Google is working on an AI that can buy things for you in Chrome

When it comes to AI Google has its hands in several different projects, including one called Project Jarvis which is said to have the capability to control your browser and make purchases for you.

Google spent a little bit of time earlier this year talking about how this sort of artificial intelligence would work. At Google I/O 2024, Google gave the public a glimpse at “agents” in its Gemini AI model that would be able to perform tasks for you. Google said it would “work across software and systems.” As well as “think multiple steps ahead” and have reasoning, planning, and memory. With these capabilities, it could remember things for or about you. Allowing it to perform tasks on your behalf. Such as purchasing a plane ticket or buying a pair of shoes online (or even returning them).

Google hasn’t talked much about these AI features since that panel at I/O back in May. However, a new report says Google has made some decent headway on this project. We know from a report earlier this week that Google is planning to announce and launch Gemini 2.0 in December. And it’s at that time that we may learn more about the AI features that Google is calling Project Jarvis.

Google Project Jarvis AI can take over your browser and automate everyday tasks

According to a new report from The Information, Project Jarvis is moving ahead at a swift pace. Google will reveal more about Project Jarvis in December alongside Gemini 2.0, the report says. It’s said to be similar to Anthropic’s Claude AI model, though it will surely differ in some ways.

Jarvis would work inside Chrome to take care of these smaller everyday tasks, but the report doesn’t mention if this would be for Chrome on desktop or mobile. That being said, Google’s demo at I/O earlier this year showed off returning a pair of shoes on the mobile version of Chrome. At the same time, Google followed that up with another potential use case that appeared to be represented by a desktop version of the Chrome browser. So it’s possible that Google has plans for Jarvis to work on both platforms.

Jarvis works slowly because it has to think for a few seconds before taking each action

In addition to Google’s plans to reveal more about Jarvis in December, the report also mentions a little about how far along Jarvis is. Not specifically in reference to how far Google is in its development cycle, but how efficiently Jarvis works. This may hint at how much longer it will be before users start to see these features available in Chrome.

According to the report, Jarvis still works slowly because it has to think for a few seconds before taking each action. This could mean that Google is still operating Jarvis in the cloud. At Google IO, Sundar Pichai said that Google was working on these AI features with security and privacy in mind. And that everything would be under the user’s supervision. So it’s likely that Google plans for Jarvis features to work on-device when they’re ready for user access.

On-device AI also works faster because it doesn’t have to reach out to the cloud and get the answers or information it needs before providing a solution. So it’s also likely that Jarvis features in Gemini will be faster by the time Google rolls these out.