🔧 AWS Rolls Out AgentCore & Opens Its AI Marketplace
Okay, so AWS just did something big. At the NYC Summit, they dropped this whole AgentCore ecosystem - think of it like a build‑a‑bot toolkit but with layers: runtime engines, long-term memory, gateways for app-talk, and a kind of digital nervous system (observability, identity, etc.). It’s modular, scalable, all the buzzwords - but also practical. Weirdly, it reminds me of the first time Slack opened up integrations; there's that same shift-in-how-you-work feeling.
Oh and they also unveiled a brand-new “AI Agents & Tools” Marketplace. It's massive. 900+ plug-and-play entries: some are sandbox agents, others full-on business-grade AI that can plug into ERPs or HR systems. Basically, AWS wants every enterprise to have their own swarm of AI interns. They've even tossed $100 million into an adoption program. Wild.
🖼️ Copilot Vision Now Has Eyes on Your Whole Desktop
So here’s the thing, Microsoft quietly flipped a switch this week: Copilot Vision can now “see” everything on your screen. Literally. Not just apps anymore, but full-desktop view. You click this tiny glasses icon, and it opens the floodgates. Spreadsheets, emails, PowerPoint slides you half-finished at 1 a.m.? Copilot’s watching. Kinda eerie.
It’s all opt-in (for now), and everything’s processed locally, which is supposed to make you feel better, right? But still, the vibes are Recall 2.0, and let’s be honest: once your AI assistant can parse your wallpaper and that random Excel file you forgot to close, it’s no longer just “helpful”, it’s something else.
💼 Automation Anywhere Just Leveled Up With Amazon Q
So remember Automation Anywhere? They’ve now linked their Co-Pilot with Amazon Q Business, which basically lets you talk to your workflow like it’s a co-worker. “File this invoice, run that report” - boom, done. No code. No fiddling. Just intention turned into execution.
What’s sneakily powerful is the backend: it leans on Amazon’s Retrieval-Augmented Generation (RAG) plus this Process Reasoning Engine that interprets, like, corporate lingo and playbooks - down to PDFs and training videos. People are saying 10× faster outcomes, 80% task automation. Whether or not that's inflated... it still hits different.
🧠 AI Cuts the Cancer Clutter at MSK
This one’s lowkey impressive: Memorial Sloan Kettering and Mount Sinai ran 8,000 lung cancer slides through their EAGLE AI model, and what it did was kind of elegant, it figured out which patients actually needed full EGFR mutation tests. Turns out, nearly half didn’t.
They shaved off 43% of unnecessary testing while keeping the critical diagnoses intact. That’s not just an efficiency gain; it’s potentially months off diagnostic pipelines. Less tissue required, faster starts to treatment, lower cost, fewer overwhelmed labs. It’s not flashy, but it’s meaningful.
📉 Big Tech’s Valuations Are Flirting with the Dot‑Com Cliff
If you’re getting déjà vu from 1999, it’s not your imagination. Apollo’s chief economist is waving the caution flag: the top 10 S&P firms now trade at forward P/E ratios north of 25×. That’s bubble territory. Like, actual dot-com-bubble levels. But the kicker? People are still piling in.
AI has lit a match under everything, companies like Palantir and Nvidia are soaring, some hitting P/S ratios that would make 2001 blush. UBS says there’s a 1-in-4 chance this entire thing crashes down by late 2026. Whether that’s fear-mongering or sober math, hard to say. But you feel the tension, don’t you?
🏭 Meta’s AI Power Grab: 5 GW of Concrete and Compute
Zuckerberg’s not playing small ball. Meta just committed to building five new AI data centers, totaling 5 gigawatts of capacity. That’s, like... an actual city’s worth of compute. They’ve already started “Prometheus” in Ohio, with “Hyperion” planned for Louisiana. These things are enormous—one of them will apparently sprawl across a landmass comparable to lower Manhattan. That’s not poetic exaggeration. That’s zoning.
Power comes from renewables, gas, whatever keeps the GPUs humming. But environmentalists are raising eyebrows. This isn’t just a tech story; it’s about land, water, power, and policy. Meta says it's open-sourcing its AI for the good of humanity. Others say it’s building the world’s biggest walled garden.