Google really, really wants you to know that we’re living in the ‘early days’

If you spoke with a Google employee this week at its I/O developer conference, I guarantee they uttered these words: “Early days.”

Mario Queiroz, a VP, started it. He introduced its Home device — “your personal Google around the house” — during the keynote address. “It’s early days,” he said.

The phrase abounded. It appeared in coverage of Google’s far-out stuff — virtual reality, augmented reality. It crept into its older, more mundane announcements, like the mobile web.

When I asked Googlers to show off Instant Apps, the Android feature that slices up apps like the web, they first assured me it was, indeed, “early days” with the product.

It was the go-to talking point in Google’s communication strategy for launching (rather, previewing) products in futuristic, untested fields of artificial intelligence and VR. The message here: If this stuff doesn’t meet your expectations, just wait; it can only get better.

That’s partly to dampen the hype. When these products launch, Google probably doesn’t want consumers demanding a full-fledged, “Iron Man”-style AI butler or incredibly high-end immersive VR out of the box.

“If you get it wrong, there’s a high cost to the user,” Aparna Chennapragada, director of Google Now, its predicative assistant tech, said onstage. “If an assistant tells you to get a car and go to the airport, and it’s not right, ‘What the hell?’”

So Google reminds the world that this tech is incredibly nascent, which is true.

But the trouble with this strategy is the general perception at I/O: Google is late to the game, not early. Amazon already has a voice-controlled home device. Facebook already has an AI-infused messaging system and a VR ecosystem.

But Google’s key ingredient is its machine learning smarts — smarts that can, in theory, push products well past the competition.

It’s not early days for that. At the final day of the conference, John Giannandrea, Google’s search chief, reminded the audience that Google has been working on this tech — like speech recognition and advanced machine learning — for years.

Still, he stayed on message. “It’s a journey,” he said. “It’s a very hard problem. We won’t be done for a long time.”

This article originally appeared on Recode.net.