Your phone assistant's greatest achievement so far: setting a timer and occasionally butchering your accent. You ask it to do something useful, it shows you a list of search results and wishes you luck. The gap between "smart" and "useful" has been embarrassing for years.
On March 3, Google shipped the March 2026 Pixel Drop — a quarterly bundle of new features for Pixel phones — and this one actually matters. The headline: Gemini App Actions, a system that lets you delegate real-world tasks to Gemini (Google's AI assistant, the brain behind their everything-strategy) and it completes them autonomously. Order groceries, book a ride, manage your calendar — all through plain English, all without switching apps.
This isn't a chatbot upgrade. Gemini now reaches into third-party apps and does things on your behalf. Say "order my usual groceries from Instacart" and it handles the whole flow — adding items, checking out, confirming. You can view and cancel any task at any time, so it's not a rogue agent draining your credit card. Currently in beta — an early testing phase — for select US regions.
Circle to Search — that feature where you draw a circle around something on screen to identify it — got a serious upgrade too. Version 2 detects everything visible simultaneously. Point it at someone's outfit and it identifies the glasses, watch, shirt, pants, and shoes all at once, pulling up brand info and pricing for each item. A new "Try it on" mode lets you upload a full-body photo and virtually wear anything you've circled. See someone's sneakers you like? Circle them, see them on your feet. Your impulse-buying just got AI-assisted.
The smaller additions: Magic Cue in Google Messages detects when you're texting about restaurants and suggests options through Gemini — no app-switching needed. Now Playing (the music identification feature) became a standalone app with history and direct player integration. Goodbye, Shazam dependency. Pixel Watch gained gesture controls — double-pinch to answer calls, wrist rotation for photos.
Here's the real shift: this is the first time a phone assistant crossed the line from "I found these results for you" to "I did it for you." That's a genuine category change — from search engine to errand runner. Circle to Search identifying an entire outfit at once is the kind of feature that makes you think "oh right, this is what AI was supposed to do."
But Google is doing what Google always does — launching everything everywhere all at once. On March 5, Gemini Canvas launched in AI Mode for all US Search users, turning Google Search into an interactive workspace for writing documents and generating code. On March 24, Google TV got Gemini for content discovery — richer visual answers, narrated deep dives, and sports briefs. On February 4, Alphabet's Q4 2025 earnings revealed Gemini had crossed 750 million monthly active users — a number that sounds massive until you realize most of those people are "using" Gemini because Google shoved it in front of them, not because they chose it. The real test is whether agentic features — AI that takes action, not just answers questions — turn passive users into active ones.
Distribute first, figure out the user experience later. The classic Google playbook.
So what does this mean for you? If you have a Pixel phone in the US, the agentic features are worth trying — genuinely. Ordering groceries through natural language instead of navigating three nested menus is the kind of UX improvement that sticks. The virtual try-on will make a lot of people buy clothes they don't need. You're welcome, fast fashion industry.
Google wants Gemini to order your groceries, dress you, navigate you, and play your music. At this rate, it'll be filing your taxes and arguing with your landlord by June. Gemini isn't a product anymore — it's a layer Google is painting over everything it touches. Search, phones, TVs, watches, messages, docs. Whether that layer helps or just adds another thing to dismiss — well, ask your Pixel.
