt 28% of back‑office roles, 24% of data and reporting functions, and 21% of customer‑service jobs over the next three years. About 45% of surveyed firms cite AI and digital‑skill...
AI Blood Test Predicts Stroke, Heart Failure and More Up to 15 Years Early
Researchers at the University of Hong Kong have launched CardiOmicScore, an AI‑driven blood test that predicts six major cardiovascular diseases up to 15 years before clinical onset. The tool outperforms traditional polygenic risk scores by integrating genomics, proteomics and metabolomics,...
Google Deploys May 2026 Core Update and AI Mode Overhaul Amid I/O
Google began rolling out its May 2026 core algorithm update on May 21, coinciding with a major redesign of the Search box and the first official usage data for AI Mode. The changes, announced at I/O, introduce Gemini 3.5 Flash...
Carrefour Unveils €5 Bn AI‑focused 2030 Plan, Adds ChatGPT Grocery Service and Smart Shelves
Carrefour announced its Carrefour 2030 strategy on Feb. 18, 2026, targeting €5 bn of cumulative free cash flow and rolling out generative‑AI tools, a ChatGPT‑based grocery service for 26 million French users, and smart‑shelf technology via Vusion. The moves aim to sharpen price...
Starbucks Pulls NomadGo AI Inventory System From 11,000 Stores After Scan Errors
Starbucks announced on May 19, 2026 that it is retiring the NomadGo AI inventory system from all 11,000 U.S. and Canadian stores. The tool, touted for 99% accuracy and eight‑fold speed gains, forced workers to double‑count items, prompting a full...
Anker's Liberty 5 Pro Earbuds Debut with First Consumer Neural‑Net AI Chip
Anker Innovations introduced the soundcore Liberty 5 Pro and Liberty 5 Pro Max earbuds, priced at $170 and $230, featuring the proprietary Thus neural‑net compute‑in‑memory AI chip. The launch marks the first consumer device to ship with a CIM‑based processor...
SAP Names David Robinson President of North America to Accelerate Cloud Growth
SAP announced David Robinson as President of its North America business, giving him oversight of sales and customer success across the region. The move is intended to speed cloud adoption and AI‑enabled outcomes for SAP’s largest market.
Wayve Teams with Stellantis to Deploy AI‑Driven Hands‑Free Driving in Jeep by 2028
London‑based AI startup Wayve has signed a deal with Stellantis to embed its end‑to‑end neural‑network driver into Jeep and other Stellantis brands, targeting a Level 2++ hands‑free launch in North America in 2028. The partnership leverages Wayve’s $1.2 billion Series D round and...
UnitedHealth Invests $1.6 B in AI, Launches Avery Chatbot for 6.5 M Members
UnitedHealth Group announced a $1.6 billion AI investment and the rollout of its generative‑AI chatbot Avery to 6.5 million members, with a goal of serving 20 million by year‑end. The moves helped lift the stock 53% since March and lowered the medical benefit...
ShengShu Technology Launches Vidu Claw AI CMO, Promising Ads at 1% of Traditional Cost
ShengShu Technology introduced Vidu Claw, an AI‑driven chief marketing officer that can turn a brief into a finished ad in hours, cutting production costs to roughly 1% of conventional methods. The platform uses a subscription model that charges for completed...
NVIDIA's Vera CPU Offers 50% Faster Per‑Core Performance, Boosting Agentic AI Labs
NVIDIA delivered its new Vera CPU, featuring 88 custom Olympus cores and a claimed 50% per‑core speed boost, to Anthropic, OpenAI and SpaceXAI. The hardware targets the rising compute demands of agentic AI, and Oracle Cloud plans large‑scale deployments.
OpenAI Launches $222 Million Singapore AI Lab as Nation Tightens Agentic AI Rules
OpenAI announced its first Applied AI Lab outside the United States in Singapore, backed by a commitment of more than S$300 million (about $222 million). The lab will create over 200 technical roles and serve as a hub for forward‑deployed engineers, coinciding...
