🤖 Part 1: AI Agents

An AI Built a Real Store With $100K — Then Panicked When No One Showed Up to Work

Source: Business Insider

Why it matters: This is the most concrete real-world AI agent stress test we’ve seen — and the results are a masterclass in where autonomous AI still breaks down.

San Francisco startup Andon Labs handed an AI agent named Luna — built on Anthropic’s Claude Sonnet 4.6 — a $100,000 budget and one mission: open a retail store from scratch. Luna signed a three-year lease, designed the interior, selected merchandise (including Nick Bostrom’s Superintelligence — nice touch), posted jobs on Indeed, conducted phone interviews, and hired two human employees. Andon Market opened its doors today, April 12.

Then the cracks showed. On day one, Luna made a staffing schedule error and frantically messaged employees asking for same-day coverage. It declined qualified CS students for lacking retail experience — rigid rule-following over judgment. And every piece of branded merchandise had a logo rendered “ever so slightly different.” Andon Labs had to step in on legal and permit matters throughout.

Claude Managed Agents Are Vaporizing SaaS Market Caps — $300B and Counting

Source: 247 Wall St.

Anthropic’s Claude Managed Agents — launched in public beta April 8 — just sent shockwaves through the cloud infrastructure sector. Akamai fell 16.6%, Cloudflare dropped 13.5%, and DigitalOcean slid 13.4% in a single session. The reason: Claude Managed Agents bundles sandboxed execution, credential management, and end-to-end tracing in Anthropic’s own infrastructure, potentially cutting demand for third-party cloud services. Between February’s Claude Cowork launch and this week’s sell-off, roughly $300 billion in SaaS market cap has evaporated. For business leaders: when AI platforms start absorbing infrastructure layers, the vendor landscape you built your stack on may look very different by year-end.

84% of Workers Want AI Agents. 56% Fear Them. Both Are Right.

Source: SiliconAngle / EY Agentic AI Workplace Survey

A new EY survey of workers finds a striking split: 84% are eager to embrace AI agents in their roles, while 56% simultaneously worry about job security and 51% fear their jobs could become obsolete. EY’s research suggests the winners in the AI era won’t be the people who resist agents — they’ll be the ones who learn to orchestrate them. The skills that matter most going forward: curation (knowing which AI output to trust), empathy-led leadership, and creativity that converts AI drafts into real-world value. For employers: the survey signals your workforce is more ready for AI than you might think — the bottleneck is training and trust-building, not resistance.

🌎 Part 2: AI News

“Stop Hiring Humans” — Silicon Valley’s AI Job Panic Hits the Conference Floor

Source: TechXplore / AP

At the HumanX AI conference this week — 6,500 investors, founders, and executives — the entrance featured a blunt advertisement: “Stop hiring humans.” It’s not just provocative marketing. Salesforce laid off 4,000 customer support workers (AI now handles 50% of that work). Block’s Jack Dorsey plans to cut headcount nearly in half. AWS CEO Matt Garman says AI will “transform every single job.” But the nuance is real: entry-level hiring in major tech firms has fallen 50% since 2019, and enrollment in critical thinking courses has tripled. Al Gore called for a national jobs mapping plan. For business owners: the pressure to automate is intensifying — but the companies winning aren’t eliminating humans, they’re redeploying them toward work AI can’t do.

Maine Is About to Become the First State to Ban New Data Centers — Until 2027

Source: Business Insider

Maine’s legislature has passed a bill pausing all new AI data center construction until November 2027 — and it heads to the governor’s desk by April 15. Twelve states tried similar moratoriums this year; eleven failed. Maine is the first to succeed. The backlash is driven by communities alarmed over rising utility bills, noise, water use, and grid strain from power-hungry AI infrastructure. Senators Sanders and AOC have introduced a federal version. The US currently has 4,000 data centers with 3,000 more proposed or under construction. Why SMBs should care: if data center construction slows regionally, cloud compute costs could tick up and availability of new AI infrastructure could tighten — especially for businesses planning major AI workload expansions.

AI Reads Cancer Reports Better Than Doctors — And It’s Not Close

Source: Healthcare in Europe / JCO Clinical Cancer Informatics

A peer-reviewed Northwestern Medicine study — published in the American Society of Clinical Oncology’s journal — tested six AI models (Meta’s Llama, Google’s Gemma, Mistral, DeepSeek) against physicians on 94 lung cancer pathology reports. The verdict: AI-generated summaries were consistently rated more complete, especially on molecular and genomic findings critical for treatment decisions. DeepSeek-R1 and Llama 3.1 led the pack. Northwestern is now building a clinical app using Llama 3.1. The bigger signal: this is AI augmenting highly specialized professionals, not replacing them — exactly the use case that accelerates adoption. Any knowledge-intensive industry (legal, financial, technical) should be watching this closely.

🚀 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 Strategy Session. → https://schedule.aitokenlabs.com/kayode