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:
Microsoft Brings C++ Smarts to GitHub Copilot in Visual Studio Code
Microsoft announced that GitHub Copilot in Visual Studio Code now leverages C++ symbol context and CMake build configuration awareness. The new capabilities are delivered through the C/C++ DevTools extension, enabling the AI agent to retrieve definitions, references, and call hierarchies at the symbol level. Copilot can also build, test, and list CMake targets using the active configuration, aligning AI actions with developers’ existing toolchain. These enhancements aim to make AI‑assisted C++ workflows more accurate and performant.
AI Shrinks Education Skill Gap by 75% on Tasks
New randomized experiment shows AI narrows skill gaps. We found this among talent levels at the same job, but this paper looks at education. They find that AI reduces the gap between more & less educated people on a business...
Scaling Config Security, Mapping AI, Token‑Efficient Frameworks
Seroter Daily Reading List – February 23, 2026 (#727): Today’s links look at how to safeguard dynamic config changes at scale, how to teach AI to read a map, and which web frameworks are the most token-efficient for AI agents. https://t.co/qg47lXCkK1
Solicitor Faces Probe After Putting Client Documents Into ChatGPT
The Upper Tribunal warned lawyers after a solicitor uploaded client emails and Home Office decision letters to ChatGPT, breaching confidentiality and legal privilege. Judge Fiona Lindsley highlighted a surge in fictitious case citations that waste tribunal resources and erode public...
Open Letter: Say No to Suno
The Music Artist Coalition’s open letter denounces AI music startup Suno for mass‑producing unlicensed tracks that dilute artist royalties and enable streaming fraud. It highlights Suno’s output of roughly 7 million tracks per day and Deezer’s finding that 85 % of AI‑generated...
Comparing Nuclear Dawn to Today's AI Revolution
Nukesletter is back on Substack: I share a short essay I wrote on the analogies often drawn between the dawn of the nuclear age and the current conversations around transformation AI. There are limits, but some useful comparisons. https://t.co/7sGgUOuBsW
Ban AI Human Impersonation to Test Global Regulation
There are nearly no good reasons for an AI to ever impersonate a human. Making that illegal in as many countries as possible, and improving state capacity to enforce it, would be a good trial balloon for human civilization's ability to...
Beyond Teacher Tools: Exploring AI for Student Success
Education technology firm TrekAi hosted a webinar highlighting the shift from teacher‑focused AI tools to student‑centered solutions. By allowing students to interact freely with AI during independent study, the platform captures authentic comprehension gaps and organic questions, generating real‑time actionable...
Deploying Open Source Vision Language Models (VLM) on Jetson
NVIDIA’s Cosmos Reason 2B vision‑language model can now be deployed on the Jetson family using the vLLM inference engine. The tutorial walks through installing the NGC CLI, pulling FP8‑quantized weights, and running device‑specific Docker containers for AGX Thor, AGX Orin and Orin Super Nano. After...
Predictable AI: Announcing the January and February Validated Model Batches
Red Hat announced the January and February 2026 batches of validated AI models alongside the Red Hat AI 3.3 release. The collections feature frontier‑class reasoning, multimodal, and NVFP4‑quantized models packaged as ModelCar OCI containers for seamless deployment. Validation delivers hardware‑specific performance baselines, integrity...
Microsoft’s AI Ambitions Could Bring Trouble
Microsoft is embedding its AI assistant Copilot deeply into Windows 11 and Microsoft 365, positioning the feature as a core productivity tool. The FTC has opened an antitrust investigation, probing whether the bundling of AI with the operating system and...
Red Hat AI Enterprise: Bridging the Gap From Experimentation to Production Scale
Red Hat AI Enterprise is now generally available, delivering a unified AI platform built on OpenShift that spans the entire model lifecycle from development to high‑performance inference. The solution targets the long‑standing production gap by letting organizations move from proof‑of‑concept...
