To understand Opera’s Neon browser, you need to recognize that it is not just a browser with an AI bot added, but a browser with three AI bots living side by side. This is both a strength and a weakness because while you don’t have to go out to tackle all of your AI tasks, knowing where to go and what AI tasks can be really confusing.
Last month, Opera started removing people from the waiting list for its artificial intelligence browser Neon. It is entering the increasingly crowded AI browser market, including Google’s Gemini-powered Chrome, Perplexity’s Comet and The Browser Company’s Dia. Unlike many of its competitors, Opera charges $19.90 per month, which sets a high standard for a product that most people get for free.
Neon has a similar configuration to other Opera browsers. It has features like a built-in ad blocker and VPN, as well as a sidebar where you can apply apps like WhatsApp and Facebook Messenger. AI features are placed front and center on the home screen and fresh tab screens. Below the search bar is a toggle with four settings: regular internet search; a chatbot called Chat; browser control agent called Do; and an artificial intelligence building agent named Make. (Opera says Neon is built on AI models from OpenAI and Google(but it doesn’t specify which model is used for which part of the Neon.)
Chat is the simplest feature, an AI assistant built into the browser that will likely already be familiar to users of any AI chatbot. You can access it at the beginning of your query, but it is also located in the upper right corner of the browser and can be accessed at any time. Chat easily handled our quick research queries and could answer specific questions about the page we were viewing, such as our request for a summary of the latest research in quantum computing. But it wasn’t perfect. Chat’s answers were so detailed that it seemed like you had to work really tough to get through them. When we asked Chat to collect and summarize the five most recent comments Edge stories, responded with about 400 words intended to inform us that there were none.
It was this exercise that showed us that Chat couldn’t always read the internet, but it gave the impression that it definitely could. Another time it was written that three of them had no comments Edge stories, when in fact there were four of them. We found it strange that Chat instead offered a guess, which is often what “early comments on tech news sites” contain.
Opera’s vice president of browsers, Krystian Kolondra, said the bot failed because we chose the wrong tool. The chat successfully recorded the number of comments listed at the top Edge articles when Opera tried it, Kolondra said. However, summarizing comments requires expanding the comments section on the page, which requires a click. Instead, we should have sent our queries to the AI Do agent, Chat’s more busy cousin that takes over your browser and performs tasks for you.
We tried a set of tasks: book a CrossFit class, book a massage for less than $50 at a local spa, and find PDFs of patterns for sewing baby onesies. While Do was working, we couldn’t switch back to Chat in the same window if we wanted to ask additional questions about the current task. There is also no way to correct Do’s course while it is running. We watched in horror as the bot scrolled past the perfectly pretty floral arrangements we had tasked it with finding a friend, only to add a monstrous funeral wreath to our cart even though we clicked on better options. On another occasion, Do stated that he had no theater tickets for the January show, even though a cursory check showed that there were plenty of them. It’s tough to trust anything after such unwavering but misguided certainty.
As with other AI browsers, performing actions with Do was also slower than doing it yourself, although it hinted at what outsourcing the general daily web surfing experience might be like. And using the To option doesn’t mean you can check out completely yet. Sometimes he encountered obstacles that only a human could cope with. When this happened, the Execute tab at the top of the screen flashed an easy-to-miss red tint, letting us know that we needed to step in and lend a hand the bot on its way.
In addition to Chat and Do, there is also an AI Make agent that can create petite web tools for you. Make exists on a virtual computer, where it downloads the necessary software, scripts or, in our case, images for your creation without cluttering up your personal computer. We asked for a elementary memory matching game with introductory Spanish vocabulary. We did it, and within a few minutes we matched the word “libro” to the book’s image. The game was clunky, but it was convenient to close the tab and know that all the pictures from the books disappeared with it.
Opera Neon’s final advantage is cards, effectively written prompts that can be used on any AI agent, which Opera says act as a “reinforcer” for interacting with the AI. Cards could theoretically save us from writing prompts, but at this point it’s tough for us to imagine the need to reuse a set of instructions. The app store-style interface is largely filled with content curated by the Neon team, from catchy prompts that rewrite pages as if they were spoken by Yoda to more solemn prompts for news aggregators. Opera hopes that as more people apply it, the platform will fill with useful user creations, but currently there are few.
At times, using Neon felt a bit like working with a hapless intern we never asked for, rather than sophisticated, time-saving technology. Often, one of the AI systems would ask for feedback and then simply perform the task without waiting for a response. Given the ability to apply a browser, it’s very basic to imagine where this proactivity could go very wrong, such as sending a ton of LinkedIn queries to people you wanted to anonymously stalk for professional purposes. One time we replied that everything looked great and that we should continue, and Neon said, “I’m glad you think so!” and it stopped working immediately. Kolondra told us that Neon will stop by for feedback in the future, but the feature is disabled because it’s not ready yet.
Kolondra admitted that work on Neon is currently in progress. “Overall, Opera Neon is in early access and is being made available to people who want to be involved in the development of this product,” Kolondra said Edge.
But Neon is also a paid subscription product, requiring $20 a month for something that’s mostly available for free elsewhere. It’s a tough sell at the moment – especially since Neon feels more like an AI browser that we have to adapt to rather than a browser that’s intelligent enough to adapt to us.