google's android show i/o 2026 landed today. one of the announcements that didn't get the front-page treatment but should have, from a distribution standpoint: create my widget. users type "show me wind speed and rain" and gemini intelligence generates a working, resizable home-screen widget — pulling from gmail, calendar, the web, whatever it needs — without your app.
read that again. without your app.
for the last five years, "we have a great widget" has been a real indie acquisition channel. the workflow was simple: user wants a widget for X, googles "best X widget android," lands on play store, installs your app, configures the widget. you get the install, the analytics event, the eventual subscription if your app is good. the widget was the wedge.
create my widget removes that funnel. on pixel and galaxy this summer, on every gemini intelligence-enabled device by year-end. it's not killing widget devs outright — your widget can be richer, more interactive, more deeply integrated than what gemini cooks up from a one-line prompt. but the lazy 80% case where someone installs a $0.99 widget app just to surface their step count? gone.
why it matters for indies#
if your app's growth model has a widget as the hook, the math is changing. specifically:
- utility widgets are commoditized. anything gemini can synthesize from gmail/calendar/web/sensors — weather, calendar, todo, fitness summary — is now table stakes. the system can build it. you can't compete on "we have this data prettier."
- deep-data widgets are the new moat. widgets that surface state from inside your app (your specific data, your specific workflows, your specific charts) are harder for gemini to replicate. those are the ones still worth shipping.
- widget-led aso just got noisier. "free widget for X" as a play store search query will see more competition, more ai-generated lookalikes, and fewer high-intent installs.
what to do this week#
three moves:
1. audit your widgets through the gemini lens. can a user describe what your widget does in one sentence and have gemini build it without your app? if yes, your widget is not a moat. if no, your widget is still defensible. write down which bucket each of yours is in.
2. update your store listing copy to lead with the data, not the widget. the new framing isn't "we have a great widget." it's "we surface data that gemini can't get to without us." rewrite your widget-focused screenshots and descriptions accordingly.
3. get your store listing tight before i/o noise drowns out everything. there's a one-week window between today and the i/o main keynote on may 19. it's the cheapest time to ship aso changes all quarter. stora handles screenshot updates, localized listing copy, and the compliance check on the new metadata — fast enough to actually fit in this week.
the bigger shift#
create my widget is a tiny feature with a big implication: the system can now generate ui that used to require an app install. it's the first instance of agentic android eating the easy cases. the apps that survive will be the ones that own the data, the trust, or the depth — not the surface.
your widget is no longer the wedge. your app has to be.