Flywire Corp (FLYW) Q4 2025 Earnings Call Transcript
Flywire reported Q4 2024 revenue of $112.8 million, a 17.4% year‑over‑year increase, and adjusted EBITDA that rose roughly 700 basis points, meeting the midpoint of its guidance. The company highlighted strong growth in its travel vertical—over 50% organic expansion making it...
DigitalOcean Holdings Inc (DOCN) Q4 2025 Earnings Call Transcript
DigitalOcean reported Q4 2024 revenue of $205 million, a 13% year‑over‑year increase, and lifted net dollar retention to 99%, with core cloud services hitting 100% retention for the first time since mid‑2023. Higher‑spend "Scalers+" customers grew 17% in count, contributed 22%...
Agentic SDLC: GitLab and TCS Deliver Intelligent Orchestration Across the Enterprise
GitLab and Tata Consultancy Services (TCS) have teamed up to deliver an Intelligent Orchestration layer that embeds AI agents into the full software development lifecycle. The partnership leverages GitLab’s Duo Agent Platform and a TCS Center of Excellence to standardize...
AI Infrastructure Cost Optimization for Scaling Teams
In 2026 AI leaders face a shift from building models to managing the cost of running them at scale. Fragmented cloud stacks create hidden expenses through data egress, idle compute, and the engineering "glue" needed to keep services synchronized. Upsun...
Beyond Simple API Requests: How OpenAI’s WebSocket Mode Changes the Game for Low Latency Voice Powered AI Experiences
OpenAI’s new Realtime API introduces a WebSocket‑based mode that streams audio directly to GPT‑4o, collapsing the traditional STT‑LLM‑TTS chain into a single, stateful connection. The protocol delivers full‑duplex communication, allowing the model to listen and speak simultaneously while maintaining session...
X Plans to Combat AI-Generated Content While Promoting Grok
X announced new anti‑AI‑bot measures, including user‑added tags for AI‑generated posts and upgraded detection tools, while simultaneously promoting its Grok chatbot across the platform. Head of product Nikita Bier emphasized the need to protect the “human pulse” of X from synthetic...
What Do Community Bankers Want? What Do Community Banks Need?
The 2025 CSBS Annual Survey reveals community banks are feeling heightened pressure from non‑bank competitors, especially in payments and wealth‑management services, while deposit‑rate volatility forces more aggressive pricing. Cybercrime emerges as the top internal risk, with card and identity fraud...
How to Build a Production-Grade Customer Support Automation Pipeline with Griptape Using Deterministic Tools and Agentic Reasoning
The MarkTechPost tutorial shows how to construct a production‑grade customer‑support automation pipeline with Griptape, combining deterministic Python tools and an LLM‑driven agent. Custom tools handle PII redaction, ticket categorization, priority scoring, SLA assignment, and escalation payload creation before any language...
Built an AI MVP in Just 30 Minutes
Facilitated an AI MVP workshop and here is what we built in 30 mins 👌🏾
WinDC Partners with Yirigaa for AI Infrastructure Deployment
WinDC has teamed up with Indigenous‑led MSP Yirigaa to roll out modular, high‑density AI data centres powered by renewable energy at generation sites across regional Australia. The partnership couples WinDC’s zero‑scope‑2 emissions compute modules with Yirigaa’s ISO‑27001‑certified workforce, delivering training,...
Two-Thirds of Independent Agencies Plan to Increase AI Use This Year, Survey Says
A new ACT survey shows that more than two‑thirds of independent agencies intend to increase AI use within the next 12 months, yet only 8% currently have AI embedded in daily workflows. Operational efficiency (60%) and staff productivity (52%) drive...
AI Writes SQL, but Fundamentals Lift Your Ceiling
Yes, AI can write the SQL. But do you understand: * Why that join works? * Why that model makes sense? * Why that metric matters? AI lowers the barrier. Foundations raise your ceiling.
