🔈 Listen on Spotify

Hi! Welcome to the 63rd 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}

/Learning /Adoption

Here is a thing we say a lot: the AI tools have raced miles ahead of most people's ability to actually use them. Anthropic clearly agrees, because they've just launched Claude Academy, a free, official site that teaches you how to get the most out of Claude, from complete beginner right through to building with the API.

There are around 20 self-paced courses, and no credit card or sign-up wall to get started. It's organised into learning paths for each product: the Claude.ai chat you already know, Claude Cowork (the agent that does multi-step work for you), Claude Code (the coding tool), Claude Tag (for pulling Claude into Slack), and the Claude Platform for people building on the API. You start with the basics and work up to the more technical stuff at your own pace.

Why does this matter? Because a tool is only as good as your ability to wield it. Most people are using maybe 10% of what Claude can do. Free, structured lessons straight from the people who built it are the fastest way we've seen to close that gap. If you only do one thing this week, start the Claude 101 path.

/Agents /Automation

For the last few weeks, Grok Bot has been the thing everyone on X won't stop talking about, and until now it was locked behind eye-wateringly expensive top-tier plans (think R5,000+ a month). This week xAI opened it up: it's now included with SuperGrok Plus, Cursor Pro+ and all Cursor Teams plans, plus there's a 7-day free trial (a credit card is required, but nothing is charged unless you convert).

So what is it, and why the fuss? Grok Bot isn't a chatbot. It's a team of always-on AI “teammates”, and here's the clever part: each bot gets its own computer in the cloud, with a browser, a terminal and file access. You message it like a colleague; it signs into your actual tools, does the work across those apps just like a person would, and keeps going even after you close your laptop. It learns a whole workflow from watching you do it once, and you can run a whole squad of bots in parallel under one supervisor.

The reactions have been something else. Investor Naval Ravikant listed off bots that research his clients before he wakes up, test his mobile apps while he's at the gym, and draft content in his exact style. Gavin Baker reckons it's “another 'Claude Code' moment” and says his personal AI usage is up something like 100x. The shift people keep describing: you stop prompting AI and start delegating to it. Assign, execute, verify, instead of prompt, answer, copy, repeat.

Quick reality check: it's still an early beta; there's no way to pick which model runs a task, and the good tiers aren't cheap. But this is the clearest look yet at where agents are heading.

/Models /Cost

China's DeepSeek has dropped an experimental new model, DeepSeek-V4-Flash-Vision-Exp, that adds proper image understanding (reading screenshots, diagrams, media) on top of its existing text smarts. The headline that got everyone's attention: on some visual-agent benchmarks, it lands close to Anthropic's Opus 4.8. And remember, this is the Flash model, the small, cheap one.

Let's be honest about the numbers, because the hype gets ahead of them. On DeepSeek's own published tests, it actually beats Opus 4.8 on only three of eleven benchmarks, and trails on several others (sometimes by a wide margin). It's also worth noting that Anthropic has since moved on to Opus 5, and nobody has benchmarked this against that. So “beats Opus” is a stretch.

But here's why it still matters, and it's the same story we keep coming back to: price. Images are billed at the same rate as the standard Flash model, which is a fraction of what US labs charge for comparable vision work. You're getting near-frontier visual reasoning at budget-tier pricing. For anyone building agents that need to “see” (reading invoices, checking dashboards, processing forms), that cost gap is the whole game.

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, a member of the Claude Code team let slip a little skill that people inside Anthropic have been using a lot: ELI5 (short for “Explain Like I'm 5”). You give it a topic, and instead of a wall of text, it builds you a simple visual explainer with big pictures and few words. It's brilliant for getting your head around something new fast, whether that's a tricky concept, a bit of code, or why a decision was made.

It runs in Claude Code (Anthropic's coding tool, which you'll have met if you started the Claude Academy path above). Here's how to add it.

Don’t believe us? Try it yourself…

Pro tip: it shines for onboarding into anything unfamiliar. New codebase, new tool, new concept: point /eli5 at it first, get the big-picture picture, then go deep.

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

This week, the payments giant Stripe agreed to buy a three-year-old startup called OpenRouter for a reported $7 billion-plus (roughly R124 billion). If you've never heard of OpenRouter, that price tag alone should make you sit up.

So what does OpenRouter do? It's a single front door to the entire AI model world. Instead of signing up separately with OpenAI, Anthropic, Google, DeepSeek and dozens of others, a business plugs into OpenRouter once and gets access to 400+ models through one connection. It then does the clever bit automatically: routing each request to whichever model is best on price, speed and reliability for that specific task (yes, exactly this week's Dictionary term in action). It's grown to more than 10 million users, and token traffic through it has exploded roughly 30,000x since launch.

Why would a payments company want this? Here's the shift worth understanding. As Stripe's CEO put it, tokens are becoming the central currency that companies building with AI spend. Stripe already helps businesses optimise money flowing out (payments, fraud, authorisation). Now it wants to help them optimise the other huge new cost line: AI tokens. Owning the router means Stripe sits at the toll booth every AI request passes through, measuring usage, applying pricing and billing, all in one place.

And here's the bigger signal for the rest of us: payments and agents are about to collide. When your AI agents start doing real work (buying things, booking things, paying for services on your behalf), someone needs to handle the money side safely. Stripe just bought itself a front-row seat to that future. When one of the world's most disciplined payments companies spends R124 billion on AI plumbing, it's telling you where the value is quietly moving.

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 👋