If Google can’t make its AI agents useful, maybe no one can

For years, tech companies have promised that AI would give everyone a capable personal assistant, but they have delivered something more like a clueless intern. Over the past six months, this has begun to change, thanks in large part to the widespread open source AI agent platform OpenClaw. And among the top AI labs now striving for similar success, there seems to be one particularly qualified to make customers succeed at scale: Google.

At I/O 2026, Google announced new AI agents for collecting information, planning events, summarizing your inbox and calendar, and more. The agents can work continuously in the background, and the company claims that they will integrate seamlessly into Google’s own and third-party tools. It also expands developer tools and revamps research with additional AI capabilities. Some will be released this week, others will be available in the coming months, but the company’s strategy seems clear: adopt some of the features that helped fuel OpenClaw’s success and amplify them with Google’s deep knowledge of our digital existence.

“Before that, I think AI agents were just an idea in research,” said Koray Cavukoglu, CTO of Google DeepMind and chief AI architect at Google. Edge In an interview. This year, he hopes, they will be “really in our lives.”

AI agents have been a buzzword since ChatGPT launched in late 2022, but it remained mostly a science fiction concept until the advent of OpenClaw, which has gained millions of users since its launch last November. OpenClaw allows people to chat with their agents via everyday apps like WhatsApp and Telegram, and (as long as the laptop is open) the agents can work around the clock. They performed well enough to handle basic tasks reliably, albeit with Some obvious drawbacks.

This immediately made all the AI ​​labs sit up and take notice, but OpenAI was one of the first companies to take action, acquiring OpenClaw (although it remains open source) in February and hiring its creator, Peter Steinberger. But Google’s existing services empire gives it a big boost. While OpenClaw has led the adoption process by integrating with tools people already use, Google can do that too via MCPbut also creating deeper links to its internal product suite, including Gmail, Drive, Docs, Photos, and Search. If anything, it’s surprising it took so long.

One of Google’s big bets this year is Gemini Spark, its new AI agent for consumers. Google promises that Gemini Spark can perform tasks across Google’s own services and more than 30 soon-to-be third-party partners, including Dropbox, Uber, and Spotify. Gemini Spark is cloud-based. It can run 24/7 without keeping your laptop open and can sync across web, Android, and iOS. The agent is rolling out to trusted testers this week, and the beta will be available in the US next week on the Google Ultra plan.

Google touts typical uses for Gemini Spark, such as shopping, searching, and coordinating with other people’s schedules and plans. Google also hopes that people will find their own uses. Josh Woodward, head of Google’s Gemini app, says he’s been using Gemini Spark to plan a neighborhood party, deploying agents to track RSVPs and what attendees bring, send reminders, and find out when the homeowner’s association will allow a giant inflatable to be put up. Outside of Spark, Google also offers the Daily Brief, which is a morning update similar to OpenAI Pulse ChatGPT.

Gemini Spark isn’t available yet, but if it works the way Google says it does, it could be a big step forward for AI agents for traditional tech companies. Google’s first proxy experiments Tasks completed at a snail’s pace While your browser is hijacked. by Gemini 3 was released last yearIts agents worked well at some tasks — such as cleaning inboxes — but failed at others. Now, Google is taking a promising step by simulating some of these things Main elements OpenClaw: Long-running agents that run around the clock in the background, giving them the ability to get more context about their tasks – and giving users the ability to text and email their agents directly.

Starting this summer, Google’s AI search is also starting to get agents — and promises to finally do more than just eat up properties on screen and recommend Pizza with glue. Its “information agents” are supposed to conduct constant basic research – such as tracking shifts in the stock market or the weather to find the best day for a picnic.

Google also announced an expansion to Antigravity, the agent development platform it introduced about six months ago. Google says the new Antigravity autonomous desktop app will serve as a central hub for agent interaction, and the entire system is now designed as a platform for building and managing autonomous agents. The expansion comes on the heels of similar tools from OpenAI and Anthropic, which have tried to expand their successful programming services to include tools more accessible to non-coders.

All of this will be supported by a new model series: Gemini 3.5, the initial Gemini 3.5 Flash version of which should be available next month. The model is supposed to have much better coding capabilities than the Gemini 3, which it did He was released with great fanfare Last November. It’s clearly aimed at overtaking updates from Anthropic, known for its programming prowess, and OpenAI. Gemini 3.5 Flash is particularly good “when deploying multiple agents simultaneously and completing long-running tasks,” Cavukoglu told reporters on Monday. It’s also supposed to be four times faster than other frontier models and less than half the price (or in some cases, a third of it) — a fact that’s relevant to AI customers working 24/7, with nominal costs that add up quickly.

In the world of AI agents, Google will still be playing catch-up with the single team behind OpenClaw. But it is A long-standing front-runner in the race for artificial intelligence and its application It has an advantage of scale: It now serves more than 900 million monthly users, executives told reporters Monday, in more than 230 countries and more than 70 languages. Compared to dedicated AI companies Under increasing financial pressureit is able to subsidize costs at least temporarily to attract users. Although its clients have not yet had to face the real world, they are heading in a promising direction. If any AI company can make agents truly useful, it’s Google. If she can’t, she won’t have many excuses to fall back on — and the whole idea may need to be rethought.