And finally, back to Rick Osterloh, for a last look at “this year’s Made by Google family”.

“They look great together, and they work great together.”

And that’s it.

One last product from Juston: Google Clips, a lifebloggy-style camera that takes pictures automatically: “Turn it on, and it captures the moment, so you can be in the moment.” Stick it to your wall while you’re playing with a pet, for instance, and the camera will snap images when you’re both in frame.

Could be creepy, but Google’s arguing it’s not: “It looks like a camera, and has an indicator when it’s on, so people know what it is and what it does,” says Juston, “and all the machine learning is on the device itself, so nothing leaves it until you choose it to.”

It’s “coming soon” for $249.

There’s more. Juston Payne comes on stage to introduce the upgraded Daydream View headset, with new lenses and new fabrics. It has more than 250 VR titles (including the Guardian, natch), and you can now cast your VR experience to the TV. That’ll cost $99.

Google’s also bringing out some wireless headphones, to make up for the lack of headphone jack on the Pixel 2. Called Pixel Buds, they’re not truly wireless earbuds – they’ve got a connecting cable running between the two ears – but they do work directly with Google Assistant.

One genuinely cool thing: the Pixel Buds allow direct, voice-to-voice translation. Your phone hears someone speak, and your earbuds play the translation in your native language. Early days, but this feels genuinely futuristic, and – for once – not in a creepy way.

Google claims it can support real-time translation in 40 languages through the Pixel Buds.

Queiroz leads with the number: DXOmark, which said last year’s Pixel was the best smartphone camera ever, says this year’s Pixel is the best smartphone camera ever. They gave it a score of 98. (DXOmark’s habit of doing this has earned it a few knocks over the past year, to be fair.)

More interestingly, the Pixel 2 camera has a portrait mode, like the iPhone Pluses, but thanks to Google’s “computational photography”, it can do it with just one camera. That means it can take a high depth-of-field image from the selfie camera, as well as the back camera on both the large and small devices.

Continuing the, er, iPhone inspiration, Google’s launched “motion photos”, which are live photos: they take some video around either side of the image.

The phones also have a really impressive “fused image stabilisation” feature for video, combining optical and digital video stabilisation to produce a very smooth looking output. Normally, this isn’t really possible – the optical image stabilisation that works for still photos tends to throw the lens about wildly for videos, but Google says it’s managed to fix that digitally.

Of course, Google is all about the back-end: Pixel users get free unlimited storage on Google Photos.

In a direct swipe at the iPhone – the first explicit mention of the device all morning – Queiroz notes that Pixel users upload twice as many phones as iPhone users, and would run out of the free iCloud storage in just three months if they were with Apple.

He then leads on to a direct pitch: using a special accessory, iPhone users can switch in just 10 minutes or less.

The Pixel 2 will cost $649, and the Pixel 2 XL $849. Both can be pre-ordered today in six countries including US, UK and Australia, although Queiroz didn’t give a shipping date – rumour has it the XL will be shipping considerably later than the smaller Pixel.

To make up for that, perhaps, Pixel 2 owners will get a free Google Home Mini for a limited time.

Chennapragada introduces Google Lens. Hold up the camera to an advert, for instance, and you can automatically pull out the email address. Take a picture of a film poster or book cover to search for it. And so on.

Apparently Google wants to make that a verb: “just Lens it”. Which, coming from a company which hates that its own name actually is a verb, is A Bit Rich.

It’s also basically a product that Google launched three years ago, when it was called Google Goggles. I don’t doubt that Lens works a bit better, but it doesn’t seem that different.

Fresher is the other half of Lens: an AR-based sticker feature. It comes with a Stranger Things sticker set, courtesy of Netflix, so you can drop a cartoon monster and tween into the real world and make them fight. Like Pokémon but bleaker.

Chennapragada hands back to Queiroz to talk more about the camera.

Ellis introduces the “Now Playing” feature, which listens for music and tells you what it hears. Interestingly, it does this without connecting to the net, through an on-device database of songs, which will be some salve to the privacy-aware among us.

She then moves to another new feature: “At a glance”, which lives on the homescreen. A permanent widget, the data display will launch with support for calendar events, but eventually become – yes – AI-powered, to include traffic, directions, weather and more.

Google has borrowed an idea from HTC, and you can now squeeze the phone to enable the assistant. Which Ellis does, then takes a selfie.

And we’re back to more Assistant features: you can use your phone to broadcast a Broadcast on your Homes that you’re coming home, and then tell your phone “let’s go home” to get it to begin your going home “Routine”.

Ellis passes over to Aparna Chennapragada to talk about the phone’s camera.

Matsuoka gives way to Chandra again, who returns to talk about a new feature for the Home hardware called “Broadcast”: tell a Home to broadcast a message, and it will rebroadcast the command to every Home in the house – so a parent can say “broadcast ‘get ready for school’”, and the whole house will hear the message. It’s not clear if the kids also get the ability to broadcast “no, I’m staying in bed” back.

Chandra also says the Home has received a lot of support for children, with new understanding of kids’ voices, and support for the limited accounts that children under 13 can get linked to their parents’ main Google accounts.

Google’s partnered with Disney to bring a bunch of stories for families and kids, including Mickey Mouse, Cars, and Star Wars, and honestly, I’m quite jealous of the people with kids for that last one.

Following an extremely hyperactive advert for the Google Home Mini which left me exhausted just watching it, Rishi Chandra is back to talk about the Google Assistant.