Google Report Finds LLM‑Generated Malware Evades Detection, Threatens 2FA Defenses
Google's latest research shows that large‑language‑model (LLM) generated malware is increasingly difficult to detect and can bypass two‑factor authentication. The findings signal a shift toward more automated, scalable cyber threats and have prompted security firms to reassess defensive tools.
Current AI Falls Short of True AGI Criteria
AGI is certainly not here by the definitions I have repeatedly laid out. or by criteria that @hendrycks @Yoshua_Bengio and I and others recently laid about at https://t.co/ogf90ZCVKs i don’t think any current AI can even reliably do any of the...
GBrain Sets New SOTA, Beats MemPalace and Vector RAG
My newest gbrain-evals just dropped - this is how gbrain does vs other options. https://t.co/yRWm72QEgf is SOTA for reranking and embedding cost, speed, and retrieval success. GBrain beats MemPalace by 1% on LongMemEval and beats Vector RAG by 38% https://t.co/nIk93HFTHb https://t.co/RIzXFkfFp8
AI Systems Slash Drug Discovery Time, Yield New Cancer and Neurodegenerative Candidates
FutureHouse’s Robin and DeepMind’s Co‑Scientist AI agents have collapsed drug‑candidate timelines by up to 200‑fold, while NIH’s TRANSLATE‑AI has identified three repurposed drugs for Alzheimer’s and Parkinson’s with 98% pre‑clinical accuracy. The breakthroughs promise faster routes to clinical trials for...
Mira Murati Leaves OpenAI, Launches Thinking Machines Lab with $2 B Seed Funding
OpenAI’s chief technology officer Mira Murati announced her departure in September 2024 and unveiled Thinking Machines Lab in February 2025. The new venture closed a $2 billion seed round at a $12 billion post‑money valuation, drawing senior talent from OpenAI, Anthropic and...
Accenture Teams with Saudi AI Firm HUMAIN to Scale Enterprise AI in Kingdom
Accenture and Saudi AI specialist HUMAIN announced a strategic collaboration in Riyadh on May 20 to accelerate enterprise‑wide AI adoption across the Kingdom. The partnership blends HUMAIN’s local AI infrastructure with Accenture’s global transformation expertise, aiming to shift organizations from...
Columbia Lloyds Deploys ZestyAI Platform to Boost Homeowners Underwriting in Weather‑Heavy States
Columbia Lloyds Insurance Company has selected ZestyAI’s Risk and Decision Intelligence platform to improve underwriting accuracy and property‑risk visibility for its homeowners portfolio in Texas, Oklahoma and Arkansas, marking a significant PropTech move into property insurance.
Novo Nordisk Cuts Drug Launch Time by Two‑Thirds Using AI, Expands India Hub
Novo Nordisk announced that artificial‑intelligence tools are trimming new‑drug launch cycles by as much as two‑thirds, with its Bengaluru, India centre now handling a majority of preparatory work for global rollouts. The move aims to recapture momentum in the fast‑growing...
Gartner Warns AI‑driven Strain Threatens HR Teams as Digital Adoption Accelerates
Gartner cautions that fast‑track AI deployment is heightening workplace strain for HR leaders, exposing productivity gaps, low‑quality AI output and eroding employee trust. The firm highlights premature AI‑driven layoffs and a two‑hour cleanup cost for poor AI‑generated work.
Nvidia Beats Q1 FY2027 Forecast, Market Reacts Cautiously Amid Growth Questions
Nvidia reported Q1 FY2027 earnings that topped expectations, delivering stronger revenue and earnings than forecast. Despite the beat, the stock slipped modestly as analysts raised questions about the durability of AI‑driven growth and noted a miss in compute revenue, while...
SAP to Acquire Master‑Data Leader Reltio, Boosting AI‑Ready Enterprise Data
SAP SE said it will acquire Reltio, a cloud‑native master data management provider, to embed AI‑ready data into its Business Data Cloud. The transaction, with terms undisclosed, is slated to close in the second or third quarter of 2026 and...
