🔈 Listen on Spotify

Hi! Welcome to the 62nd 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}

/Agents /xAI /DigitalLabour

We've spent most of 2026 talking about chatbots that answer questions. Grok Bot, which xAI launched in beta on 11 August, is a different animal. Instead of a chat window that goes quiet the moment you close your laptop, each "Bot" gets its own computer in the cloud, signs into the tools you already use (with your own logins), and works through multi-step jobs on its own, only pinging you when it needs a sign-off.

Think of it less like a smarter assistant and more like a junior colleague you hand a task to and forget about. xAI's own example is a sales bot that researches accounts overnight, scores leads, and leaves a queue of drafted emails waiting for your approval by morning. It's xAI's answer to OpenAI's ChatGPT Work: the race is no longer about who writes the best reply, but who can actually get the work done.

The catch? This is firmly a professional tool, not a free toy. Access is bundled into the pricey paid tiers (SuperGrok Heavy and the Cursor Ultra and Teams plans), with no free option at launch, and it's an early-days beta. But the direction of travel is clear: the frontier labs are all sprinting from "AI that drafts" to "AI that does".

/Anthropic /Regulation /Transparency

Here's one that has divided people. From 2 August, Anthropic started weaving an invisible watermark into everything its newer Claude models write. You can't see it; it doesn't change the meaning of the words, but it's woven into the text itself, so it survives a copy-paste and may even survive light editing. Generated files like PNGs and SVGs get a signed provenance tag instead. The point is to make it possible to tell, later on, whether a piece of content came out of Claude.

Why now? The EU AI Act's transparency rules kicked in on the same day, and Anthropic (along with OpenAI, Google, Meta and others) signed up to mark AI-generated content. Anthropic is just the first to actually ship a text solution, and it's doing it worldwide, not only in Europe.

Not everyone is thrilled. Some paying users are unhappy about the idea of their work carrying an "AI-made" tag, especially if Claude only lightly touched it (a proofread, a translation, a summary). A few have cancelled in protest. The balanced view: heavy rewriting is meant to dilute the mark, Anthropic is clear that a detected mark is not proof of anything on its own, and the other big labs signed the same code, so expect this to become normal rather than a Claude-only quirk. Worth knowing if you build on the API: the mark passes through to your end users, and the compliance obligations still sit with you.

/OpenAI /Productivity

OpenAI shipped a handful of handy upgrades this week (announced 14 August). Nothing earth-shattering, but the kind of small stuff that quietly makes the tool more useful day-to-day. Here's the run-through:

Quizzes. Say "quiz me on [topic]", and ChatGPT builds an interactive quiz right in the chat, questions, hints and all. A genuinely nice way to learn something new. This one is available on all consumer plans, free tier included.

Reservation search. Describe what you want in plain language ("a table for two next Saturday somewhere with a good outdoor patio"), and it searches booking platforms for open slots. Heads-up for local readers: coverage depends on which booking partners operate in your market, so it's patchier here in South Africa than in the US.

Google Drive in your Library (paid users). You can now pull Google Drive files into ChatGPT and even open and edit Docs, Sheets and Slides without leaving the chat. Hit the + menu and "Add from Library".

Better home-page suggestions (paid users). Higher-quality prompts on your ChatGPT home screen to help you discover what it can actually do.

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

This one's for our South African readers. 🇿🇦

Imagine sending a voice note that says "make me a lasagne and restock the fridge", and a minute later your Woolworths or Checkers cart is built, itemised and priced, ready for you to check out. That's Pelicart, a WhatsApp grocery concierge built right here in Cape Town.

What makes it clever is the model: it doesn't try to be a grocery store (no stock, no drivers, no warehouses). It sits on top of the retailers you already use - Woolies Dash and Checkers Sixty60 - as a conversational layer, and strips out all the tapping and scrolling. You talk to it the way you'd text a friend who happens to know every aisle. This is where a lot of commerce is heading: you'll talk to it, not click through it.

One honest note so you're not surprised: Pelicart builds your cart; it doesn't secretly buy anything. You link your own Woolies or Checkers account, review the cart, and check out yourself. It can see and fill your cart, but it can't touch your bank card. Fair.

Don’t believe us? Try it yourself…

It's free for your first 2 orders, then Pelicart Pro is R99/month for unlimited orders. Your groceries themselves are always paid to the store on your own account. (Not sponsored, just a smart local build worth backing.)

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

This week the story isn't a product; it's a number: $2 trillion. That's the valuation investors are reportedly pinning on Anthropic (the maker of Claude) ahead of a possible stock-market listing as soon as October. If it lands, it would be the largest IPO in history, bigger even than SpaceX. For a five-year-old company, that's staggering. In rand terms, we're talking somewhere north of R35 trillion, which is many times the size of the entire JSE.

So where does a number like that come from? Growth, mostly. Anthropic's annualised revenue was around $47 billion (roughly R850 billion) back in May, and its backers are betting that figure hits $100 to $120 billion by year-end. When a business is growing at that pace, investors start doing the kind of maths that produces trillion-dollar headlines. Anthropic has quietly won a lot of that revenue by selling Claude to businesses - think Claude Code for developers and enterprise deployments - rather than chasing consumer hype. That B2B moat is a big part of why serious money takes the figure seriously.

But it's worth keeping your feet on the ground. A few important caveats: the $2 trillion figure comes from investors, not from Anthropic itself, which has set no such target publicly. Running frontier AI is brutally expensive (every query costs real compute), open-weight rivals are undercutting on price, and turning that eye-watering revenue into actual profit is far from guaranteed. Some analysts are openly asking whether this is the AI equivalent of the dot-com bubble.

The lesson for the rest of us? Two things. First, the real enterprise value in AI is being built quietly in the boring places: developer tools and business workflows, not the flashy consumer demos. Second, valuation is a story people tell about the future; revenue is what's true today. Worth remembering whenever the next big AI number flashes across your feed.

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 👋