The AI PA is getting a few new routines, allowing you to trigger a bunch of actions with just a few simple commands. For instance, you can now say “good morning” to get a news briefing, make your coffee, and then put on a playlist, all at once.

More generally, Google says it now supports more than 1,000 smart home products. And the next step for Google – with its Nest corporate sibling – is to drop voice control entirely, and move to a more proactive, AI-led world.

Yoky Matsuoka, Nest’s chief technology officer, comes on stage to talk about the ways this partnership works. Apparently you can ask your Google Home Mini to show you your backdoor video camera to find out if that weird noise you heard was your pet pig eating your son’s packed lunch. (Yes, that is the real example given. Apparently Matsuoka has an adorable pet pig.)

The Home can also tell you who is at your front door, by name, if you have a Nest doorbell, and the Assistant can shut your home down for the night as one of the new routines.

These things are nice, but they aren’t really that new. Most of the Nest features were already announced, and they all also work with the Google Home hardware that’s already available.

Response is a little muted from the crowd in San Francisco so far.

There’s a small but very enthusiastic group of mobile network partners at the front who are clapping and whooping at the minor announcements.

The Home is launching in Japan this week, Chandra says, to muted applause. Perhaps not the best announcement to lead with.

But then he moves on to one of the line’s biggest differentiators to Amazon’s Echo: the Voice Match feature, which lets the Home work out who’s speaking to it and give different answers depending on the questioner. So it’s rolling out to a further four countries on top of the UK, US and Canada.

In another move against Amazon, Home will also receive calling functionality in the UK “later this year”. Amazon just launched that last week.

Chandra hands over to Google Home designer Isabelle Olsson, who introduces the first new product of the day: the Google Home Mini.

Rick Osterloh opens the real product keynote, noting the purchase of HTC’s hardware wing, before launching into a video detailing the company’s successes over the past year: Google Home, Pixel, the 4K Chromecast, Daydream VR and more.

It’s all very much pumping music and #influencers, and I’m a bit overwhelmed.

Osterloh says the Google wifi is the #1 selling mesh wifi router in the US, which is nice but it’s still only a router, and he says 100m new answers have been added to the Home since it launched.

With a slight nod at the supply issues – you frequently couldn’t buy a Pixel for love nor money – he turns to the future. “The competitive field for smartphones is levelling off,” he says. “It’s going to be tougher and tougher for people to develop new and exciting products each year, because that’s no longer the timetable for big leaps forward in hardware alone”.

“Smartphones might be reaching parity in their specs, but we’re seeing huge breakthroughs in the kinds of experiences we’re able to deliver to users.”

We’re being set up for a phone with, perhaps, underwhelming hardware but Google-style top-class AI, by the sound of it. Osterloh gives the example of the Google Home, which has two microphones compared to the four, six or eight you find on competitors, but still manages the same standard of voice-recognition through leveraging machine learning.

“Radically helpful” is the tagline for today’s event it seems. “Everything is designed for you, to keep the tech in the background and out of the way … Our products are constantly getting faster and more helpful the more you interact with them, thanks to machine learning.”

The phones will come, but first, we move on to Google VP Rishi Chandra, to hear about the Google Home.

And so Pichai introduces the new triumvirate for Google’s products: “AI + software + hardware”. Most may think of Google as strong on software and weak on hardware, but this framing pretty clearly wants us to see the company as super strong on the first two, making up for a (perceived) weakness in the last two.

But for now, back to some (very) detailed looks at Google’s AI technology. Pichai introduces AutoML, a new tech that the company has built to make the process of creating and training AI easier. It’s made image classification and object recognition quicker, more efficient and more accurate, he says.

One final look at where Pichai wants to go, sharing a cute pic of some child baseball players chatting through Google Translate – he wants to improve it, so they can do it without needing to look at their phones – before he leaves the stage, to bring on Rick Osterloh, the company’s SVP of Hardware.

We’re off, as Google chief executive Sundar Pichai enters the stage. He begins by noting the terror attack in Las Vegas, and the suffering caused by the three Atlantic hurricanes that have hit the US over the past few weeks, before moving on to Google’s ongoing shift from being a mobile-first to AI-first company.

He cites examples like mapping 5,000km of roads in Lagos, Nigeria, using machine-learning technologies to read road names and house numbers from street-view images; using Google Maps data to predict parking spaces in San Francisco; or AI-powered machine translation to do millions of translations a day.

“This is why we’re excited about the shift to an AI-first world,” Pichai says. “At a higher level, it should be about computers adapting to people, not the other way round.”

He thinks this will happen in four ways: AI will become conversational, it will become ambient and always on, it will be “thoughtfully contextual” and it will constantly learn and adapt.

That last point “applies to things like security and privacy, as well”. Currently, he says, we focus too much on giving users options about security and privacy, but a good AI system will learn to treat a doctor’s appointment differently from a work meeting.

All this talk about how Google is AI-first, not mobile-first, may seem odd immediately before the company launches two new phones. But I think it will all make sense …

Hello and welcome to the Guardian’s live blog of today’s Google press event. It’s a big day for the company, which is expected to reveal its new pair of flagship smartphones, the Pixel 2 and Pixel 2 XL.

They’ll be the first new phones from the company in a year, since the original Pixel came out last August. It’s also the first we’ve heard from Google’s hardware team since they sort-of acquired HTC two weeks ago, bringing the team who built the Pixel in-house.

The main event starts at 9am in San Francisco (or 5pm in the UK), so you’ve got a little while to run off and grab a snack.