Revolutionizing Productivity: How Jace AI Aims to AUTOMATE Your Browser Work

A new technology comes along that could reduce by 90 per cent the time our brains spend on digital labour. Imagine an AI that could crawl the web for you, that could discern what you need done, and then go off and do the thing without your close supervision. Sound like the plot of a sci-fi movie? Not the next one, but the one after that. With its most recent AI agent, called Jace, the startup company Zeta Labs is poised to change how humans and computers work together. It’s a novel use of large language models (LLMs), the same kind of AI that has powered recent feats such as ChatGPT and GPT-4. Founded by three former Meta engineers, Zeta puts an LLC driving a user’s browser – all while a human monkeys around finding things on the internet.

The Genesis of an Idea

It started when Peter Albert, one of the two co-founders of the company, grew frustrated by some of the mundane chores of running an ecommerce business. When the Greater-Than-Person (GPT) era started to come into clearer view, it became possible to consider whether such things could be automated by AI. Albert, Fryderyk Wiatrowski, his former colleague at Meta, and other members of the Zeta Labs team saw that it could, seeding the idea for the company, and for Jace. They wanted to liberate themselves from the burden of menial digital labour. Zeta Labs was founded with that goal in mind.

Unleashing Jace: The Autonomous Assistant

Jace sets itself apart from the crowded field of AI by offering to carry out in-browser tasks autonomously if you give it a command. It is already drawing early adopters. Recent $2.9 million pre-seed funding, led by Daniel Gross, formerly of Asana, and Nat Friedman, formerly of Microsoft and GitHub, is a signal that Jace has an exciting prospect. If you want to book a hotel or hire an entire football team, Jace can do it for you. It’s what the next stage of AI might look like.

A Closer Look: Jace at WORK

Under the hood is Jace’s dual-model architecture: the LLCM for comprehension and planning, and its own AWA-1 model, which provides the actions that can take place on the page. Give Jace any of a range of tasks – managing emails, setting up marketing flows, launching a startup – and it will do so, resulting in an AI chatbot that isn’t just there for conversation. It is, instead, a web agent.

Unlocking New Possibilities with Jace

Jace’s potential use cases are manifold. For example it could allow any size to medium enterprise to automate repetitive web-based tasks, freeing up time and giving them a productivity boost across practically any business sector, from recruitment and ecommerce to marketing and sales.

Evolving with Jace: What the Future Holds

Though Jace has already shown immense promise, Zeta Labs is not resting on its laurels. Realising the potential in the technology, the company is continuing to invest in improving Jace’s abilities and sharpening it for general release. The second generation of its AWA model will be even more speedy and adept at handling more image-heavy, task-focused workloads. The company also plans to offer the technology on a subscription basis, making it more easily accessible to the public and businesses alike.

A Step into the Future

In this way, it becomes a reflection of the potential of cooperation, of ingenuity and imagination. Not simply about automation, but subordination: not just about leaning on AI, but about learning it – how to make it an extension of ourselves, taking charge, making it work for us, not the other way round.

Understanding "MOVE" in the Context of Jace

This qualifier ‘move’ is an important one. Jace’s ‘moving’ stands for all sorts of things that we are going to associate with AI in general and the evolving Jace in particular. It’s our sign for an end of the work we have to do, but it also involves the end of a way of characterising and thinking about AI: putting things in place, settling for a final version, and then putting it to work. It’s a sign for the end of static theory, whether it’s the conceptual and social AI models discussed earlier, or more recent systems and frameworks for Machine Learning. Jace is a moving and evolving thing. Earlier types of AI, including the most recent versions, are static. Think of them as spread out over a sheet of paper. Jace and his successors exist in an open loop in the cloud.

The Epitome of Efficiency

Jace is more than just a tool – it is an agent, the first of what TechCrunch’s Alexia Tsotsis in 2013 dubbed the ‘hands-off’ computer: that is, a piece of technology that exists to act, rather than simply to inform. In allowing users to outsource an ever-growing range of browser‑based chores to an autonomous service bot, Jace essentially eliminates the old, click-heavy, attention-intensive human element of our digital work lives in favour of new and potent efficiencies.

It’s a glimpse of a world where AI agents will be able to live more autonomously in our digital lives – a boon for users and businesses alike In the months following that demo, Zeta Labs has further refined Jace’s functionality. But more than that, the company has started devising real tasks for it to do, worries for it to handle on behalf of customers. ‘Imagine your job,’ Murray says, whether it’s selling movie tickets or picking out mattresses online, ‘and consider all the tiny little details you have to keep track of. And then multiply that times 10.’ If Jace and its successors are to do more than simply move data from point A to B, if they’re to lend a hand in performing tasks for the rest of us, their cognitive boundaries are going to keep expanding to take on more and more of those messy human worries. It’s a glimpse of a world where AI agents will be able to live more autonomously in our digital lives – a boon for users and businesses alike. Need a café table booked? No problem. Want someone to sift through hours of footage to find a dog? Done. Murray believes that it will only be a matter of time before such software tools are doing so many things that we can’t even conceive of them. ‘I’m always astonished by how rapidly things progress in this area,’ he says. ‘As long as you can articulate something in words, it’s something that could be done.’ And for Zeta Labs, all of this functionality is a long way from its original, experimental goal: to make the cobbling process more interesting for humans.