Meet Simular, the startup that wants to build autonomous computers

Can computers become autonomous? That’s the bet behind Simular, a startup that creates AI agents that can use computers like humans do. IBM Think discussed agentic computers with Simular’s CEO and Cofounder, Ang Li, who worked at Google DeepMind and Baidu’s autonomous driving.

According to Li, when we interact with a computer, humans spend a lot of time making useless moves. For example, according to a study he conducted with his team, the average user spends up to five hours a day just moving a mouse.

“The keyboard and mouse—they are not natural,” he said in an interview with IBM Think. “Our generation is still used to that, but the previous one was not, and maybe the upcoming one won’t.” Instead, Li said, we should look inward at how humans operate to inform how AI evolves. “Technology tries to create something that’s easy to communicate, like voice speech, and tries to make this machine behave like humans, so we can interact in a more natural way.”

Simular envisions a future where computers function as fully autonomous systems. “Everyone thinks AGI [artificial general intelligence] acts like humanoid robots, self-driving cars,” Li said. “I feel it’s definitely one of the directions, but what if computers can do all the work I do? That’s already human-level intelligence.” Li believes this is achievable now, thanks to technological advancements like open-source frameworks and multiagent systems.

Simular launched on macOS earlier this spring as a native browser, aiming to take over the more mundane, repetitive tasks that can bog people down in work that isn’t fulfilling or impactful. Instead of running on the cloud, Simular’s agent runs locally via a native WebKit browser, operating autonomously behind the scenes and allowing users to step in at any time.

In Li’s vision, humans will stay in the loop and coexist with agents. “The agents still make a lot of mistakes, so it’s important for humans to have the ability to immediately intervene,” he said. “It’s just like the CEO of a company. If you want to send a paycheck or send money to another company, you always need someone to sign off, right?”

Agentic browsing is a crowded space, and the competition is heating up this year. The Browser Company just opened Dia, its own take on agentic browsing, to users of its former flagship product, Arc.

“AI is going to change how we interface with our computer and the things we turn to it for,” said Josh Miller, CEO of The Browser Company, on a recent podcast.

In January, OpenAI launched Operator, an autonomous agent for browser-based tasks. A few weeks later, Perplexity teased its own AI browser, Comet, which has yet to be launched. “The reason we’re doing the browser is that it might be the best way to build agents,” said Perplexity CEO Aravind Srinivas in an interview with The Verge last April.

Li, who published a paper with this team last fall about an open agentic framework that uses computers like a human, sees the pace of agentic AI innovation moving fast. “I think it’s good that more people are working on this,” he said.

He is convinced that humans, in the near future, will simply be able to focus on creative innovation. “If the technology is good enough, then everyone will carry a small box, or something even smaller,” he said. “Then [you] just talk to the box, and the box will do the job for you. You don’t have to look at screens. That’s the future I’m seeing.”