😺 Insane Google I/O news

PLUS: these demos make us question reality...

Welcome, humans.

It's another massive day in AI.

P.S.A. to Big Tech: Could we get a breather between releases? OpenAI and Google I/O back-to-back feels like celebrating Christmas Eve right after Thanksgiving. Seriously, I’m still recovering from last night’s dessert—let us process!

Here’s what you need to know about AI today:

Google demoed its most advanced AI assistant yet, Project Astra.
Gemini 1.5 Pro is making its way to the sidebar of Google Workspace.
Google launched new video, image, and music AIs.
Gemini 1.5 Pro’s latest context window can handle 1.4M words.

On Tuesday’s podcast: Pete breaks down GPT-4o and hidden gems from the launch that people aren’t talking about enough (Apple Podcasts, Spotify, YouTube).

Gemini is coming to all your Workspace apps, assuming it works.

Yesterday at its I/O developer conference, Google introduced a buffet of AI innovations. We’re at I/O IRL, diving deep to gather all the tea you need.

AI is here in Google Workspace. This means paying users will have a ChatGPT-esque assistant right beside your screen that knows everything your Google apps know about you (so literally everything).

Powered by Gemini 1.5 Pro, it now boasts a super-duper big context window of 2M tokens—meaning it can "remember" 1.4M words.

Why it matters: when you're working in Docs/Sheets/Gmail/Slides, you can ask Gemini to retrieve or summarize any content from all these apps:

“Hey Gemini, fetch the latest budget numbers from Sheets and pop them into this email.”
“Hey Gemini, condense the main points from the email chain with our marketing team into a new Doc.”

BTW: The killer use case from yesterday that no one’s talking about could be Gemini’s ability to search through all your emails, saving you the headache of ever having to search for an email again.

That is all assuming these features materialize...Google has a bad habit of inflating its demos.

While these features aren't exactly new, they’re available today—try Gemini for Workspace via Google One AI Premium for consumers or with a Gemini for Workspace add-on for business customers.

Looking forward, Gemini will take action, too: It’s not just about finding info; Google is prepping basic versions of AI agents that can perform tasks.

An agent to categorize all receipts in your inbox into a GSheet (link).
An email agent to return an order and schedule a UPS pickup (link).
An agent to manage tasks like updating addresses across multiple websites (link).

The next big thing (though probably not anytime soon) is AI Teammate. Picture it as a coworker popping up in chat groups, emails, and documents. If it knows the answer to a question, it’ll answer just like any other employee would.

FROM OUR PARTNERS

Work with the team that changed the way Warren Buffett does business.

Warren Buffett might be the greatest investor of all time, but even he could use a hand integrating AI into the way he works.

AE Studio’s team of stacked engineers, designers, and data scientists helped change the way Warren and Berkshire Hathaway do business.

And you can work with them, too.

Whatever your biggest problem at work, AE Studio can probably solve it. Their secret is treating your project as if it were their own.

Tell AE Studio about your work challenge or big AI idea today.

Other major AI announcements from I/O.

There’s the work stuff from I/O, which we find most practical, and then there are the mind-blowing demos that truly make us question the future of reality.

Let’s dig in…

Project Astra: Google’s ultimate AI assistant that can see and reason about what's around you. It’s similar to ChatGPT-4o and might power a future pair of glasses. Watch the demo above.
Veo, Google’s new AI video generator. We would be impressed if it weren’t for Sora, OpenAI’s (far superior) text-to-video model.
Imagen 3, Google’s newest text-to-image model. Genuinely impressed by this one. The people look very, very real (see here).
Music AI Sandbox, Google’s take on creating music using AI. Listen to the demos here.
NEW MODELS for the geeks:
1. Gemini 1.5 Flash—a smaller, speedier version of Gemini 1.5 Pro.
2. Gemma 2—Google’s best open-source models.
3. PaLI-3—a fresh open-source vision model.
4. Gemini Live—a talking feature for Gemini similar to ChatGPT Voice Mode coming later this year.

Around the Horn.

OpenAI cofounder and chief scientist Ilya Sutskever has officially departed OpenAI following the Sam Altman coup.
All users should have received access to ChatGPT-4o yesterday. The new voice/vision features will be rolled out over the coming weeks.
- Plus users will be able to send 80 messages every three hours on GPT-4o.
xAI is in talks to lease $10B in cloud servers from Oracle over the next few years.
Google’s AI search SGE is launching for all users in the US today and will reach over 1B users by EOY.

Wednesday Wirings.

Glue AI

Crazy stat: Half of all global venture funding for AI startups was allocated to companies in the Bay Area. SF remains the tech Mecca.

Atlan, which uses AI to bring order to “data chaos” at enterprises, raised $105M.
RunPod, which offers GPUs for AI workloads, snagged $20M.
Artisan, a builder of AI employees, $7.3M.
OpenAI’s Startup Fund raised another $5M.
Glue AI, an AI chatbot assistant informed about your business, raised from Craft Ventures.