Gemini goes agentic

Plus: Anthropic's CEO won't budge

February 27, 2026

Welcome back! You can now tell Alexa to chill out, literally. Amazon just dropped three new personality styles for Alexa+: Brief (straight to the point), Chill (your laid-back buddy), and Sweet (warm, encouraging, basically your AI hype person). Each one is tuned across five dimensions: expressiveness, emotional openness, formality, directness, and humor. 

It's a tiny feature with big implications. We're entering an era where AI doesn't just answer your questions but mirrors the energy you want back. Which Alexa+ mode would you choose as your default? Hit reply and let me know!

Google Gemini Can Now Order Dinner For You

Via Google Blog

This week, Google crossed a new frontier from chatbots into agents. Gemini is now rolling out task automation on Pixel 10 and Samsung Galaxy S26 devices, letting the AI independently navigate apps like Uber and DoorDash to complete tasks on your behalf.

How it works: Gemini uses computer vision to "see" app interfaces the way humans do, recognizing buttons, form fields, and navigation patterns without needing custom API integrations. When it finishes, you get a notification.

Why Uber and DoorDash first: Hailing a ride or ordering dinner involves repetitive steps that rarely require creative judgment. Plus, they’re high frequency and relatively low stakes, making automation valuable. If Google proves the tech works reliably here, expansion to travel booking, reservations, and shopping is the logical next move.

The bigger picture: This is the clearest signal yet that we're moving from AI you talk to into AI that handles your digital busywork. OpenAI has talked about agents, Microsoft is building them into Copilot, Anthropic's Claude can navigate computers directly. But Google is first to put agentic AI into the hands of mainstream consumers through devices they already own.

Anthropic Draws a Big Red Line With the Pentagon

Anthropic CEO Dario Amodei is putting his foot down. Yesterday, Amodei said his company “cannot in good conscience accede” to the Defense Department’s demand for unrestricted access to its AI within classified military networks.

The ultimatum:

  • Defense Secretary Pete Hegseth gave Anthropic until 5pm today to comply or lose out on a $200 million contract (and potentially be labeled a “supply chain risk, typically reserved for foreign adversaries).

  • We’ll see what happens in just a few hours—whether the contract is scrapped or the two head back to the negotiating table.

Silicon Valley is picking sides: A growing number of tech workers from OpenAI and Google signed an open letter supporting Amodei’s stand. Even retired Air Force Gen. Jack Shanahan (who led Project Maven) wrote that he’s “sympathetic to Anthropic position” and that the company’s red lines are “reasonable.”

The timing is messy: Earlier this week, Anthropic announced it’s replacing its two-year-old Responsible Scaling Policy with a more flexible "Frontier Safety Roadmap," sparking concern the company was softening its safety stance. Anthropic says this change is unrelated to the Pentagon dispute—but both developments paint a picture of a company feeling pressure from multiple directions.

Why it matters: Anthropic has long positioned itself as the AI company with a "soul"—the responsible adult in a room full of move-fast-and-break-things competitors. The eventual result of its high-stakes battle with the Pentagon could set the precedent for how other tech companies do—or don’t—bend safety rules related to their models.

Wearable Makers Go All-In on AI Health Coaches

The next wave of wearables isn't just tracking your health. It's telling you what to do about it.

This week, CUDIS launched its newest smart ring with an AI "agent coach" that generates everything from personalized exercise programs to referrals to licensed medical professionals.

  • Beyond standard metrics like sleep quality, stress, and heart rate variability, CUDIS calculates your "Pace of Aging"—whether your body is aging faster or slower than your chronological age. The AI spots negative trends (like chronic poor sleep or elevated resting heart rate) and either suggests lifestyle changes or escalates to a healthcare professional.

  • The company has sold 30,000 units across two models and grown its app to 250,000 users in 103 countries.

Meanwhile, Oura just unveiled a women's health AI model trained on data from over 600,000 users. The model using menstrual cycle data to provide context-aware guidance. Instead of generic sleep or activity advice, the model tailors recommendations based on where a user is in their cycle.

The bigger picture: People are increasingly turning to AI for health insights they'd traditionally get from doctors, trainers, or nutritionists. Wearables are betting that if they already have your biometric data, they should be the ones interpreting it for you. Whether consumers—and doctors—agree with that bet is another question.

Find the clip and cut the chaos

Via Wideframe

Wideframe is an AI video workflow tool that indexes massive footage libraries using transcripts and semantic search. 

How you can use it:

  • Search footage by dialogue or intent

  • Auto-generate rough cuts from large media libraries

  • Export Premiere-ready sequences

  • Speed up pre-edit and first-pass assembly

Pricing: Paid with free trial

Turn a prompt into apps in minutes

Via Fabricate

Fabricate is an AI-powered no-code platform that transforms a written prompt or template into a ready-to-deploy web app. It scaffolds the frontend, backend, and database automatically.

How you can use it:

  • Spin up an MVP for a SaaS product without hiring engineers

  • Prototype internal dashboards or booking systems for quick validation

  • Create marketplace concepts with pre-populated demo data

  • Generate privacy-safe test databases for product experiments

Pricing: Free and paid plans available

Jobs, announcements, and big ideas

  • Google unveiled Nano Banana 2, a next-generation image generator offering pro-level features with significantly faster generation speeds (more details in my video👇).

  • OpenAI has raised $110 billion in investments from Amazon, Nvidia, and Softbank.

  • Apple released Xcode 26.3, introducing autonomous coding agents that can analyze projects and make edits directly within development environments.

  • Burger King launched “Patty AI,” a headset assistant designed to monitor and improve staff politeness in real time.

  • QuiverAI opened the public beta for Arrow 1.0, describing it as the first AI model purpose-built for generating scalable SVG graphics.

  • President Trump said major tech companies should supply their own power for new AI data centers.

  • Google updated Circle to Search to allow users to explore multiple objects within a single image at once.

Nano Banana 2 just dropped again. Here’s what changed.

That’s a wrap! See you next week for more.

—Matt (FutureTools.io)