The Day Google Search Fought Back

Welcome to Briefly AI. Today: The day Google Search fought back.

Google overhauled Search last week at I/O, replacing blue links with AI agents and conversational interfaces. The backlash has been swift, vocal, and measurable. DuckDuckGo app installs are up 30% as users scramble for the exits. Let's talk about what just happened.

Right, so here's what Google actually did. At I/O on Tuesday, they formally retired the search box as we've known it for 25 years. You can now feed it images, PDFs, videos, even open Chrome tabs — not just keywords. The interface is multimodal, conversational, and deeply AI-driven. On paper, it's impressive. The problem is that nobody asked for it, and quite a lot of people actively don't want it.

DuckDuckGo's numbers tell the story. Installs of their app jumped 30% in the days following the announcement. That's not a blip. That's people voting with their feet — or their downloads, rather. The sentiment online has been consistent: users feel like they're being force-fed AI when what they wanted was, well, search results. Fast ones. Accurate ones. Not a chat.

Now, why does this matter beyond a bit of user grumbling? Because Google's entire business model rests on you using their search box. If that erodes — even slightly — the advertising revenue tied to it erodes too. And if competitors like DuckDuckGo can position themselves as the "just give me the links" option, they've got a wedge. It's early days, but Google's bet that everyone wants conversational AI in their search experience is now a testable hypothesis. And the early data suggests the hypothesis might be wrong.

What to watch: whether Google makes the new interface opt-in or keeps pushing it as default. Also whether this is a vocal minority or the start of a genuine user revolt. And whether Microsoft sees an opening with Bing.

Meanwhile, on a completely different note, a startup called OpenRouter just raised $113 million at a $1.3 billion valuation. If you haven't heard of them, here's the short version: they're a multi-model AI routing platform. You send them a prompt, they figure out which model — GPT, Claude, Gemini, whatever — is best suited to answer it, and they route it there. No vendor lock-in. No picking sides.

The reason this matters is that usage has grown five times in the last six months. That's not hype. That's enterprises and developers hedging their bets. The early AI era was characterised by picking a model and building around it. The current era is characterised by not wanting to be stuck with one. OpenRouter's valuation doubling in a year tells you the market agrees: the future is multi-model, not monogamous.

For businesses, this is good news. Prices come down when there's competition. Flexibility goes up when you're not locked into OpenAI or Anthropic or Google. Bad news if you're one of those three and you were counting on platform lock-in to keep customers captive. The AI market is maturing fast, and that maturity looks a lot like commoditisation.

Now this next one's a bit uncomfortable, but it's worth sitting with. Uber's president, Andrew Macdonald, said publicly this week that AI spending is getting "harder to justify." The company reportedly blew through its entire annual AI budget in four months, and now they're asking the question every CFO is quietly asking: are we actually seeing returns on this?

Here's what's interesting. Uber isn't saying AI doesn't work. They're saying they can't connect rising token consumption — the actual usage of AI tools like Claude Code — to measurable business outcomes. More AI usage isn't translating to more revenue, or faster delivery times, or happier customers. It's just translating to higher bills.

This is the conversation that's happening behind closed doors at a lot of companies right now. AI has been sold as transformative, and the pressure to adopt has been immense. But if you're spending millions and you can't point to where the value is, that's a problem. Not a technical problem. A business problem.

The psychological bit here is worth noting. There's enormous social pressure to be "doing AI." Nobody wants to be the executive who missed the boat. But Uber's candour — saying the quiet part out loud — might give others permission to ask the same question: is this actually working, or are we just afraid to stop?

What to watch: whether other large companies start publicly questioning AI ROI, or whether Uber becomes an outlier for admitting it. Also whether the AI vendors respond with better ROI measurement tools, because right now, it's harder to measure than it should be.

That's your lot for today. Google's redesign is driving users away, the multi-model future is here and funded, and Uber just said what a lot of boardrooms are thinking. Three stories, five minutes, and not a single use of the word "transformative." You're welcome. See you next time.