Today's AI Pulse
Salesforce to acquire AI‑agent platform Fin for $3.6B
Salesforce announced a $3.6 billion purchase of Fin, the AI‑agent platform formerly known as Intercom, to strengthen its Agentforce suite. The deal comes as Salesforce reported strong Q1 results and a $50 billion buyback program. Fin’s technology enables automated customer support across chat, email, WhatsApp, phone and Slack.
Also developing:
By the numbers: Waymo acquires Apple’s Arizona Proving Ground for $220M
637: The AI Automation Stack Behind The Fastest Growing Ecommerce Brands With Leo Sgovio
In this episode, Leo Segovia reveals his AI‑driven automation stack that handles influencer outreach, video ad creation, and product launches without giveaways or paid ads, enabling him to sell 5,000 units before Christmas. He explains how his system scrapes and scores creators using 114 data points, then uses bots to send personalized messages at optimal times, dramatically improving response rates. Leo also discusses cloning winning competitor ads with AI tools and shares his three‑phase influencer strategy for brands of different sizes. His background includes leading innovation at Viral Launch and building marketing automation platforms like ComboMat and Spliced.io.
ABS, HD Hyundai and Anduril Industries to Advance Autonomous Surface Vessels
ABS, HD Hyundai and Anduril Industries have signed a memorandum of understanding to develop autonomous surface vessels. The agreement creates an end‑to‑end pipeline covering ship design, construction, AI‑driven autonomy and classification certification. ABS will provide classification and technical advisory services,...
AI Chatlogs Promise Discovery Risk, Windfall
Judges across several federal districts have begun issuing rulings that treat AI‑generated chat logs as discoverable evidence in civil litigation. The emerging jurisprudence forces law firms to preserve and produce interactions with tools such as ChatGPT, Claude, and Gemini, creating...
Toyota Reveals Grand AI Vision for Vehicles and Beyond
Toyota’s software arm Woven by Toyota introduced the AI Vision Engine, a large‑scale vision‑language model that fuses visual, behavioral and environmental data. The engine will underpin the next‑generation Anzen driver‑assist suite and the Arene software platform, enabling hardware‑agnostic updates for...
Dyna Software's AI Assistant Promises to Massage Your Toughest ServiceNow Configs
Dyna Software unveiled Platform Copilot, an agentic AI that lets business users configure ServiceNow applications through natural language or uploaded form images, bypassing the need for developers. The tool reads a customer’s instance schema, generates wireframes, validates changes, and builds...
LightSeek Foundation Releases TokenSpeed, an Open-Source LLM Inference Engine Targeting TensorRT-LLM-Level Performance for Agentic Workloads
The LightSeek Foundation has launched TokenSpeed, an MIT‑licensed open‑source LLM inference engine built for agentic coding workloads. Its architecture combines a compiler‑backed SPMD modeling layer, a C++ finite‑state‑machine scheduler that enforces KV‑cache safety, and a pluggable kernel system that works...
Your Showcase Primer: Specter, Flow Engineering, General Matter
The May 20th San Francisco Startup Showcase highlighted three founders tackling the physical‑world bottlenecks that limit AI‑driven growth. Xerxes Libsch’s Specter adds a low‑cost, wireless sensor mesh to give enterprises real‑time visibility of critical infrastructure. Pari Singh’s Flow Engineering layers AI agents...
Kodiak Robotics Secures $100M to Scale Driverless Trucks
Physical AI has moved from "if" to "how quickly" and @KodiakRobotics is hitting that next gear. With a $100 million capital raise, 28 driverless trucks already in our customer's hands, and new developments across the long haul, defense and industrial verticals,...
Nvidia Vera Rubin Used by Google Could Next and Thinking Machines Lab
NVIDIA unveiled its Vera Rubin platform, pairing the Rubin GPU with the Vera CPU to deliver a full‑stack AI system that claims ten‑fold better inference performance per watt than the Blackwell generation. At Google Cloud Next, the company announced the...
