Google's Lyria 3 Pro lands in the Gemini app and lets anyone create full 3-minute songs from a text prompt - no music degree required. OpenAI quietly kills Sora as the AI video wars get brutal. Shopify merchants wake up with 800 million new potential customers overnight. And Anthropic's most powerful AI model leaks before it was supposed to exist - and it's so capable, they're scared to release it. What a week.

Hi! Welcome to the 42nd 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.

_{~5 mins read}

/Music /Gemini /CreativeAI

Last month, Google dropped Lyria 3 into the Gemini app and let people create 30-second AI music clips. Cute. This week, they went full album mode. Lyria 3 Pro is here, and it lets you generate full songs of up to 3 minutes, complete with proper song structure - intros, verses, choruses, bridges and all.

This isn't just longer background noise. Lyria 3 Pro actually understands how music works. You can now prompt it for specific song parts, layer in your preferred vocal style, tempo, and instruments, and end up with something that sounds like a real track. Grammy-winning producers are already using it for film scores. For the rest of us, the real use case is immediately obvious: royalty-free, custom music for your content, your company presentation, your podcast intro, or even just a personalised birthday song that'll make your girlfriend think you've secretly been to music school.

It's available now in the Gemini app for paid subscribers (AI Pro and Ultra plans), with daily track limits tied to your subscription tier. Importantly, all tracks are watermarked with SynthID so AI-generated audio can always be identified. The future of music production just got very, very democratised.

/Commerce /ChatGPT /Ecommerce

Here's a number that should make every Shopify merchant sit up straight: ChatGPT has roughly 800 million users. As of this week, every single one of them can now discover and buy from Shopify stores without ever leaving their chat window.

Shopify has launched Agentic Storefronts, which automatically plugs your store into major AI platforms - ChatGPT, Microsoft Copilot, AI Mode in Google Search, and the Gemini app - all managed from your existing Shopify Admin. No separate integrations. No extra apps. No additional transaction fees. Your products simply become visible to anyone browsing or asking questions across these AI channels.

This is significant. We're watching the birth of an entirely new sales channel - one where people don't search for a product and then find your website, but where an AI recommends your product in the middle of a conversation. Shopify maintains that merchants stay the merchant of record, keep their customer data, and retain full control of the checkout experience. If you're on Shopify, your store is essentially now inside ChatGPT by default. If you're not on Shopify, they've launched an Agentic Plan that lets any brand add products to their catalogue to access these AI channels. The era of agentic commerce is officially open for business.

/OpenAI /VideoAI /Strategy

Sora was supposed to be OpenAI's blockbuster video product. When it launched in 2024, the hyper-realistic AI-generated videos genuinely made jaws drop, and millions of clips were generated across the platform. There was even a headline-grabbing deal where Disney pledged a $1 billion investment and the prospect of licensing beloved characters to the platform. Big stuff.

This week, OpenAI quietly announced they're shutting Sora down. No clear timeline, no real explanation - just a brief post on X saying they know the news is "disappointing." It turns out Disney's $1 billion pledge had never actually been paid, and no formal licensing deal had been signed. Sora hit 1 million downloads in its first week, but regular users eventually found little day-to-day reason to keep coming back to a dedicated AI video app.

What killed it? Partly the market - Google Veo and ByteDance's Seedance have both pushed hard into AI video, making it a brutally competitive space. Partly strategic refocus - OpenAI has been channelling serious energy into enterprise and coding to compete with Anthropic's Claude, which has become the go-to tool for developers. The video dream is dead. The coding arms race, very much alive.

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

You know that feeling at the end of January when the southeaster has finally settled, the mountain is out, you've just done a trail run around Lion's Head, and there's an Aperol waiting for you at the Lawns? That specific Cape Town summer feeling that's almost impossible to explain to anyone who hasn't lived it? Well, now you can turn it into a song.

Here's how:

Pro tip: Keep refining the prompt - swap in your own memories, add specific instruments, or change the vibe entirely. The more detail you give it, the better it gets.

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

Anthropic - the company behind Claude - had quite the week. A "human error" in their content management system accidentally made a cache of internal documents publicly searchable. Cybersecurity researchers stumbled upon it and found draft blog posts describing a model the world hadn't yet heard of: Claude Mythos.

What is Mythos? According to the leaked material, it's a genuine step-change above anything Anthropic has released before - larger and more capable than Claude Opus (their current most powerful model), with dramatically higher scores on coding, academic reasoning, and cybersecurity benchmarks. Anthropic has confirmed the model's existence, stating it represents "meaningful advances in reasoning, coding, and cybersecurity." They've already quietly given a small group of early-access customers the chance to test it.

Here's the part that's genuinely unprecedented: Anthropic is deliberately holding back the release because of what Mythos could do in the wrong hands. The leaked blog post described it as "currently far ahead of any other AI model in cyber capabilities" - capable of exploiting vulnerabilities in ways that "far outpace the efforts of defenders." That's an extraordinary thing for a company to say about its own product. The plan is to release it first to cybersecurity defenders, giving them a head start in hardening their systems before the model reaches the open market.

The financial markets reacted immediately. On the day the leak went public, cybersecurity stocks took a serious knock: CrowdStrike, Palo Alto Networks, Zscaler, and SentinelOne all fell around 6%, Okta dropped 8%, and Tenable tumbled nearly 10%. It's a stark reminder of just how much weight AI announcements now carry in global markets - every time a frontier model leaks or drops, entire sectors move. Anthropic has since secured the exposed data. As for when Mythos officially launches - watch this space.

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 👋