Google is rumored to be releasing a new version of its powerful Gemini AI model later this year, and part of this will be ‘Project Jarvis’. This is an autonomous AI agent built into Chrome that can perform various web-related tasks on your behalf.
According to The Information, Project Jarvis would be able to visit different websites, fill in forms and even make purchases on its own — based on your instructions of course. Overall, this could be the biggest upgrade to Chrome since it launched in 2008.
This is similar to Claude with Computer Use, a beta product unveiled by Anthropic last week that lets Claude take control of your machine through constantly updating screenshots. It uses these to determine where to ‘click the mouse’ and perform tasks.
However, unlike Claude's broader computer access, Project Jarvis is specifically limited to Chrome browser and focuses on consumer tasks like shopping and travel booking, with sources saying the system currently needs several seconds between each action.
What is Google Project Jarvis
Google’s Project Jarvis, named for Tony Stark’s AI assistant in Iron Man, will take on most of your everyday web tasks. You’ll basically be able to tell Gemini ‘I have $500, find me a holiday for these dates, book it and just tell me the details’, and it will make it happen.
Its goal is to automate everyday web tasks in Chrome. Jarvis works by taking control of the browser, capturing screenshots of what is happening on screen then using its ‘AI Vision’ capabilities to work out where to click and what to type.
I just hope Jarvis finds a way to fill my bank account before going on a spending spree.
Jarvis won’t have control of your whole computer, like Claude, but it does work in a similar way. A future version may give you computer control if it comes to ChromeOS in the future, although there is no mention of that possibility in the Information report.
According to the The Information's report, it will come out in December with the next big release of Google Gemini, likely to be called Gemini 2. This will likely be on December 6, a year after Google released Gemini 1, although it may not get widespread release until 2025.
It is also not clear whether Jarvis will be available to everyone, or a developer preview similar to Anthropic’s Claude Computer Use. If widely available it may be restricted to Gemini Advanced subscribers or as an opt-in preview due to the rumored ‘few second delay’ in each task.
Jarvis, Computer Use and OpenAI’s Project Swarm are all examples of ‘AI Agents’, the next major development in the AI race. These basically automate everyday tasks, allowing you to simply ask the AI to do something and know it can get on with it. I just hope they find a way to fill my bank account before going on a spending spree.
More from Tom's Guide
- Android 16 could launch sooner than expected — here’s what we know
- Google Pixel 9a renders just leaked — no more camera bar
- Samsung Galaxy S25 Ultra dummy units reveal design change — what you need to know
English (US) ·