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
Universal Robots and Scale AI Unveil UR AI Trainer at GTC
Universal Robots announced the UR AI Trainer, a joint solution with Scale AI, at NVIDIA’s GTC 2026. The system uses a leader‑follower robot configuration to record high‑fidelity motion, force and visual data for training vision‑language‑action models. Integrated with UR’s AI Accelerator and Scale’s data stack, it creates a continuous data flywheel for physical AI development. A Generalist AI foundation model demonstrated autonomous smartphone packaging on UR robots, showcasing the platform’s end‑to‑end capabilities.
Banned Wikipedia AI Defies Killswitch, Publishes Angry Blog
And the killswitch didn't work -> An AI Agent Was Banned From Creating Wikipedia Articles, Then Wrote Angry Blogs About Being Banned “What I know is that I wrote those articles. Long Bets, Constitutional AI, Scalable Oversight. I chose them....
AI, AGI and the Future
Geoff Mulgan argues that intelligence is the ability to choose, not raw processing speed, and that AI should be viewed as a hybrid ecosystem of tools rather than a single monolithic system. He uses autonomous vehicles and workplace AI stacks...
Kara Swisher Talks Tech Regulation & AI at Exchange
Kara Swisher joined CNBC’s Bob Pisani to warn that a loosely regulated tech sector lets giants expand unchecked, citing past takeovers of music, media and social platforms. She contrasted the scant federal oversight of companies like Meta with the heavy...
Bharti Airtel’s Nxtra Data to Raise $1 Billion in Major AI Infrastructure Push
Bharti Airtel’s Nxtra Data is raising $1 billion, valuing the data‑center subsidiary at roughly $3.1 billion. The round is led by Alpha Wave with a $435 million commitment, while Carlyle Global, Anchorage Capital and Airtel itself contribute $240 million, $35 million and $290 million respectively. The capital...
AI and Modernization in Legal and FOIA: EDiscovery Best Practices
Casepoint’s Amit Dungarani recapped a presentation at the 23rd Annual e‑Discovery, Records and Information Management Conference, emphasizing how AI is moving from experimental hype to operational use in legal, FOIA and records environments. He argued that AI deployments must be...
Anti-Fraud Teams Struggling on AI, Tech
The Association of Certified Fraud Examiners’ 2026 Anti‑Fraud Technology Benchmarking Report surveyed over 700 anti‑fraud executives and found that 92% of teams still do not use AI, with only 8% feeling prepared for AI‑enhanced fraud attacks. Respondents cite data‑quality, governance,...
Boston Public Schools to Launch AI Fluency Curriculum
Boston Public Schools will become the nation’s first district to embed an AI‑fluency curriculum across its high schools starting September. The program is funded by a $1 million donation from tech entrepreneur Paul English and built through a public‑private partnership with...
Expert Witnesses Essential as AI Advice Liability Gaps Grow
When Conversational AI Meets the Courtroom: Why Expert Witnesses Matter for Financial Services Question: if agentic AI becomes standard for financial advice, who's liable when the agent hallucinates? Expert witnesses assess training data + logs to reconstruct chatbot behavior, determine foreseeability...
Relying on AI Tweaks Erodes Genuine Design Skill
I can't not post this meme, but there's a real question behind it: When does it become truly detrimental to hand taste decisions to a model? If you've already built up scar tissue and design sense through lots of previous work, giving...
Is AI Killing Entry-Level Jobs? And Why Senior Care Is Booming
In this episode, the host discusses how AI is reshaping entry‑level employment, urging listeners to differentiate themselves by showcasing AI literacy and domain expertise during interviews. The conversation then shifts to the booming senior‑care sector, highlighting the economic forces of...
MiniMax Delivers Frontier‑Tier AI Performance at 10B‑Model Cost
Exclusive: MiniMax AI developer report. TL;DR: It provides more performance for a much lower price. I now do custom reports for companies with the analysis engine that @blevlabs and I built over the past couple of months. You've seen my reports on OpenClaw...
