The central pitch of AI browsers like Perplexity’s Comet, Opera’s Neon, and DIA from Browser Company is that it helps you complete daily tasks more efficiently. These are agents that are limited to one browser, but Composite aims to build an agent solution that will help experts with tasks, regardless of which browser you are using.
The startup was launched earlier this year by Yang Hwang Yun and Charlie Dean. Yun is Uber’s former product manager, and Deane founded a company that sells proxying for servers. When he was on Uber, he noticed that many people around him were groaning in their browsers.
“I saw people from a variety of roles, including marketing, sales, recruitment, security engineers, and more doing a lot of boring jobs in the browser, which made me feel like this would stop me from using my education and skills completely.
The company said today it raised $5.6 million in seed funding in the round led by Nat Friedman and Daniel Gross venture NFDG, and has received participation from Menlo Ventures and Anthropic’s Anthology Fund.
Composite currently offers solutions for Mac and Windows. The setup is very easy. The agent simply installs a browser extension to use that browser. You can issue different commands between tools used on the web. Composite completes the work. For example, it can be useful to use related documentation to look at the Jira backlog for a bug, leave comments about high-priority bugs, or mark replicated bugs as resolved.
According to the startup, recruiters can search for candidates across sites and use them to draft personalized emails, security engineers can create vulnerability tickets based on alerts, and marketers can extract reports from a variety of sources to create short insight reports.
Yoon said other AI browsers and agents from companies like Openai and Prplexity are solving non-professional needs, such as shopping and ticket booking.
TechCrunch Events
San Francisco
|
October 27th-29th, 2025
“We are the ideal tool for professionals who want to set up workflows without technical knowledge. Composite is very good at atomic actions such as clicking on different elements of a website or typing boxes.
He added that it works on a browser where the agent is already logged in to the service, so no connector is required and it can work on different sites.
The tool has already proposed several tasks based on user patterns. However, in the coming months, the company is aiming to develop a better mechanism for automatically surface the tasks that composites can do on behalf of users. Startups are also working on how to schedule recurring tasks.
The startup said the tool is suitable for experts as it allows administrators to restrict tools, perform tasks locally, and users can define which websites are out of scope, as users do not need to switch browsers.
There is a lot of competition in the field where agents can work for professionals. Companies like Openai use their browser to perform actions. The concept relies on the context of the user within the app and is combined with other connectors. A common catalytic support highlight is to use the entire desktop as a context. Like a spreadsheet, there are a lot of startups working in context. Many of these startups are in the early stages, and there are questions about the efficiency of AI agents in the long run. Investors are ready to put money in, but startups working in this field have a lot to prove that justification.
Matt Kraning, a partner at Menlo Ventures, is confident in Composite’s outstanding abilities. He told TechCrunch that the tool is not overly technical and is very intuitive to experts.
“The composite material handles a variety of modalities and sites very well and is designed with professional use cases in mind. The tool is suitable for those who have to experience many tasks in a day across a variety of functions,” he said.
