🔈 Listen on Spotify

Hi! Welcome to the 56th 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}

/Microsoft /Enterprise /Deployment

Here's a plot twist nobody put on their 2026 bingo card. This week Microsoft launched a whole new operating business called Microsoft Frontier Company, backing it with a $2.5 billion investment and roughly 6,000 engineering and industry experts. Its job isn't to build a shinier model. It's to get the AI tools that already exist working properly inside real companies.

And Microsoft isn't alone. Just two days earlier, Amazon's AWS committed $1 billion to its own deployment venture, and both OpenAI and Anthropic have set up similar outfits in recent months. That makes Microsoft the latest of several giants to plant a flag in the same patch of ground.

So why does this matter to you? Because it's a quiet admission from the people building this stuff: the model is no longer the hard part. Getting a business to actually adopt it, wire it into messy real-world workflows and see a return - that's the hard part. Everyone underestimated just how tricky the "last mile" of AI would be, and now billions are being thrown at fixing it. The tech has arrived. Making it useful is the real race.

/Claude /Coding /Productivity

If you've used Claude, you'll know Artifacts - those live, interactive pages Claude builds for you right inside the chat. They've been part of Claude Chat and Claude's Cowork tool for a while. Now they've landed inside Claude Code too, the version of Claude that developers use to build software.

Here's the neat part. You ask for an artifact, Claude writes the code, publishes it live to a private link, and keeps updating it in real time while it carries on working in the background. Think a walkthrough of the changes Claude just made to your project, a living dashboard that tracks a build as it happens, or a self-contained little page you can share with your team without emailing screenshots around. The pages stay private to your account.

Best of all, it's not locked behind the enterprise paywall. Artifacts in Claude Code are now available on the Pro, Team, Max and Enterprise plans, so solo builders and small teams get to play too.

/Google /Gemini /Agents

Google's Gemini Spark is a 24/7 agentic assistant, meaning it doesn't just answer questions; it goes off and does things for you. This week it landed on the Mac desktop app, putting it head to head with the likes of Claude Desktop and Microsoft's Copilot. On your Mac, it can now dig through your files, then use them as the raw material for a new Google Doc or spreadsheet. Google's example: point it at a folder of invoices and ask it to build you a budgeting worksheet.

The more interesting upgrade is who Spark now talks to. It's added support for third-party apps like OpenTable, Instacart, Canva, Dropbox and Zillow Rentals, which means you can ask it to book a table, order the weekly groceries, design a flyer or line up apartment viewings. It can also now keep an eye on things in real time - sports scores, share prices, breaking news, even the weather - and it's rolling out custom MCP support so you can plug in your own favourite apps (see this week's AI Dictionary for what MCP actually means).

One honest catch for local readers: Gemini Spark for macOS is in beta and, for now, only available to Google AI Ultra subscribers in the US. So this one is a "watch this space" for South Africa rather than something you can switch on this afternoon. Given how fast Google has been rolling things out, though, we'd expect wider access before long.

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

Let's be honest with each other: we all know you're a little bit addicted to scrolling YouTube Shorts and Reels. No judgement, us too. But every now and then a Short actually teaches you something, and you catch yourself thinking "why isn't all my studying this easy to watch?"

Well, now it can be. Google's NotebookLM can take your own study material and spin it into short-form video overviews - bite-sized clips about the exact thing you're trying to learn, served up in the format your brain is already hooked on. Cramming for an exam, wrestling with a dense report, or upskilling on something new? Feed the material in and get back a Short that teaches it right back to you.

Don’t believe us? Try it yourself…

Pro tip: the more focused your uploaded material, the sharper the video. Feed it one topic at a time rather than your whole semester in one go, and you'll get something you can genuinely revise from. (We tested this one ourselves, and yes, it actually works.)

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

Everyone has a hot take on AI and jobs. Almost nobody has actually measured it. A new study from Ramp and Revelio Labs (June 2026) finally does - and the findings cut hard against the doom headlines.

The clever bit is how they measured it. Most research on this has had to guess who's really using AI, either from "exposure scores" that rate which jobs AI could theoretically do, or from surveys that wildly disagree with each other (the US Census reckons 17-20% of firms use AI, while executive surveys claim closer to 78%). Ramp took a different route: it followed the money. Because Ramp is a corporate card and bill-payment platform, the researchers could see every actual payment to AI vendors - OpenAI, Anthropic, GPU clouds, coding tools - across 21,559 US companies, matched against monthly headcount records. You can fudge a survey answer. You can't fudge a cleared invoice.

Then they split those companies by how seriously they were spending. "Light" adopters were paying around $2.78 per employee per month - a few chat subscriptions and some dabbling. "Heavy" adopters were spending $33.67, about twelve times more, the level you only hit when AI is wired into real workflows with agents and proper integrations.

That split changed everything. Over the two years following adoption, the heavy adopters grew their total headcount by 10.2% compared to similar firms that hadn't adopted yet. The light adopters? No measurable change at all. In plain terms: buying a few subscriptions did nothing. Rebuilding how the work actually gets done grew the company.

Two findings really jumped out at us. First, the payoff takes time. Nothing much happens in the first quarter; the curve only starts bending around month six and then compounds. So that three-month pilot a company quietly killed for "no ROI" may simply have been switched off before the good part kicked in. Second, and most surprising given all the panic: heavy adopters actually grew their entry-level hiring by 12%, with juniors growing faster than the rest of the firm. And it wasn't just engineers - sales, admin and customer service, the exact roles AI is "supposed" to wipe out, all expanded. These firms weren't swapping people for software. They were scaling the whole operation.

Now, the honest caveats, because that's how we roll. This isn't a clean lab experiment, and the sample leans towards modern, US-based Ramp customers, so it's not a perfect mirror of a business in Cape Town or Joburg. The measurable gains are also concentrated in the tech sector for now, where coding is the most mature use of AI. And crucially, firm-level growth doesn't rule out job losses elsewhere: if the heavy adopters are winning market share, some of that is coming out of competitors who may well be shrinking. This study measures the winners' side of the ledger, not the whole board.

Still, the practical lesson is hard to ignore. The line between firms that grew and firms that saw nothing wasn't whether they "used AI" - it was whether they invested in genuine capability: agents, integrations, rebuilt workflows and teams who actually know how to use the tools. Subscriptions are not a strategy. If you're weighing up your own next move, the takeaway is to budget for a 6 to 12 month learning curve, think expansion rather than replacement, and treat the next "AI made us do layoffs" headline with a healthy pinch of salt.

Source: "A New Look at AI's Impact on Jobs", Ramp × Revelio Labs, June 2026.

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 👋