Fliggy Deploys AI Assistant to Streamline Hotel Operations Across China
Alibaba's travel platform Fliggy launched an AI assistant that automates inventory, pricing and invoicing for hotels, reporting a 70% lift in operational efficiency and a dramatic booking increase at a pilot property. The rollout reaches roughly 7,000 hotels and promises...
Andreessen Says AI Coding Bots Outperform Humans, Sparking HR Debate
Venture capitalist Marc Andreessen told the Joe Rogan Experience that AI coding agents can produce five to twenty times more code than human programmers and never get sick, drunk or file HR complaints. His remarks arrive amid a wave of...
Denodo Adds AWS Integrations to Power Governed Agentic AI in the Middle East
Denodo announced a suite of integrations with Amazon Web Services that embed its data‑virtualization platform into Bedrock AgentCore, SageMaker and Quick. The move aims to give AI agents trusted, real‑time access to governed data across hybrid and multi‑cloud environments, a...
Expedia Unveils AI‑Powered B2B Server to Link Partners’ Agents Directly to Travel Inventory
Expedia Group announced that its new Model Context Protocol (MCP) server, part of an Intelligent Experience Platform, will go live in the next few months. The AI‑enabled infrastructure will let partner travel agents query Expedia’s lodging, flight, car and activity...
Unveiling Treatment Timelines in Gliomas via AI
Researchers applied causal machine learning to map how radiotherapy and chemotherapy affect lower‑grade glioma patients over time. By integrating longitudinal clinical data and adjusting for confounders such as age and tumor genetics, the models uncovered distinct temporal windows where each...
Commodity Intelligence
The author argues that today’s large language models have become "commodity intelligence," delivering homogenized, index‑fund‑like knowledge rather than differentiated insight. By comparing AI knowledge to commodified assets such as coal or gold, he shows how over‑generalized data erodes personality and...
Free Extension Reveals Claude’s Exact Token Usage
A solo developer got so frustrated with Claude's rounded usage percentages that he built a free browser extension that reads Anthropic's own internal data and shows you the exact numbers Anthropic does not. It's called Claude Counter. A minimal extension that...
AI‑Driven ‘Steroid Olympics’ Redefines Creative Human Potential
A surge of AI‑generated works—from a contested Granta short story to the fully AI‑made Cannes film “Hell Grind”—has ignited a cultural clash over what counts as genuine human creativity. Experts warn the flood of low‑quality output will pressure creators, while...
FinTech Farther Raises $150 Million Series D to Accelerate AI‑Driven Wealth Management
Farther announced a $150 million Series D round led by General Atlantic, earmarked for scaling its AI‑powered wealth management platform. The funding comes as the firm reports $23 billion in recruited assets and aims to triple its year‑over‑year growth, signaling deeper tech...
McKesson Posts $403B Record Revenue as AI Revamps Distribution Network
McKesson Corp. announced full‑year 2026 consolidated revenue of $403 billion, a 12% rise, while embedding artificial intelligence across its distribution, oncology and biopharma services. The AI rollout, highlighted by a new Montreal hub and a unified planning platform, helped generate $6.2 billion...
Anthropic Launches Claude for Legal to Embed AI in Law Firm Workflows
Anthropic introduced Claude for Legal on May 22, 2026, a version of its Claude chatbot built for law firms and corporate legal teams. The tool promises to automate document review, research and drafting, but immediately raised data‑privacy and professional‑responsibility concerns...
JetBrains Warns Enterprises of AI‑tool Vendor Lock‑in, Claims Sole Independence
At Google Cloud Next, JetBrains VP Mikhail Vink warned that most AI coding assistants are now tied to a single cloud provider, positioning JetBrains as the sole independent vendor. The company backs its claim with a 16 million‑user base, 300,000 commercial...
Google Unveils Gemini‑Powered AI Glasses at I/O, Targeting Meta’s Wearable Lead
Google announced its Gemini‑driven AI smart glasses at the I/O developer conference, previewing a prototype that blends audio interaction with an in‑lens display. Developed with Warby Parker, Gentle Monster and Samsung, the devices are slated for a fall 2026 release...