SWE‑Bench Verified Flawed Tests Reveal Data Leakage Issues
Am currently putting together an article, and yeah, the SWE-Bench Verified numbers are definitely a bit sus across all models -- the benchmark suggest they are more similar than they really are. So, I went down a rabbit hole looking into...
American AI Industry Trembles as Deepseek Prepares to Release New Model
Chinese AI startup DeepSeek is set to launch its V4 model, following the market‑shaking debut of V3 last year. V3’s release knocked the Nasdaq composite down 3% and sent Nvidia shares tumbling 17%, wiping out roughly $600 billion in value. Analysts...
Anthropic Says DeepSeek, Moonshot, and MiniMax Used 24,000 Fake Accounts to Rip Off Claude
Anthropic disclosed that three Chinese AI labs—DeepSeek, Moonshot AI and MiniMax—used roughly 24,000 fraudulent accounts to conduct over 16 million interactions with its Claude models, targeting reasoning, coding and tool‑use capabilities. The coordinated distillation attacks extracted large‑scale training data, effectively stealing...
More Than Dashboards: AI Decisions Must Be Provable
Enterprise leaders are demanding proof of what AI systems actually did, not just what they were designed to do or what dashboards report. As AI moves into regulated, high‑risk environments, boards and auditors expect decision‑level evidence for each action. The...
Sharon AI, Cisco and NVIDIA Bring AI Factory to Australia
Sharon AI, Cisco and NVIDIA are launching Australia’s first Cisco Secure AI Factory, a sovereign AI infrastructure that keeps data and processing within the country. The facility will run on Cisco UCS servers, Nexus Hyperfabric, and 1,024 NVIDIA Blackwell Ultra...
Seedance 2.0 Videos Feel Like a Nightmare Dreamscape
I’ve been watching Seedance 2.0 videos so you don’t have to and they are a nightmare dreamscape https://www.techradar.com/ai-platforms-assistants/ive-been-watching-seedance-2-0-videos-so-you-dont-have-to-and-they-are-a-nightmare-dreamscape
From Clicks to Impactful Visibility: SEO’s 2026 Evolution
#SEOFOMO x Berlín SEO & Content Club: SEO, AI Search and the New Playing Field in 2026 - An excellent recap by @mjcachon going through what was discussed last week at the event: * The Real Shift: From Clicks to...
Chinese Ship Completes First Autonomous Docking and Cargo Handling
Four years after launching the world’s first autonomous electric feeder, China’s 300‑TEU vessel Zhi Fei completed its first fully autonomous docking at Qingdao Port, completing the maneuver in just 30 seconds. The ship, capable of crew‑managed, remote‑controlled or unmanned navigation, operates...
Three AI Glasses Tiers: Ultra-Light, Middle, Heavyweight Picks
I did a live stream over on LinkedIn this morning on another show and saw @auradeluxe asked me what my latest thinking on AI-driven glasses are. For me, it splits up into three different categories: 1. Ultra-lightweights Where...
Before We Predict The End Of Lawyers, Let’s Take A Deep Breath
Generative AI is being touted as a productivity breakthrough for law firms, but the author argues it may actually increase workload. Drawing on the Solow paradox, the piece suggests that more information leads to more analysis, verification, and billable hours...
Before We Predict The End Of Lawyers, Let’s Take A Deep Breath
Stephen Embry argues that generative AI, while touted as a productivity miracle for law firms, may actually generate more work rather than reduce it. He frames this counter‑intuitive outcome with the Solow paradox, which observes that new technologies often raise...
How GPs and LPs View AI Disruption: Thoma Bravo’s Holden Spaht and New York’s Anastasia Titarchuk
At the PEI Group’s NEXUS 2026 summit, Thoma Bravo partner Holden Spaht and New York‑based LP specialist Anastasia Titarchuk discussed how artificial‑intelligence disruption is reshaping private‑equity dynamics. Both highlighted accelerating deal flow in AI‑enabled software and the need for deeper technical diligence....
