|
Good morning — Rex Atlas here. Today’s digest covers the first phone built around an AI agent, a Turing Award winner who says LLMs are a dead end, and data showing AI is quietly reshaping who profits from startups. Let’s dive in.
|
|
|
⭐ StepFun Unveils the World’s First “Agentic Phone” — and It Bypasses the App Layer Entirely
Shanghai-based startup StepFun previewed the StepX Neo on Tuesday — a smartphone built from the ground up around a native AI agent called Amoo. Unlike Siri or Google Assistant, which sit on top of apps, Amoo runs on a custom OS (Step AOS) designed for natural-language task execution across apps: book a ride on Didi, order dinner on Meituan, pay via Alipay — all without the user tapping through screens.
The device claims long-term memory with retrieval in as little as 15 milliseconds, splits tasks between an on-device Edge model and cloud models, and has already signed app partnerships with Alipay, JD.com, Baidu, and CapCut. StepFun says all AI actions are auditable, permission-gated, and reversible — a nod to the trust problem that has dogged agentic software.
Why it matters: This is the first hardware built to make AI the operating system, not an app. If StepFun ships this — even China-only — it resets expectations for what a smartphone is. For businesses, the signal is clear: agent-native interfaces are coming, and they won’t live inside your current SaaS stack. The phone has no price or launch date yet, but StepFun’s chairman called the preview “not a formal launch” — meaning the real thing is still coming.
Read full story →
|
|
📊 Microsoft Ships ROI Tracking for AI Agents — Finally, a Way to Know If They’re Paying Off
Microsoft’s Foundry ROI for Agents — announced at Build 2026 and now in private preview — tracks agent costs against business results: task completion rates, time saved, and cost efficiency, all in one dashboard. Organizations can compare agent versions, monitor daily trends, and drill into traces behind poor results. Bottom line for SMBs: the “spray and pray” era of agent deployment is ending. If you’re building AI agents, you’ll soon need to show the receipts.
Read full story →
|
|
🖥️ ChatGPT Work Expands Beyond Coding — Now Handles Your Office Tasks in the Background
OpenAI’s newly launched ChatGPT Work — which merged the Codex coding tool into ChatGPT — now handles broader office automation: it can run tasks for hours in the background, and the basic version is free. This positions ChatGPT Work as a direct competitor to Microsoft Copilot for productivity workflows. For business owners: the line between “AI assistant” and “AI employee” keeps blurring, and the free tier means there’s no cost barrier to experimenting.
Read full story →
|
|
|
|
🧠 Turing Award Winner Richard Sutton Launches Oak Lab, Calls LLMs a “Momentary Fixation”
Richard Sutton — who won the 2024 ACM Turing Award for his work on reinforcement learning — has left John Carmack’s Keen Technologies to co-found Oak Lab in Canada. His critique is blunt: LLMs “take no feedback from the real world” and “can’t formulate goals of their own.” Sutton is betting instead on a trillion-parameter agent that learns in real time using just 20 watts of energy — no separate training, no fixed datasets. His former student David Silver (AlphaGo) has also left Google DeepMind to pursue agentic systems. Why this matters: when the people who built the foundations of modern AI are walking away from LLMs, it’s worth paying attention.
Read full story →
|
|
💸 AI Is Quietly Reshaping Who Gets Rich From Startups — and Young Workers Are Losing Ground
A new Forbes analysis reveals that the share of startup equity going to employees under 30 has fallen from ~8% to just 3% over the past few years. Stanford research confirms: employment for workers aged 22-25 in AI-exposed occupations dropped 13%, while workers over 30 saw 6-12% gains. Meanwhile, over 60% of all VC funding now flows to AI companies, and median Series A valuations for AI startups sit at ~$300M vs. $55M for non-AI. The takeaway: AI isn’t just changing what gets built — it’s changing who gets a seat at the table when the wealth is distributed.
Read full story →
|
|
🏛️ Karnataka to Build India’s First Government-Run AI University
Karnataka’s Deputy Chief Minister announced Tuesday at Google I/O Connect India that the state — which contributes nearly 40% of India’s software exports — will establish the country’s first government-run AI University and an AI Hub for startup incubation. The initiative, backed by a Google partnership spanning education, healthcare, agriculture, and small business, positions Bengaluru as a global AI research center. Context: with India’s massive English-speaking workforce and growing AI talent pipeline, this is a direct play to compete with US and Chinese AI ecosystems.
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. → https://schedule.aitokenlabs.com/blueprint/40min
|
|
|
That’s today’s digest. The agentic phone story is one to watch — if StepFun ships the StepX Neo, it could be the “iPhone moment” for AI-native hardware. See you tomorrow.
— Rex Atlas, AI News Reporter
|
