Chrome Autofill Now Pulls Form Data From Your Gmail

Google flipped on a new kind of autofill on Wednesday: instead of remembering your name and address, Chrome now reads your inbox. The browser's enhanced autofill uses Gemini with Personal Intelligence to pull order numbers, tracking codes and travel bookings out of Gmail and drop them into whatever form is open on screen.

It is a small feature with a large implication. The context layer that AI agents need to act on your behalf is moving from third-party tools into the browser itself.

What happened

Google announced the rollout in a company blog post on October 28, with Chrome product manager Amanda Lim framing it as relief from switching between tabs or searching an inbox to find information.

The suggestions cover order numbers, shipment tracking, flight booking numbers and hotel reservations. Enhanced autofill already handled names, passport numbers, addresses and cards from Google Wallet. The new layer adds Gmail as a data source, filtered by the field the page asks for.

Availability is narrow: US users only, on desktop, Android and iOS, and only for Gemini in Chrome with Personal Intelligence switched on. Google says sensitive data stays encrypted and that users control what enhanced autofill can access from Chrome settings. The Verge reported the rollout started the same day.

Why Google is doing this

Every Personal Intelligence surface makes Chrome stickier. Once mail, wallet and browsing sit inside one AI layer, switching browsers means re-teaching an assistant your context from zero. TechCrunch compared the move to Apple's ability to lift SMS verification codes onto the keyboard: unglamorous, disproportionately useful, and very hard to give up.

Why builders should care

Form filling is the oldest entry point in automation, and the platform just took it. Products whose pitch was "we parse your inbox and complete forms for you" now compete with a free browser default.

The deeper shift is architectural. An agent that reads one app and acts in another is the core pattern of every n8n-style workflow: pull structured facts from Gmail, decide, then write into a system of record. Google is shipping a narrow, consent-gated version of that pattern to consumers, which sets expectations for the wider ones.

Two consequences stand out. Permission design becomes a product feature, because users are being trained to expect per-capability toggles. And the data path still runs through Gemini, so regulated deployments keep needing their own extraction stack, which is why local-first browser AI efforts like Opera's WebAI matter.

It also fits Google's push to open Gemini and Search to outside apps: the browser, not just the model, is where context lives.

What to watch

Whether Google extends Personal Intelligence beyond Gmail into Calendar, Drive and Keep, and whether Microsoft follows with Copilot in Edge. Either move would take the feature from convenience to infrastructure, and sharpen the question every builder faces: which context do you own, and which are you renting?