Ottonomy Unveils Platform Connecting Robots, Drones and Smart Infrastructure on NVIDIA
Ottonomy unveiled Ottumn.AI, a cloud‑based platform that orchestrates autonomous robots, drones, and smart infrastructure across healthcare, manufacturing, and commerce. The solution runs on NVIDIA AI hardware, leveraging Jetson edge devices, Isaac Sim digital twins, and cloud GPUs for sub‑30 ms decision...
New AI Agent Logs Directly Into College Platform Canvas to Do Your Homework for You
Companion.AI unveiled "Einstein," an autonomous AI agent that logs into Canvas, watches lectures, writes essays, participates in discussions and submits assignments on a student’s behalf. The startup markets the tool as a full virtual computer that can perform any browser‑based...
Why Connecting AI Everywhere Fails, Embrace Autonomous Agents
Built an MCP server recently? Did developers actually use it, or is it collecting dust? 😵 I'll be speaking at @theleaddev London (June 1-2) about why "connecting AI to everything" doesn't work and what actually does when building tools that...
AI Winners Will Be Infrastructure, Not Software
AI winners aren’t who you think. Dell’s margin shift changes the AI narrative. Infrastructure, not software, may drive the next leg of leadership. Most investors are looking the wrong way. Read this: https://t.co/nY4GbIKIfM @graniteshares @willrhind
AI Instantly Uncovers Price‑move Catalysts in Seconds
I just recorded a new @Deepvue Tutorial discussing one of our key AI updates🪄 Now traders can research catalysts and the context behind a gap or significant price move in seconds Here's how.... and it feels like magic 👇 https://t.co/meg4B1IrUQ
AI Governance, Trust, and Safety Drive Future Innovation
The new frontier is not just adopting AI. It is mastering its governance. The companies that prioritize trust and safety will be the ones that lead the next decade of innovation. https://t.co/MqiUbVHQVO
Clarity Beats Uniform AI: Define Your Customer Identity
If everyone has the same AI, AI isn’t the advantage. Clarity is. Who are you to your customers? : #AI #Strategy #Leadership #Business #Innovation https://t.co/JZtALylZ7J
AI Ambition Outpaces Governance, Threatening Shareholder Value
74% of firms will deploy agentic AI by 2028. Only 21% have mature governance. A 53-point gap between ambition and readiness. That is a ticking time bomb for shareholder value. https://t.co/TsiQpfm0tL
Cyabra Mislabels Real Palestinian Accounts as Bots
If we want to talk about manipulation I think it's important to note Cyabra (company POLITICO's story is based on) is an Israeli AI startup that does not have a good record of identifying who is and who isn't bots....
Anthropic Fined Billions for Massive Training Data Theft
Anthropic is guilty of stealing training data at massive scale and has had to pay multi-billion dollar settlements for their theft. This is just a fact.
AI Surge Makes Software Competition Easier, Threatening Companies
“I’m more bullish on AI than I’ve ever been. And that’s exactly why I’m bearish on most software companies. Not because their customers will leave, but because their next thirty competitors just got a lot easier to build.”
Old AI Giant Falls as Young Startup Takes Over
IBM, a 115-year-old company, pioneered AI with Watson a decade ago. Today, a 3-year-old startup (Anthropic) just wiped $28 billion off IBM’s market cap. Mercy is for the weak.
AI's Task-Specific Skills Don't Equal General Intelligence
The field of AI is still struggling with the fact that task-specific skill is not the same as general intelligence
OpenAI Teams Up With Consulting Firms for Enterprise AI
🚨 ANALYST TAKE: OpenAI just partnered with the very management consulting firms AI is supposed to replace. Here’s why this is NOT really a contradiction. It’s actually one of the key endgames of enterprise AI. Thread 🧵👇 https://t.co/wTKfR4kW6t