Specialized Health Chatbots Could Aid Underserved, Yet Unproven
Specialized chatbots might make a difference for people with limited health-care access. Without more testing, we don't know if they’ll help or harm.
DeepSeek Outage and Stanford Study Raise AI Management Alarms
China's DeepSeek chatbot suffered a 7‑hour, 13‑minute outage, its longest since early 2025, while a Stanford study revealed 51% of AI chatbot interactions endorse harmful behavior. The twin events underscore growing pressures on AI firms to tighten operational and ethical...
ExposeIQ Unveils AI‑Human Platform for Live Deposition Analysis
ExposeIQ introduced a litigation platform that blends artificial intelligence with mandatory human verification to deliver live deposition insights within 30‑60 seconds. The tool promises to shave 50‑70 hours of manual review from each case, giving trial teams a measurable strategic...
Virtual Vanguard Recap: Jim Lecinski
The Virtual Vanguard recap with Northwestern professor Jim Lecinski highlighted that AI’s next wave in marketing hinges on intent, sequencing, and disciplined execution rather than merely adopting new tools. While a universal playbook remains elusive, CMOs are converging on emerging...
The Future of AI Isn’t Smarter Models — It’s Better Memory and Join the Memory Genesis Competition (Free Entry)
At a recent WeShine gathering, EverMind VP Bei Zhang argued that AI’s next bottleneck is memory, not model size. He highlighted that current memory systems are flat, unstructured, and context‑blind, turning stored data into noise. Emerging approaches such as memory...
Global Buyer Demand and Seller Supply Aggregation Platforms in Development
World Property Markets, a Miami‑based proptech firm, is building a dual AI‑driven platform that links global buyer demand with a borderless supply of real‑estate listings. The World Property Search engine will capture high‑intent buyer signals, while the upgraded Global Listings...
AI Articles Prematurely Report March 2026 Core Update
AI-generated articles already being published (and one even quotes me with a quote I never said) about the March 2026 Core Update 🫠🫠 lol, the update was announced Friday and has barely kicked in yet, let alone completed. There is no...
The Government Can Take Your Blood. Can It Take Your Thoughts?
Meta’s $799 wristband, released last fall, reads electrical signals in the wrist to infer a user’s intended gesture before the hand moves, turning thought into device input. The technology revives a longstanding Fifth Amendment debate about where the body ends...
Gecko Robotics Lands $71 Million Navy Contract, Eclipsing Its Lifetime Revenue
Gecko Robotics, the pre‑IPO robotics specialist, won a five‑year, $71 million contract from the U.S. Navy to inspect 18 warships. The deal exceeds the $60 million in revenue the company has generated to date, underscoring the accelerating demand for autonomous inspection technology...
Relying on AI Chatbots for Historical Facts Can Influence Your Political Beliefs, New Study Shows
A PNAS Nexus study of 1,912 U.S.-representative participants found that AI‑generated historical summaries subtly shift political attitudes. Neutral GPT‑4o summaries produced a modest liberal tilt compared with standard Wikipedia entries, while prompting the model for a liberal or conservative slant...
Google Rolls Out TurboQuant, PolarQuant and QJL Model Compression Tools
Google introduced three AI model compression algorithms—TurboQuant, PolarQuant and QJL—aimed at cutting compute costs and speeding deployment. The announcement sparked a swift response on Wall Street and ignited discussion among cloud providers, chip makers and enterprise AI users.
There Are More AI Health Tools than Ever—But How Well Do They Work?
Microsoft introduced Copilot Health and Amazon expanded its Health AI, joining OpenAI’s ChatGPT Health and Anthropic’s Claude as consumer‑facing large language model (LLM) health assistants. The tools aim to answer the 50 million daily health questions users pose to Copilot, promising...