Design for Agentic Scale: Speed, Power, Governance
“Your most important user no longer has a pulse.” My colleague, Yasmeen Ahmad, said this at Google Next. Not to be provocative, but because it’s true. Agents don’t work like humans. They can simultaneously command dozens of API calls. They can...
OpenAI Debuts a Codex Plugin for Chrome
OpenAI released a Chrome extension for its Codex platform, enabling AI‑assisted coding directly in the browser. The plugin can test web applications, pull context from multiple open tabs, and operate alongside Chrome DevTools. It runs on both Windows and macOS,...
AI Graveyard: 142 AI Companies so Far Buried, Many More to Come
The AI Graveyard, a side‑project of ToolDirectory.ai, now tracks 142 artificial‑intelligence tools that have either shut down or been absorbed by larger products. The list highlights a rapid churn rate as the AI boom fuels a flood of startups, many...
The AI Industry Is Where Banking Was in 2006. (We're Hiring)
CeSIA, the French Center for AI Safety, is hiring three senior staff—Head of Policy Analysis, Head of Communications, and Operations & Executive Associate—by May 22, 2026, with options for remote work across the EU or UK. The organization aims to shift AI...
AI Reads Retina to Diagnose Multiple Metabolic Diseases
In 2014 I predicted we'd diagnose metabolic disease just by looking into your eyes. Nature Medicine just published the proof. Researchers built an AI called Reti-Pioneer. You look into a camera. It scans your retina. No blood draw. No needles. No lab. What it can...
OpenAI Makes Its Rival to Anthropic's Mythos More Widely Available to Cyber Defenders
OpenAI has launched a limited preview of GPT-5.5-Cyber, a less‑guarded version of its GPT-5.5 model, for vetted cyber‑defenders protecting critical infrastructure. Early testing shows the model performs on par with Anthropic’s Mythos in finding and exploiting software bugs, with Mythos...
Agentic Coding Powers Instant Data Visualizations, UIs, CLIs
A few major use cases for agentic coding for me: 1. Adhoc data visualizations. Anytime I have a question that can be answered quantitatively, I generate some code to make a plot. 2. Adhoc data annotation UIs. In ML, "make your own...
CERN: Smarter Decisions at the Speed of Collisions
CERN’s ATLAS and CMS experiments are embedding machine‑learning models directly into their Level‑1 trigger hardware to cope with the High‑Luminosity LHC’s unprecedented data flow. By converting AI algorithms into FPGA firmware, the detectors can evaluate up to 40 million collisions per...
Gmail Adds AI Inbox to the Bottom Bar as Help Me Write Becomes More Personal
Google has extended its AI Inbox feature to Gmail’s Android and iOS apps, placing a dedicated tab in the bottom navigation bar alongside Mail, Chat and Meet. The mobile AI Inbox is currently in beta and limited to Google AI Ultra subscribers,...
Meta's AI Agent Plans Reportedly Include an OpenClaw Competitor that Can Shop on Instagram
Meta is developing an AI agent called Hatch, modeled after the open‑source OpenClaw platform, to embed autonomous shopping capabilities within Instagram and other Meta apps. The agent has been trialed on simulated versions of services such as DoorDash, Reddit and...
AlphaFold’s Surprising Revelation: AI Shifts Bottlenecks, Not Workers
AlphaFold’s Nobel‑winning protein‑fold predictions sparked fears that AI would replace structural biologists, echoing broader job‑pocalypse anxieties. A new NBER study finds experimental structure work and related publications stayed steady after the 2021 release of hundreds of thousands of AI‑generated models....
Can You Hear Me Now?
Venture capitalists are flocking to AI‑powered voice agents as the sector is projected to exceed $35 billion in revenue by 2033. Recent funding rounds have poured billions into startups that combine large‑language models with speech synthesis, targeting customer service, healthcare, and...
CEOs Deploy 20+ AI Agents, Younger Founders Hold Back
For the last few months I’ve been interviewing CEOs about their personal AI stacks 🤖. What helps them run the company better / faster / smarter/ The most AI-pilled ones are running 20+ agents that basically run their entire day....
