Hi! Welcome to the 16th 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.

_{~8 mins read}

/Ecommerce /Integration

Yesterday, OpenAI and Stripe dropped what might be the most significant shift in e-commerce since the "Add to Cart" button. Starting today, US ChatGPT users can buy products from Etsy sellers (and soon over a million Shopify merchants like Glossier, SKIMS, and Spanx) without ever leaving the chat.

Here's why this is massive: 800 million people use ChatGPT every week - that's double Amazon's traffic. Every "find me running shoes under £100" is a search not happening on Google. Every purchase is a transaction not flowing through Amazon. And OpenAI takes a small fee on each one.

The experience is seamless: Ask ChatGPT for gift recommendations, tap "Buy" on a product, confirm your details, and you're done. No browsing through endless pages, no fighting with checkout forms, no app installations required. The merchant handles everything (orders, payments, fulfilment) using their existing systems and keeps the customer relationship - but the transaction happens in chat.

OpenAI open-sourced the Agentic Commerce Protocol that powers this, meaning any merchant or developer can build integrations. For Stripe customers, it's literally one line of code to enable agentic payments.

This isn't just a feature - it's a fundamental reimagining of how commerce works. The question isn't whether AI will disrupt online shopping, it's who owns the buy button of the future.

🔗 Read the full announcement from OpenAI

/Design /Image /Ideation

Google just threw its hat into the creator ring with Mixboard, and it's clear they're coming for the design-obsessed crowd. Think Pinterest had a baby with a scrapbook, then gave it superpowers - that's Mixboard.

This experimental tool (currently in public beta in the US) lets you create moodboards using text prompts powered by Gemini 2.5 Flash. Type something like "show me cups, bowls, and plates in Memphis style" or "plan an autumn party in my living room," and watch as AI generates visuals to match your vibe. You can upload your own images, combine them with AI-generated ones, and edit everything using natural language with their new "Nano Banana" image editing model.

What makes this exciting for creators? It dramatically accelerates the concepting phase. Instead of spending hours hunting through stock photos and Pinterest boards, you can visualise ideas in minutes. Whether you're planning home decor, brainstorming product designs, or mapping out your next DIY project, Mixboard turns vague ideas into visual possibilities fast.

The catch: Not available in South Africa yet. You'll need a VPN to access it for now.

📹 Watch Mixboard in action

🔗 Try it at labs.google/mixboard

/Agent /Benchmark

Anthropic just released Claude Sonnet 4.5, and if you thought AI was impressive before, buckle up. This isn't just an incremental update - it's a paradigm shift in what AI agents can actually do.

The headline stats that matter:

Claude Sonnet 4.5 is now the best coding model in the world, scoring 77.2% on SWE-bench Verified (the benchmark that tests real-world software engineering skills). It can maintain focus for more than 30 hours on complex, multi-step tasks without losing the plot. For computer use - literally controlling your computer like a human would - it hit 61.4% on OSWorld, up from 42.2% just four months ago.

But here's what really matters: the price hasn't changed. You're getting frontier intelligence at the same $3/$15 per million tokens as Claude Sonnet 4. That's roughly 10-20x cheaper than running comparable extended thinking models from competitors, whilst matching or beating them on most benchmarks.

What makes this a breakthrough:

The real magic is in agentic capabilities. Claude Sonnet 4.5 can handle tasks that previously required human intervention at every step. The new Claude Agent SDK - the same infrastructure powering Claude Code - is now available to developers, meaning you can build production-grade AI agents with the same tools Anthropic uses internally.

What this enables:

We're talking about AI that can actually do things, not just talk about doing them. Software engineers are already reporting that Sonnet 4.5 completes tasks they'd normally budget days for in hours. The model's ability to reason through complex, multi-step problems whilst maintaining context means we're entering an era where AI agents can genuinely act as junior teammates

rather than glorified autocomplete.

🔗 Read the full release from Anthropic

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

This week’s challenge: Creating hyper-realistic AI videos that don't scream "I'm AI"

If you're in marketing, content creation, or just love experimenting with bleeding-edge tech, this one's for you. ElevenLabs just dropped a tutorial showing how to combine three powerful AI tools - ElevenLabs (voice), Google's Nano Banana (image editing), and Runway (video generation) - to create videos so realistic, your audience won't believe AI made them.

This isn't your typical AI slop. We're talking professional-grade output that can rival traditional video production, without the crew, equipment, or weeks of editing.

High-level workflow:

Why this matters: You can now produce marketing videos, explainer content, or social media posts at a fraction of the traditional cost and time. The quality bar just jumped several notches.

📹 Watch the full tutorial here and see exactly how it's done.

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).

In this section, we aim to spotlight something weird & wonderful in the world of AI.

This week: AI just aced the CFA Level III... and finance bros everywhere are sweating

The CFA (Chartered Financial Analyst) exam is notoriously brutal. Level III, the final boss of the series, involves complex essay questions that test analytical reasoning, investment strategy, and professional judgement. Humans typically spend 1,000+ hours studying across multiple years to pass all three levels.

AI models just waltzed through it in minutes.

Here's what happened: Researchers from NYU Stern and GoodFin tested 23 language models on mock CFA Level III exams. Nine of them passed. OpenAI's o4-mini scored 79.1% on the essay portion. Gemini 2.5 Pro and Claude 4 Opus hit 75.9% and 74.9% respectively. Even more interesting? Human graders consistently scored AI essay responses 5.6 points higher than automated grading systems - meaning the AI wasn't just technically correct, it was persuasive.

Two years ago, these same models failed the essay sections entirely. The leap in analytical and reasoning capabilities is staggering.

What this means for finance: The technical stuff - research reports, investment rationales, portfolio analysis - is increasingly AI territory. The human edge? Client relationships, contextual judgement, and the messy, emotional side of money that algorithms can't replicate (yet). If you're in finance and your job is mostly cranking out reports, it's time to lean into the human skills AI can't touch.

🔗 Read the full research paper

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 👋