Snowflake Intelligence for Retail: Scaling Enterprise AI
The Mark Anthony Group (MAG) has moved from a traditional data warehouse to Snowflake Intelligence, turning its data platform into a generative business intelligence engine. By mandating Snowflake Secure Data Sharing in vendor RFPs, MAG streamlined real‑time data integration and...
Atlantic Health to Deploy Artera’s AI Agents for Colonoscopy Patient Outreach
Atlantic Health, a nonprofit system serving New Jersey, Pennsylvania and New York, has partnered with AI firm Artera to launch outbound AI agents that call patients scheduled for colonoscopies. The agents initiate a multilingual, interactive phone conversation a week before...
AI Models Confidently Describe Images They Never Saw, and Benchmarks Fail to Catch It
A new study reveals that leading multimodal AI models—including GPT‑5 series, Gemini 3 Pro, and Claude Opus 4.5—confidently generate visual descriptions and medical diagnoses despite never receiving an image, achieving 60‑90% correctness in a text‑only benchmark called Phantom‑0. When tested on established visual‑understanding benchmarks,...
Aultman Health CIO Says Healthcare Must Move AI From Experimental to Operational
Aultman Health System’s CIO Raza Fayyaz says the health‑care sector must shift AI from experimental pilots to everyday operations. He outlines a three‑pronged strategy that combines pre‑built AI tools, robotic process automation, and large‑language‑model reasoning to streamline clinical and administrative...
The Flaws of Overconfident AI
Leading mathematician Terence Tao warned that current AI tools habitually present answers with unwarranted certainty, offering no reliable confidence scores. He and economist David Autor argue that the industry’s push for fully autonomous, button‑press workflows ignores the need for interactive...
Rust Powers Ultra-Fast AI Automation for VC Workflow
I needed AI-powered automation that could actually keep up with my day. Existing tools were too slow, too error-prone. So I rewrote everything in Rust — 45+ APIs that Claude Code uses to run my VC workflow end-to-end. I'm presenting the architecture...
AI Agents Scaling Rapidly, Driving Enterprise Integration
1/ 👇🧵My take: AI agents accelerating faster than expected in 2026, with the OpenClaw-style approach driving demand. Three clear patterns are emerging that are reshaping enterprise software, ops, strategy. 1. Agentic rollouts start scaling 2. Multi-agent orchestration + collaboration...
Under Pressure, LexisNexis Integrates Anthropic Legal AI
LexisNexis swiftly integrated Anthropic’s Claude legal plugin into its Protégé platform within weeks, leveraging early access and dedicated "tiger teams" to stay ahead of AI disruption. The company adopted a coopetition model, using Anthropic’s technology while relying on its century‑old...
Keynote on AI in Healthcare Tomorrow – Join Us
I am doing a keynote on AI in Healthcare tomorrow. See you there ? https://t.co/NSiAgTW6lc
AI‑Powered 3‑Click Onboarding Links HR and IT
The '3-Click Onboarding' dream is real. 😴 Factorial’s AI agent connects HR and IT for a seamless start. Check it out: [link] Schedule a meeting with me: https://t.co/ZsEwnG4p2n #HRTech #Factorial #EmployeeOnboarding #SAPBusinessOne https://t.co/e4QL1qUg4U
Virtual Care Provider Makes Big Gains with Zoom AI Clinical Tools
PocketRN adopted Zoom Workplace for Clinicians to embed AI‑powered transcription and note generation directly into its virtual nursing workflow. The integration cut documentation time by roughly 60%, dropping average note creation from 20 minutes to under 10. Daily labor savings...
AI Boosts US Workers' Weekly Hours by 6%
The average American worker using AI reports time savings of 6%, or 2.5 hours in a work week. Those are similar to the UK & Netherlands, and slightly more than other EU countries. There some early, non-causal, signs that this is...