Ollama Vulnerability Highlights Danger of AI Frameworks with Unrestricted Access
Researchers at Cyera uncovered a critical flaw (CVE-2026-7482) in the popular Ollama AI framework that enables unauthenticated attackers to trigger an out‑of‑bounds heap read via a crafted GGUF file. The bug can leak process memory—including prompts, API keys, environment variables,...
Google Wants Non-Commodity Content. These B2B Brands Show You What It Looks Like.
Google Search Central highlighted a split between commodity and non‑commodity content, declaring that only the latter thrives in AI‑driven search results. Non‑commodity pieces are defined by uniqueness, specificity, and authenticity—attributes that founder‑led B2B brands like 6Sense, Clarify and Beehiiv already...
How to Get AI to Sell For You
The post announces an AI Toolkit Bundle that helps creators make their sales pages, articles, and outreach AI‑discoverable, letting assistants like ChatGPT, Claude, and Gemini recommend them to buyers. It cites a 2026 study showing 67% of purchasers start their...
ICE Plans to Develop Own Smart Glasses to ‘Supplement’ Its Facial Recognition App
Immigration and Customs Enforcement is evaluating the creation of smart glasses to supplement its Mobile Fortify facial‑recognition app, which scans individuals and cross‑references a database of roughly 200 million biometric records. The agency argues the wearable heads‑up display would free officers’...
Docusign Intelligent Agreement Management Brings Clarity to Workflows
DocuSign introduced Intelligent Agreement Management, a SaaS platform that extends its e‑signature core with identity verification, centralized workspaces, multichannel delivery and AI‑driven analytics for healthcare providers. The solution consolidates patient intake forms, consent documents and vendor contracts into a single,...
Semafor’s New AI Tool Helped Boil Down Its Entire Flagship Conference Into Nine Takeaways
Semafor introduced Semafor Intelligence, an AI‑assisted editorial product that condenses the five‑day World Economy conference into nine key themes. Journalists fed a custom‑built AI tool with transcripts from more than 250 sessions, using vector embeddings to cluster ideas and extract...
Before You Build an AI Agent, Do This First (Most People Skip It)
Entrepreneurs often dive straight into AI tools like ChatGPT, Zapier, or n8n without a clear plan, leading to tangled workflows and broken agents. The article argues that the first step should be a simple paper sketch that defines the exact...
How to Successfully Integrate AI Into Claims Operations: Best Practices, Pitfalls, and Keys to Lasting Impact
Artificial intelligence is moving from pilot projects to enterprise‑wide use in workers‑comp claims, promising faster interventions and lower costs. Adjusters can now spot risk signals early, shifting the process from reactive to predictive and enabling more informed conversations with injured...
AI Won’t Fix Your Supply Chain. Humans Will.
The article argues that AI in logistics will not replace people but will give them more time and clarity. AI excels at spotting patterns, flagging risks early, and handling repetitive tasks, creating space for human judgment. However, decisions that affect...
Virgin Atlantic Becomes First Airline With a ChatGPT App
Virgin Atlantic has become the first airline to launch a dedicated app inside ChatGPT, letting travelers search for flights using plain‑language queries. The AI‑driven interface returns a concise list of options and then routes users to Virgin’s website or mobile...
US Tech Official Calls for ‘Transformational’ Use of AI in Scientific Discovery
White House Chief Technology Officer Ethan Klein urged a "transformational" rollout of agentic AI to accelerate scientific discovery, emphasizing tighter integration from data collection through prototyping and scale‑up. A Market Connections survey shows 53% of federal agencies are already exploring...
Teradata Unveils New Product, Exec Says Its ‘Ready to Move Beyond AI Pilots’
Teradata chief product officer Sumeet Arora announced a new AI‑centric offering built on the Vantage cloud data platform. The solution, dubbed Vantage AI Ops, promises to move enterprises beyond experimental pilots to production‑grade AI workloads. Early customers include major banks...
