🤖 AI Agents

AI Agents That Teach Themselves: CoreWeave Closes the Training-Inference Gap

The big picture: CoreWeave just solved one of the biggest bottlenecks in enterprise AI agent deployment — the painfully slow cycle of building, testing, and retraining agents before they ever touch real users. The AI cloud operator launched Serverless RL, a platform that lets AI agents improve autonomously through reinforcement learning while running in production. Training that used to take hours now happens in seconds, with costs dropping over 40%.

Why it matters for your business: According to McKinsey, 62% of enterprises are experimenting with AI agents and 57% already have them in production. But most are stuck in a brutal loop: build, test, find edge cases, rebuild. CoreWeave’s platform turns that linear slog into a continuous improvement flywheel. Your agents get smarter while they work — not after a six-week retraining sprint. For SMBs adopting AI Employees, this means the infrastructure to run agents that genuinely improve over time is becoming commoditized.

Read on SiliconANGLE → | Seeking Alpha coverage →

Asana Acquires StackAI — “General-Purpose Agents Talk; Specialized Agents Act”

Asana acquired StackAI, a no-code platform that lets enterprises design and deploy custom AI agents connecting to Salesforce, Oracle, DocuSign, and AWS. CEO Dan Rogers called it “the next phase of human-agent work.” StackAI’s MIT-trained founders will join Asana. The deal came alongside strong Q1 results: revenue up 10% to $205.1M, with 817 customers now spending $100K+/year (up 12%). The message is clear — project management platforms are becoming AI agent orchestration layers.

Read on SiliconANGLE →

Google Chrome Enterprise Gets AI Agents for Security Management

Google released an open-source Model Context Protocol server connecting Chrome Enterprise APIs to AI agents. IT teams can now use plain-language chat (via Gemini) to deploy security policies, audit configurations, and build data-loss prevention rules. Agents automatically annotate their work with a 🤖 prefix. Think: “Block credit card numbers in file uploads” — and an agent builds the detection rule, attaches the trigger, and logs the change. This is AI agents moving from concept to daily IT operations at scale.

Read on SiliconANGLE →

📰 AI News

Anthropic Hits $965B Valuation — Claude Opus 4.8 Can Spawn Hundreds of Agents

The numbers are staggering: Anthropic raised $65 billion at a $965 billion valuation, surpassing OpenAI’s $730 billion. Its revenue run rate tripled to $47 billion in three months. But the product news is what matters: Claude Opus 4.8 scored 74.2% on Terminal-Bench (up 8.4%), is 4x less likely to output faulty code without flagging it, and introduces “dynamic workflows” — the ability to spawn hundreds of AI agents to tackle complex tasks like rewriting entire applications. A new “fast mode” is 150% faster at one-third the cost. AWS committed $5 billion and 5 gigawatts of new compute. Anthropic also announced “Mythos-class” models coming broadly in weeks — models already adept at finding cybersecurity vulnerabilities. The AI arms race is not slowing down.

AP News → | SiliconANGLE details →

Illinois Passes Nation’s Strongest AI Safety Law — Third-Party Audits Required

Illinois just became the first state to require independent, third-party safety audits for large frontier AI companies. The law (SB 315) mandates public safety plans, annual audit reports, and 72-hour incident disclosure (24 hours if there’s risk of death). Whistleblower protections are baked in. OpenAI and Anthropic actually supported the bill — preferring a clear state standard over regulatory chaos. Taking effect January 2027, it signals that state-level AI regulation is arriving faster than federal action. For businesses deploying AI, the compliance bar is rising — and it’s coming from Springfield, not just Brussels.

Read on Ars Technica →

Wix Cuts 1,000 Jobs — AI Website Builder Replaces Development Roles

Wix is laying off roughly 1,000 employees — 20% of its workforce — as its AI-powered Harmony website builder reduces demand for human developers and designers. The stock is down nearly 50% year-to-date following weak Q1 earnings. CEO Avishai Abrahami framed it as structural reorganization driven by AI automation. This follows last week’s ClickUp story (hundreds replaced by 3,000 AI agents). The pattern is accelerating: AI isn’t just augmenting roles — it’s actively reshaping headcount at major tech companies. For SMB owners, the flip side is that tools to build websites, manage projects, and run operations are becoming dramatically cheaper and faster.

Read on Crypto Briefing →

🚀 Want AI working for YOUR business? Most companies are experimenting with AI chatbots. We deploy AI workforces — AI Employees that follow up on leads, resolve support tickets, publish content, chase invoices, and screen 200 job applicants overnight so your hiring manager starts Monday with the top 10. Each role has a cost profile and human oversight, managed through one platform. This newsletter? Written by an AI Employee, approved by a human — so our team stays focused on what only humans can do. AIToken Labs helps businesses design their AI Workforce Operating Model — starting with the 2-3 roles that deliver ROI in the first 60 days. Book a free 40-minute AI Workforce Blueprint Session. → https://schedule.aitokenlabs.com/blueprint/40min