🔈 Listen on Spotify

Hi! Welcome to the 50th edition of the TomorrowToday newsletter. 🎉

We’re here to decode the AI chaos so you don't have to. Think of us as your friendly neighbourhood tech translators - we cut through the chaos, translate the jargon, and spotlight new AI tools that matter for founders, builders, and curious minds.

Buckle up, because the future's moving fast and we're here to make sure you don't get left behind! ⚡

If you enjoyed today’s newsletter, please forward it to a friend & subscribe by following this link.

_{~6 mins read}

/Gemini /AI /Google /Search

Last week, Google held its annual developer conference, Google IO, and it was nothing short of the biggest AI mic drop moment of 2026. The company dropped over 100 product updates and announcements, almost all of them AI-powered. If there was any doubt about whether Google is serious about winning the AI race, that doubt is firmly gone.

Here are three things worth knowing.

Gemini 3.5 Flash is Google's new flagship model - four times faster than other frontier models, with benchmark scores that beat the previous Gemini 3.1 Pro across the board. It's already available to everyone in the Gemini app and via the API today. Developers are calling it the best price-to-performance model currently available.

Gemini Omni is the bigger story. This is Google's new multimodal powerhouse that can generate and edit video using nothing but conversational language. Think: "move the background to a beach" and it just does it. The first version, Omni Flash, is already live for paid Gemini subscribers. Google's Demis Hassabis says Omni will eventually be able to create any output from any input. That is a very large statement.

Google also unveiled their Intelligent Eyewear, developed in collaboration with Samsung, Gentle Monster, and Warby Parker. These are the best-looking AI glasses we've seen so far - a world away from the chunky tech-bro aesthetic of Google Glass. Audio glasses are launching this fall, with display versions to follow. They run on Gemini, support real-time translation and hands-free navigation, and are designed to actually look like... glasses.

/Voice /AI /Interaction

If you don't know Mira Murati yet, here's the quick version: she was the Chief Technology Officer at OpenAI - one of the most important people at the company that built ChatGPT. She left in September 2024, founded a new lab called Thinking Machines, raised $2 billion, and went quiet for a while.

This month, she resurfaced - and what she showed us is something genuinely different.

Thinking Machines released a research preview of what they're calling Interaction Models. Right now, every AI assistant follows the same walkie-talkie pattern: you speak, it waits, it processes, it responds. There's usually a 1-2 second lag that makes conversations feel robotic and unnatural. Thinking Machines' model - TML-Interaction-Small - processes audio, video, and text simultaneously, in real time, without waiting for you to finish. It responds in 0.40 seconds (compared to 1.18 seconds for GPT Realtime). It can interrupt. It can listen while it talks. It's what a phone call with AI could actually feel like.

Watch the demo here and read their technical blog here. This is not yet publicly available - it's a limited research preview with broader access expected later in 2026. But it's a clear taste of what the future of interacting with computers is going to look like. It's going to be next level.

/Productivity /Meetings /AI

If you've been following this newsletter for a while, you'll know we're big fans of Granola. It's our favourite meeting note-taking tool - it works offline, doesn't paste a clunky bot into your Google Meet, and captures everything without you having to lift a finger.

This week, Granola launched a new feature called Briefs. Before each meeting, Granola automatically searches your emails, the web, and your previous meeting notes - then distils everything into three crisp bullets: just what you need to know, right when you need it.

No more frantic Googling someone's name right before you jump on a call. No more realising halfway through a client meeting that you'd actually spoken six months ago. Granola does the prep for you.

In this section, we aim to spotlight an incredible AI tool or use case and guide you on how you can try it.

Google's Omni Flash is live right now and it lets you create and edit videos through simple conversation. Upload a video, a photo, or start from a text prompt - then shape the output just by chatting with it. Here's how:

Once it generates, keep refining through conversation: "Add a sunrise over the Drakensberg" or "Make it feel like a vintage Super 8 film"

Pro tip: Gemini Omni Flash is also available to YouTube Shorts and YouTube Create users at no extra cost, so if you're already on those platforms, you're already in.

In this section, we're spotlighting real businesses using AI to solve actual problems.

For nearly 80 years, a famous unsolved problem in mathematics sat gathering dust. The planar unit distance problem, first posed by Hungarian mathematician Paul Erdős in 1946, asked a deceptively simple question: given a set of points on a flat surface, how many pairs of points can be exactly one unit apart? For eight decades, the mathematical community believed the optimal answer looked something like a square grid.

This month, OpenAI revealed that one of their general-purpose reasoning models had done what no human had managed: it disproved that belief entirely. The model discovered a completely new family of geometric constructions that performs better than anything mathematicians had previously found. This marks the first time in history that an AI has autonomously solved a prominent open problem at the heart of a mathematical field.

What makes this particularly striking is that the model wasn't purpose-built for geometry or for this specific problem. It was a general reasoning model - the same type you'd use to draft an email or summarise a report. It just happened to be capable of holding together a long, difficult chain of reasoning, connecting ideas across distant fields, and finding a path that human researchers had missed for generations.

OpenAI is careful to note that this doesn't make human expertise less valuable - if anything, it makes it more so. AI can search the solution space, surface candidate proofs, and verify results at machine speed. But humans still choose the problems that matter, interpret what the results mean, and decide what to pursue next.

The real implication here is the trajectory. If a general-purpose AI can crack an 80-year-old unsolved maths problem in 2026, what does it do for drug discovery, materials science, or climate modelling by 2028? We're watching the beginning of something significant. And the researchers and engineers who learn to work with these tools - rather than alongside them in parallel - will be the ones who define the next era of science.

AI is full of jargon, and we’re here to decode it. Each week, we’ll give you a plain-English definition of a buzzy term you’ve probably seen (but never fully understood).

Thanks for reading TomorrowToday! We’d love to hear from you:

➡️ What would you like us to cover next?
➡️ Have a tool or topic we should feature?

We’re building this with (and for) you. 🚀
See you next Tuesday 👋