Ex-Google employees say we need ‘an Android-like moment for AI’

Hugo Barra, Google’s former VP of Android product management, announced Wednesday that he is leading a new startup with aims to develop an Android-like operating system for AI agents.

“[We’re] going back to our Android roots, building a new operating system for people & AI agents,” Barra wrote in a post on X.

I’m starting a new company with some of the best people I’ve ever worked with, and could not be more pumped. We’re calling it /dev/agents.

Going back to our Android roots, building a new operating system for people & AI agents. Check out @dps‘s post below for more.

We’re…

— Hugo Barra (@hbarra) November 26, 2024

The company, called “/dev/agents,” is working to develop a cloud-based “next-gen operating system for AI agents” that will “work with users across all of their devices,” company co-founder and CEO David Singleton wrote in a post on X. He argues that AI agents will “need new UI patterns, a reimagined privacy model, and a developer platform that makes it radically simpler to build useful agents.”

As the current generation of large language models like GPT-4o, Llama 3.1, and Gemini 1.5 face diminishing performance returns despite developers pouring more and more amounts of training data, compute power and resources into them, AI agents are increasingly seen as the next major advancement in generative AI technology. These agents, unlike traditional apps, are designed to autonomously process information, make decisions, and perform specific actions on their user’s behalf. That could be anything from generating complex computer code to booking flights and hotel accommodations, to transcribing business meetings then generating actionable tasks based on what was discussed.

Here’s how the new company’s website describes its mission: “Modern AI will fundamentally change how people use software in their daily lives. Agentic applications could, for the first time, enable computers to work with people in much the same way people work with people. But it won’t happen without removing a ton of blockers. We need new UI patterns, a reimagined privacy model, and a developer platform that makes it radically simpler to build useful agents. That’s the challenge we’re taking on.”

The industry’s leading companies are already racing to deploy their own branded agents. Microsoft recently announced that it will incorporate agents into its 365 Copilot ecosystem in early 2025. Google’s Project Jarvis, which is expected to arrive with the next Gemini update, leverages the AI’s capabilities to execute common tasks, such as visiting websites and filling out online forms, at the user’s command.

OpenAI’s agent, code named Operator, will function in much the same way when it is releases in January as a research preview through the company’s developer API. Anthropic has already released its agent, dubbed Computer Control, which empowers Claude to emulate the keyboard presses and mouse clicks of a human user.

“We can see the promise of AI agents, but as a developer, it’s just too hard to build anything good,” Singleton told Bloomberg, noting that the industry needs “an Android-like moment for AI.”