🤖 AI Agents

Automation Anywhere Unveils EnterpriseClaw — AI Agents That Run Where the Work Actually Happens

Why it matters: Most enterprise AI agents live in the cloud, but 83% of the world’s business data sits on-premises. Automation Anywhere’s new EnterpriseClaw platform — built with Cisco, NVIDIA, Okta, and OpenAI — lets AI agents operate across cloud, desktop, on-premises servers, and secured legacy systems from a single orchestration layer. This is the “last mile” problem solved: agents can now reach into SAP, mainframes, and air-gapped systems without routing sensitive data through external APIs.

The partnership stack is formidable: Cisco provides AI Defense for agent security, NVIDIA contributes OpenShell runtime and Nemotron models for on-prem inference, Okta handles cross-agent identity and authentication, and OpenAI supplies frontier models like GPT-5.5. Automation Anywhere’s own Process Reasoning Engine and Contextual Intelligence Graph tie it together.

The takeaway for your business: If you’ve been waiting for AI agents that can actually touch your ERP, CRM, or legacy systems without a six-figure integration project, EnterpriseClaw signals that plug-and-play enterprise agents are arriving. Preview now, GA later in 2026. Start mapping which internal systems would deliver the highest ROI from agent automation.

Read full story →

⚡ Visual State Cards Double Small AI Model Success Rates on Real Desktop Tasks

Researchers from Shanghai Jiao Tong University and Xiaohongshu published a framework called MMSkills that bundles screenshots (“visual state cards”) alongside written agent instructions. The result? Small models like Qwen3-VL-8B more than doubled their success rates on real desktop tasks — from 10.78% to 25.40% on OSWorld benchmarks. Even frontier models like Gemini 3.1 Pro improved from 44% to 50%. The framework is open-source on GitHub. Why this matters: you don’t need a massive, expensive model to get reliable desktop automation — you need better context. This is a practical, cost-lowering breakthrough for SMBs.

Read full story →

⚡ Devenex Launches First “Execution Control Plane” for AI Agents — Because Governance Is Now a Precondition, Not an Afterthought

Launched at Google Cloud Next, Devenex is the first governance infrastructure purpose-built to control AI agent execution. Deloitte’s 2026 survey found 80% of leaders now cite security and compliance as the top barrier to AI agent adoption (up from 68% in 2025). Devenex ensures every agent action is policy-evaluated, identity-bound, explicitly authorized, and recorded as audit-grade evidence before execution. With the EU AI Act fully applicable August 2, 2026, this category is about to get very crowded — and very necessary.

Read full story →

📰 AI News

🔍 Google I/O 2026: Gemini 3.5 Flash, a 24/7 Personal AI Agent, and the Biggest Search Overhaul in Years

Google came out swinging at I/O 2026. The headlines: Gemini 3.5 Flash is now the default model — 4x faster than competitors and outperforming GPT-5.5 and Claude Opus 4.7 on multi-step tool use and financial analysis. Gemini Spark is a new personal AI agent that runs on dedicated VMs in the background, handling tasks autonomously (beta next week for Ultra subscribers). Antigravity 2.0 brings multi-agent orchestration and 12x faster coding to Google’s agentic IDE. And Google Search is getting a dynamic expandable search box that accepts text, images, files, and video — merging AI Overviews and AI Mode into one unified experience. The Gemini app has surpassed 900 million users.

So what: Google is betting that AI agents — not chatbots — are the future of search and productivity. The new $100/month AI Ultra tier signals that premium AI experiences are the new normal. If your business relies on search traffic, the unified AI search experience means your content strategy needs to account for how Gemini surfaces and synthesizes information.

Full Google I/O roundup → | Antigravity 2.0 details →

🧠 Andrej Karpathy, OpenAI Co-Founder and “Vibe Coding” Inventor, Joins Anthropic

In a move that shook the AI world Tuesday, Andrej Karpathy announced he’s joining Anthropic’s pretraining team. The OpenAI founding member, former Tesla AI director, and creator of Eureka Labs will lead a team focused on using Claude to accelerate pretraining research — essentially teaching AI to help build better AI. His March “autoresearch” experiment, where an AI agent ran 700 unsupervised experiments and found 20 optimizations that cut training time by 11%, previews the kind of work he’ll be doing. Karpathy’s move is the latest sign that Anthropic — reportedly in talks at a ~$1 trillion valuation — is consolidating top talent in the race toward AGI.

Read full story →

💾 Dell COO: Token Consumption Up 320x — Agentic AI Is Breaking Cloud Economics

In a candid Forbes interview, Dell COO Jeff Clarke revealed that token consumption for AI reasoning has surged 320x as agentic systems proliferate — even as per-token costs fell 80%. His warning: cloud-only AI strategies will become “economically indefensible and thermodynamically impossible to sustain at scale.” Dell’s answer is a “deskside to data center” continuum running agents locally where data lives. The company posted $9 billion in AI server revenue in Q4 (up 342% YoY) and projects ~$50 billion for FY2027, with a $43 billion AI server backlog. Clarke’s core message: “Move AI to the data, not the data to AI.”

For your business: If you’re running AI agents that chain multiple reasoning steps, your cloud inference bill may already be compounding silently. The architectural decision — cloud vs. local vs. hybrid — needs to happen before the finance team sees the numbers, not after.

Read full story →

🚀 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 →