The AI Media Arena: Journalism Is The Biggest Source For Query Answers
A Muck Rack study of over 25 million AI‑generated links finds that 99% of citations come from non‑paid sources, with journalism leading at 27% of all references. Corporate blogs follow at 24%, while aggregators, owned media and government sites make up...
Claude Code's Downtime Reveals Frustrating Unreliability
Something going on with my Claude Code right now, it just took my site down (second time in 12 months, good score), couldn't fix it by itself, I had to manually fix it, I did within 5 seconds, so that's...
Easy Management-Fee Gains on Anthropic Face Market Retribution
1. Anthropic may be a $5T co in a couple years. 2. Never in the history of markets has ridiculously easy money (a decade’s worth of income surpassed by management fees brokering a deal for Anthropic shares you sourced via tweet)...
Gemini 3.1 Flash‑Lite Launches: Fast, Affordable Premium AI
Gemini 3.1 Flash-Lite is GA and ready to use. It's a fast and cost-effective way to access premium AI. https://t.co/ELGpnllkIv
Enterprise AI Redefines System Integrators: Key Customer Questions
I’ll be talking about this very topic tomorrow with @jonerp and @josheac live. How is Enterprise AI changing the role of SIs? What should customers be asking right now? Lots to unpack.
AI Success Depends on Robust QA and DevOps Maturity
RT 77% of orgs are confident in AI outputs? while many still lack secure CI/CD, automated audit trails, and even basic defect escape metrics. Your QA and DevOps maturity will make or break AI agents in production. #AI #DevOps #QA @Star_CIO ...
AI Coach Imperfect Personalizes Daily Fitness Using Your Data
Imperfect (@imperfectco_) is an AI fitness coach that's personalized based on your health data. It adapts your training, recovery, and nutrition each day to how your body is actually responding. Congrats on the launch, and welcome back @matin (W11)! https://t.co/5N9sjEQSKG https://t.co/UL5vwXUngk
Tiny Model Beats Opus 4.6 at Spreadsheet Navigation
we trained a .35b-parameter model to navigate spreadsheets better than opus 4.6. normal corporate card company stuff.
OccupyAI Looms Soon—Invest in Your Humanity
Based on recent developments across both corporates and policymakers, I am convinced #OccupyAI is going to arrive far sooner than even I had originally expected. Start investing in your humanness now.
Robotics Data Startup Turns Video Into Training Gold
The developers behind the future of robotics industry. San Francisco is stunning today. I am here to learn more about robot data to train future robots with @keval_shah14 who built a system to prepare video data to sell to robotics...
Most Companies Boost AI; 40% Knowledge Work at Risk
92% of companies are increasing AI investment over the next 3 years. 40% of knowledge work could be disrupted in that window. What’s your AI strategy? We have a few spots left in the AI Business Lab Mastermind. https://t.co/B3ac1JrQNc
OpenAI's Partnership Hype Outpaces Financial Clarity
“they hadn’t figured out how OpenAI would pay for it” may turn out to be the epitaph for an entire era. 🪦 scoop from @anissagardizy8 @theinformation, and credit her with the great line “OpenAI has made a habit of announcing landmark...
AI Agents Threaten Relevance of Expert-Created Courses
Do AI agents make high quality online courses from proven experts more or less relevant?
Foundation Models Enable Safer, Flexible Robot Swarm Decision‑Making
Implementing foundation models as robot swarm designers and operators could unlock more flexible decision-making and enhance human-robot interactions, with careful monitoring mechanisms to ensure safety. Learn more in Science #Robotics: https://t.co/VCOZ6sAKOa https://t.co/nsKs5pkib5
Opus 4.7 Underperforms; Compute Limits Hinder Mythos Rollout
Or.... they don't have enough compute space for Mythos publicly and an awesome marketing campaign. Opus 4.7 still doesn't perform as well as 4.6 on its release. Might change with new data center expansions with Google and xAI, but as...
AI Could Secure Software by Eliminating All Vulnerabilities
Maybe powerful AI will favor the cyber defense. Vulnerabilities are finite in number, so if you can fix them all, maybe software just becomes much more secure.