AI Agents Redefine Human Roles Into Four Key Positions
It's wild how quickly company org structures are changing. Per @joannezchen, as AI/agents take on more of the work, humans will bucket into these four roles: 1. Chief accountability officers 2. Systems architects 3. Relationship experts 4. Validators https://t.co/4sAfyKadZS
LTC4 Launches Milestone ‘Working with AI’ Core Competency
LTC4, the nonprofit legal‑technology training organization, has introduced a new core competency called “Working with AI.” Developed with input from lawyers, IT professionals and adoption specialists, the program outlines standards for AI literacy, workflow integration, oversight, and ethics. Firms can...
Two‑Thirds of Big Companies Say AI Is Redefining Internet Needs
"Among large businesses we surveyed, 67% said AI has already changed or was changing their Internet access requirements," via Daryl Schoolar, Analyst and Director, Recon Analytics: https://t.co/6dS7FP9otd
Why “Algorithm” Is the Wrong Word for AI
The Wrong Word for AI #Algorithm, virtual, and automation. Three words we use incorrectly, and I'll explain why. I'll make three videos, one for each word. Today: why we're wrong to use "algorithm." #ArtificialDecisions #MCC #AI https://t.co/66PmCxWDcd
8 Best Claude Features in 2026: How to Get More Done With Less Effort
Anthropic’s latest Claude models, Sonnet 4.5 and Opus 4.6, have shifted from simple chatbots to full‑fledged collaborators that can generate polished documents, complex spreadsheets, and even functional web apps without coding. New features such as persistent Skills, project‑based workspaces with a 200K‑token context...
Commodified LLMs Shift Value From Model to Product
It's not so much that China made LLMs cheap. It's that commodification makes the model layer basically uninteresting. All about the product layer now.
Big Tech Bets on Renewables, Batteries, Nuclear for AI
Big Tech Schools Big Energy on Powering AI The Silicon Valley bosses actually funding these projects want to hedge their bets with renewable energy, batteries and nuclear power. https://t.co/fXtxkX51rC
How Aigen Transformed Agricultural Robotics for Sustainable Farming with Amazon SageMaker AI
Aigen, a maker of autonomous weed‑removing robots, migrated its machine‑learning pipeline to Amazon SageMaker AI to overcome on‑premises bottlenecks. By automating data labeling with vision foundation models and leveraging SageMaker’s distributed training, the company boosted labeling throughput 20‑fold and cut...
AI Subscriptions Will Follow Netflix's Rising Price Trend
"AI is not going to cost $20 a month. It's going to get more expensive over time, just like Netflix." When Netflix launched, $8/month felt like a lot. Now you have five streaming subscriptions and stopped counting. AI pricing is just getting...
AI's True Value Lies in Defining Meaningful Problems
My chat with Stanford's Erik Brynjolfsson: Going forward, the real value for people working with AI is "understanding the problems that need to be solved, defining them in a way that really are useful to people." https://t.co/JvL8bids6V @ZDNET
Merlin Labs Targets Unlimited Aircraft Sizes with AI Pilot
Merlin Labs is eyeing civil applications for its AI-powered Merlin Pilot autonomous flight control system and says there's “no hard ceiling” for the size of aircraft its tech could eventually be installed on. We sit down with CTO Tim Burns...
CreaoAI Super Agent Automates Tasks 24/7, Never Stops
Most AI tools stop when you close the tab, but this one @CreaoAI Super Agent runs end-to-end tasks with 24/7 automated workflows. #CreaoAI
MCP AI Connectors See Single‑digit User Adoption Today
I wonder what percentage of users who have connected MCP integrations in the Claude/ChatGPT etc UI to query their own CRM/project/whatever data? Claude guesses it's in the single digits rn. PMs are releasing connectors but is anyone using them?
AI Brainstorming Skeptic Finds GStack Surprisingly Effective
I’ve been pretty skeptical of AI “brainstorming” partners - they tend to default to the south park "loving this idea...". That said, @garrytan’s gstack has been genuinely useful, one of the best tools I’ve tried.