💼 Salesforce Cuts 4,000 Jobs Due to AI Agents
Salesforce CEO Marc Benioff says AI agents replaced nearly half of the company’s support staff, cutting 4,000 roles. Now, bots handle 50% of customer interactions and follow up on every lead. Despite the cuts, customer satisfaction stayed level. Benioff called the AI transition a “rebalance” and a “stunning” success.

🦾 Runway Eyes Robotics for AI Simulation Expansion
Runway, known for generative video tools, is pivoting toward robotics as companies tap its world models for training simulations. Its realistic AI-generated environments help robotics and self-driving firms test actions at scale—without the cost of real-world trials. Runway won’t launch new models, but will fine-tune existing ones for this sector.

🧠 Hermes 4 Outsmarts ChatGPT—with No Guardrails
Nous Research’s Hermes 4 models rival top proprietary AIs in math and reasoning, all while minimizing refusals and safety filters. The open-source release uses novel training systems like DataForge and Atropos to create dynamic reasoning models with transparent thought processes. It’s a bold challenge to Big Tech’s control—and its content restrictions.

🦿 NVIDIA Pushes Robotics Forward with Blackwell GPUs
NVIDIA’s Jetson AGX Thor platform brings Blackwell GPU power to robotics, enabling real-time decisions and major performance boosts. It’s a leap toward “physical AI,” where robots mimic human actions. Industrial adoption looks imminent, but high costs may slow humanoid robots’ entry into everyday homes.

Enjoying our newsletter? Forward it to a colleague—
it’s one of the best ways to support us.

Zapier’s pricing is built to grow with you—no hidden costs, no “success tax.” Smarter, simpler workflows mean you save more as you scale.

Check out the Zapier vs n8n comparison today.

The tiny Caribbean island of Anguilla is cashing in on its country code “.ai” as artificial intelligence fever drives demand for domain names. In 2024 alone, Anguilla earned nearly $39 million from domain registrations—almost a quarter of its total government revenue—up from under 50,000 .ai domains in 2020 to more than 850,000 today.

The windfall is helping diversify an economy long reliant on tourism and vulnerable to hurricanes. By partnering with Identity Digital and avoiding a one-time licensing deal, Anguilla is playing a smarter long game than predecessors like Tuvalu—banking on .ai staying hot, and hoping not to get burned.→ Read the full article here.

AWS sits at the heart of the generative AI boom, powering everything from LLM training runs to global-scale inference. In this sweeping conversation, AWS CEO Matt Garman discusses the future of work, engineering, open vs. closed models, and why agentic workflows—not just raw models—will be where the next wave of value is created.

Garman is bullish on the economic upside of AI, skeptical of doomer narratives, and refreshingly candid about infrastructure bottlenecks, engineering culture, and how Amazon uses its own silicon to support customer choice. → Read the full article here.

Jim O’Leary, North America CEO at Weber Shandwick, has turned his office into a functioning AI lab—equipping himself with tools that summarize meetings, surface critical intel, and draft communications in his voice. The goal: scale himself and set an example for clients navigating AI adoption.

After six months of setup and customization, O’Leary estimates he's saving one to two hours a day—time now spent on strategy, leadership, and even his kids. It’s not about replacing the CEO, he says, but making his knowledge and decision-making more accessible across the company. → Read the full article here.

Tencent AI Lab and Washington University researchers have unveiled R-Zero, a training framework that allows large language models to evolve without any human-labeled data. By pitting two models—Challenger and Solver—against each other in a self-reinforcing loop, R-Zero builds its own curriculum of reasoning challenges, driving measurable performance gains across tasks like math and general-domain reasoning.

This could upend the economics of AI development, particularly in domains where labeled data is scarce or costly. But the system’s long-term stability is still in question, with self-generated data quality declining over time—a critical challenge for the future of self-improving AI. → Read the full paper here.

🩺 AI Stethoscope Diagnoses in Seconds: An AI-powered stethoscope from Imperial College can spot heart failure, valve disease, and arrhythmias instantly, making early detection 2–3.5x more likely in UK trials.

⚡ AI Demand Sparks Nuclear Revival: NextEra plans to restart the Duane Arnold reactor in Iowa by 2028, tapping rising AI data center demand and bringing 600MW of clean power back to the U.S. grid amid shifting energy economics.

🛠️ Netstock Eases AI Into Warehouses: A 16-year-old inventory software firm is helping small businesses adopt AI gently, serving 1M+ tailored insights without replacing human judgment or overwhelming users.

🧹 Humans Clean Up AI’s Mess: Freelancers are cashing in on fixing sloppy AI work—rewriting dull articles, redrawing wonky logos, and debugging botched code as companies realize automation still needs a human touch.

📢 Want to advertise with Forward Future? Reach 550K+ AI enthusiasts, tech leaders, and decision-makers. Let’s talk—just reply to this email.

Thanks for reading today’s newsletter—see you next time!

Matthew Berman & The Forward Future Team

🧑‍🚀 🧑‍🚀 🧑‍🚀 🧑‍🚀