OpenAI’s public launch of ChatGPT in November 2022 is widely viewed as the event that kicked off the generative AI boom, which remains the dominant theme in the tech industry almost three years later.

Notion jumped on the bandwagon early.

Two weeks before ChatGPT hit the market, the productivity software startup announced its own artificial intelligence feature using an OpenAI model. Notion AI was designed to be a “writing assistant” that could help a user with brainstorming, editing and summarizing, the company said.

“We’re at an important inflection point,” CEO Ivan Zhao wrote in a blog post at the time. “The potential of artificial intelligence has grown exponentially, and will continue to grow.” 

The AI wave has pushed Notion past $500 million in annualized revenue, the company now tells CNBC, which ranked the company 34th on its 2025 Disruptor 50 list. The latest developments landed on Thursday as Notion launched a customizable agent that can create documents to pull in data from many sources, using models from the likes of OpenAI and Anthropic.

Akshay Kothari, Notion’s co-founder and operating chief, said in an interview that the company is racing to keep up with enterprise demand for AI tools. Corporate clients include Kaiser Permanente, Mitsubishi Heavy Industries, Nvidia and Volvo Cars.

“We’re doubling this year and likely going to double the sales team next year,” Kothari said. He added that about 90% of the business comes from “multiplayer usage,” or teams of workers.

Notion was founded in 2013, two years before OpenAI was created as a nonprofit AI lab. The company, which now has about 1,000 employees, launched the first version of its product in 2016 and says it has over 100 million users.

But unlike most startups that have boomed with the rise of generative AI, Notion hasn’t raised outside capital in a long time. Its most recent fundraising round came in 2021, when the big driver for cloud-based collaboration software was the Covid pandemic and remote work. In October of that year, Notion raised $275 million at a $10 billion valuation.

Kothari says the company has more cash on hand than the $330 million it’s raised to date.

In May, Notion introduced Ai products for summarizing meetings and searching through corporate files. Annual revenue growth has accelerated every month since then, Kothari said, though he declined to provide growth rates.

Thursday’s announcement includes the rollout of a preview of an additional feature called custom agents, which can perform actions in the background. As an example, a custom agent can be instructed to produce and send a list of articles that are relevant to a person’s interests every week.

Kothari said that last year 10% to 20% of Notion customers were paying for AI add-ons. That shot up to 30% or 40% earlier this year and recently crossed 50%, he said.

At that point, the company started including AI in its business and enterprise plans, without charging extra, Kothari said, adding that the company is talking with clients about a fair pricing model for custom agents.

Productivity software is a highly competitive space, with Microsoft and Google at the center.

Weeks after Notion’s big financing round in 2021, Microsoft announced Loop, an app for working on documents. The product, which resembles Notion, became available to organizations with Microsoft 365 productivity software subscriptions in 2023.

Microsoft is also pushing Copilot, an AI assistant that can spit out Word documents and Outlook emails. Google, meanwhile, offers the Gemini AI option for its Google Workspace applications.

Ramp, a business credit card startup, pays for the Gemini AI option for Google Workspace apps. But the company has encouraged people to migrate documents and project tracking to Notion, said Ben Levick, Ramp’s head of operations, in an email.

Levick said that nine out of 10 employees at Ramp, which has a workforce of 1,200, now use Notion’s AI features every month, and the company is testing custom agents to answer internal inquiries and to connect sales feedback with forthcoming products that could address requests from clients.

WATCH: Salesforce CEO Marc Benioff on what the market is getting wrong about AI

Google is adding more artificial intelligence into its Chrome browser as the search giant tries to fend off burgeoning competition from AI startups OpenAI and Perplexity.

In a blog post Thursday, Google said it’s rolling out Gemini in Chrome to users of Mac and Windows computers in the U.S. as well as to mobile devices. Users will be able to ask Gemini for help understanding the contents of a particular webpage, work across tabs, or do more within a single tab, such as schedule a meeting or search for a YouTube video.

“We are evolving the browser to help you get the most from the web – in ways we didn’t think possible even a few years ego,” said Rick Osterloh, Google’s senior vice president in charge of platforms and devices, in a statement. “And we are doing it while keeping the speed, simplicity and safety of Chrome that so many people love.”

Internet browsers are at the center of the battle for consumer AI supremacy because they serve as a key gateway to accessing information and content online.

Google and Apple have for years controlled most of the internet distribution points, which is a big reason the U.S. Department of Justice tried to force Google to divest Chrome as part of its antitrust case.

However, the judge in the case recently decided Google could keep Chrome, in large part because generative AI has dramatically changed the competitive landscape.

AI rivals are rolling out browsers to try and control more of the user experience. In January, OpenAI announced Operator, an agent that uses a browser to complete tasks, such as shopping using the Instacart app. The company is reportedly working on its own browser built upon the open-source Chromium code base.

Last month, Anthropic launched a browser-based artificial intelligence agent powered by its Claude AI models. Perplexity debuted its browser Comet in July for AI tasks, making it available to paying subscribers.

The new Gemini in Chrome integrates deeper with Google apps like Calendar, YouTube and Maps, so users can access those services without moving to a different webpage.

In the coming weeks, it will also be available to users of its enterprise productivity product Google Workspace, where users will have “enterprise-grade data protections,” Mike Torres, Google vice president of product, wrote in a blog post.

Google also announced new agentic capabilities for Gemini in Chrome. Agentic AI allows users to build customized services that can perform specific jobs.

In the coming months, users will be able to ask the Gemini agent to do certain tasks, such as booking a haircut or ordering weekly groceries. The agentic features were previously part of an internal project called “Project Mariner,” which was popular with employees.

Before Thursday’s announcement, Google required users to be signed up for certain Google subscriptions to access Gemini in Chrome. Now it’s more widely available with far more features.

WATCH: Dan Niles says Google will have the best AI app long term

Meta Chief Product Officer Chris Cox said on Thursday that smart glasses are the future of computing devices.

“We talk to them, we will see with them, we will use gestures the same way we interact with each other to interact with our computers,” Cox told CNBC’s Julia Boorstin. “The interfaces will get more natural, and so we certainly believe that the next really important wearable technology is going to be a pair of glasses.”

The $799 Meta Ray-Ban Display glasses, which were revealed on Wednesday, have a small in-lens display that is controlled with hand movements using a neural wristband.

Users will be able to record videos, as well as send messages via voice or physically using handwriting gestures on their knee, Cox said.

“We’ve started with just the basics, which is messaging, which we know is the thing people want to do in a more fluid way,” Cox said.

Unlike Meta’s previous audio-only Ray-Ban Meta smart glasses, the Displays allow people to see messages and watch videos.

During a demo, CEO Mark Zuckerberg unsuccessfully attempted to answer a video call from Meta tech chief Andrew Bosworth, as the button to accept the call failed to appear on the display.