🔈 Listen on Spotify

Hi! Welcome to the 64th 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}

/Enterprise /CRM /Claude

For months, the loudest take in tech has been "SaaS is dead" - why pay for clunky software when an AI can just do the thing? But that misses the point. The software was never the valuable bit. The database was. All that customer history, the deals, the workflows, the rules your business has built up over decades - that's the moat. So when the world's biggest CRM and the maker of Claude partner, you should pay attention.

Last week, Salesforce and Anthropic announced Claudeforce, and it flips the model on its head. Instead of you logging into Salesforce and clicking around, you point Claude at your Salesforce data, and it does the work - pulling deal history, updating your pipeline, drafting the follow-up, tailoring the proposal - all while your business rules and permissions stay enforced in the background. It launches with a plugin carrying 37 prebuilt sales skills. The UI, effectively, becomes the AI.

Salesforce CEO Marc Benioff put it bluntly on the earnings call: "This nonsense of the SaaSpocalypse, I think it's time for it to stop." Reading between the lines, he is not wrong. The value was never the screens. It was the data underneath, and now that data is far more useful because an agent can reason over it.

And here is the fun part: they made the best AI launch video we have ever seen, with Matthew McConaughey walking you through a salesperson's Monday morning while Claude quietly clears the pile. Genuinely worth the two minutes - go give it a watch.

/Productivity /Agents /Automation

Up to now, if you wanted Claude to actually do something on a website, you had to hand it access to your own browser through an extension. A bit fiddly, and a bit nerve-wracking. That has changed.

Claude Cowork now has its own browser built into the desktop app. When a task needs a website, a browser opens right there in the side panel, and Claude navigates, reads the page, clicks, and - the big one - fills in forms. You can watch it work and jump in whenever you like. There is nothing to install, and it stays separate from your personal browser and logins unless you choose otherwise.

Why does this matter? Because most of the boring admin in your life lives on some portal that has no neat integration - a supplier login, a government form, an internal dashboard nobody ever built a connector for. This is exactly where Claude can now pull numbers, submit forms, and work through the tedious multi-step stuff while you get on with something that actually needs a human. It is rolling out across the paid plans on the desktop app.

/Commerce /Agents /Payments

Small announcement, massive implications. Grok Bot has teamed up with Link so it can now buy things on your behalf. You send your bot shopping, it finds what you need, and it completes the purchase. To be clear on the scary-sounding part: you still approve every single spend request, and the bot gets a secure, single-use card for each payment. It is not about spending your money unsupervised. For now it is US-only, with mobile coming soon.

But park the mechanics for a second, because the shift underneath is the real story. Shopping is moving from search to delegation. And here is the thing marketers are only starting to wake up to: marketing has always been built for humans - the pretty packaging, the emotional ad, the brand you feel good about. An agent doesn't care about any of that. It isn't nudged by a clever billboard or a limited-time sticker. It scans, compares, and buys.

So if your product isn't discoverable to an agent, or isn't the first or best thing it surfaces, it simply doesn't get bought. No amount of brand warmth saves you. This is a genuinely new way of shopping, and it quietly rewrites the rules of how you sell online. Worth thinking about now, before your competitors do.

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

Heads up - this one is for the more technical crowd. If "toggle" and "local model" don't scare you off, read on.

Here is the pitch: normally, every message you send to Claude runs on Anthropic's servers. But Ollama just launched a Claude toggle that lets you run open models (think Qwen, DeepSeek, Kimi, GLM) locally on your own machine, right inside the Claude Desktop app you already use. If your computer is beefy enough, that means effectively unlimited tokens with no per-message cost, and your prompts never leave your machine. For now, the one-toggle setup is on macOS.

Don't believe us? Try it yourself…

If the model picker comes up empty, you can point Claude Desktop's third-party inference at Ollama yourself using the default local address below.

A quick honest caveat: a local model won't match Opus or Sonnet on the really hard stuff. But for everyday tasks, keeping it private and free on your own hardware is a lovely thing. The full guide is on the Ollama blog.

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

We love a proudly local win, and this one is a cracker. Verascient, a Cape Town-founded AI startup, has raised $1.2 million (about R19.5 million) in an oversubscribed round led by Founder Collective - the firm that backed Uber and Airtable early - alongside Andrena Ventures, Cambridge Enterprise and Summit Ventures. That is serious, top-tier international money backing two South Africans.

And what a two. Co-founder and CTO Emile Ferreira taught himself to code at 12, was employee #6 at Replit (now valued north of $9 billion), built one of the world's first on-device AI assistants (200,000+ users, an innovation award) while still in high school, and then went and earned a distinction in his MPhil in Advanced Computer Science at Cambridge. He is, quite plausibly, one of the best AI engineers this country has produced. His co-founder and CEO, Keagan Stokoe, was part of the founding team at township fibre company Fibertime before running an AI consultancy, PurpleDorm, across the UK and SA.

Here is why the product matters, and why we are genuinely excited about it: The problem with enterprise AI was never the models - they are already brilliant. The problem is context. Your business knowledge is scattered across emails, documents, meetings, spreadsheets and the heads of long-serving staff, and an AI can't help with what it can't see. Verascient's answer is a temporal knowledge graph (yes, this week's AI Dictionary term) that pulls all that scattered knowledge into one place - preserving history, permissions and provenance - so agents can actually reason over your business. As Ferreira puts it, there is a big difference between giving your staff an AI tool and building a company that can actually operate around AI.

That is the real unlock, and it is being built by South Africans. Watch this one. We think they are going to do very, very well 🚀

Side note: If you want to implement an AI brain for your company, reach out to us, and we'll gladly connect you with the Verascient team.

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 👋