Hi! Welcome to the 11th 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.

_{~7 mins read}

/Productivity /Excel

Remember when Microsoft first promised AI would make Excel autonomous? Well, they're finally delivering - but not in the way they initially promised. Instead of a chatbot floating around your spreadsheet, Microsoft has introduced the COPILOT function that works exactly like any other Excel function.

Here's the game-changer: you can now type =COPILOT("Generate 10 realistic customer names", A1:A10) or =COPILOT("Classify this feedback", B4:B18) directly into cells. It's like having a VLOOKUP that can think, understand context, and generate content on demand.

The function integrates seamlessly with Excel's calculation engine, meaning when your data changes, the AI results automatically update too. No manual refreshes, no add-ins - just pure Excel magic powered by GPT-4.1-mini.

Where can you use it? Currently rolling out to Microsoft 365 Copilot license holders on Windows (Version 2509+) and Mac (Version 16.101+). Excel for the web is coming soon through Microsoft's Frontier program. For South Africa specifically, Copilot Studio agent builder is now available, but the COPILOT function rollout timing hasn't been announced yet - typical Microsoft keeping us waiting! For more information, please see the release video.

Bottom line: This is what Excel's AI integration should have been from day one - simple, powerful, and actually useful for real spreadsheet work.

/Design /Productivity

GenSpark, the super agent platform, just dropped its design agent, and it's genuinely impressive. We're talking about creating an entire brand identity - logo, colour palette, interior design concepts, packaging, websites, and social media assets - all from describing your small business idea in a single sentence.

The AI Designer uses a multi-agent architecture with nine specialised agents working together. One handles market research, another tackles visual identity, while others focus on packaging, web design, and marketing materials. It's like having an entire design agency in your pocket, minus the R50,000 monthly retainer.

We tested it ourselves, and the results are surprisingly cohesive. Sure, it's not quite at the level of a seasoned human designer with years of experience and that special creative flair we humans bring to the table. But it's remarkably close, and for a startup trying to build a professional brand identity without the budget for a proper design agency, this could be a complete game-changer.

The platform offers hundreds of customizable templates and can even design coffee shop interiors that look genuinely inviting. For bootstrapped founders who need to look professional from day one, this levels the playing field significantly.

Don’t believe us, look at their release video.

/MachineLearning /Data /Benchmark

Meet NEO, the first fully autonomous machine learning engineer that actually works like a real ML engineer. This isn't just another AI tool that helps with coding - it's a human-in-the-loop system designed to handle complete ML pipelines from data exploration and feature engineering to training, tuning, deployment, and monitoring.

Powered by 11 specialised agents, NEO can autonomously handle the entire ML workflow while keeping you in the driver's seat. You can guide, inspect, and override any step, but the system handles the heavy lifting. Think of it as having a brilliant ML engineer who works 24/7 and never gets tired of hyperparameter tuning.

The benchmarks are impressive: tested on 75 Kaggle competitions, NEO scored medals in 34.2% of them, significantly outperforming Microsoft's RD Agent. It's currently in beta, but the system is production-ready with native integrations for Snowflake, Databricks, BigQuery, and can be deployed in your own VPC.

This could genuinely democratise machine learning. Instead of needing years of ML expertise, you can now describe what you want to predict and let NEO handle the technical complexity.

If you are interested to learn more, watch their YouTube video.

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: Scrape property data like a pro with Firecrawl V2

Firecrawl V2 just launched, and it's the easiest way to scrape data from any website. Let's say you want to build your own property search tool by scraping data from Property24 to find flats in Cape Town.

Here's how to get started:

The beauty of Firecrawl V2 is that it handles all the technical complexity - JavaScript rendering, anti-bot measures, and data cleaning - so you can focus on building your actual application.

Pro tip: Always respect robots.txt files and website terms of service when scraping. Use the data ethically and consider reaching out to website owners for permission when building commercial applications.

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.

Who is behind the new nano-banana image model? 🍌

A new and mysterious model, dubbed nano-banana, has taken the AI world by storm, leaving a trail of hyper-realistic images and speculation about its origins. The model has appeared on LMArena, a platform where different large language and image generation models are tested against each other to determine which performs best in blind head-to-head comparisons.

Nano-banana has quickly become a standout performer on LMArena, consistently outperforming its peers in head-to-head comparisons. Its most striking feature is its ability to create stunningly hyper-realistic images that push the boundaries of current generation capabilities. But here's where it gets interesting - nobody knows who's behind it.

The emergence of this model has been met with significant intrigue, particularly due to subtle clues hinting at its potential creator. Following its appearance on LMArena, several prominent Google executives have been observed posting nothing but a banana emoji on their X (formerly Twitter) accounts. This peculiar trend is being widely interpreted as a cryptic nod to the new model, with fans and analysts alike reading between the lines of the fruit-based digital breadcrumbs.

Further fueling the speculation is the model's name itself. The "nano" prefix is highly consistent with Google's own naming conventions for its lightweight and on-device models, such as the Gemini Nano. This has led many to believe that nano-banana is not a standalone product but rather the first public glimpse of a hyper-realistic, on-device image generation model from Google.

But Google isn't saying anything. No official announcements, no press releases, no confirmations. Just banana emojis from executives and a model that's quietly dominating the leaderboards. It's like watching a tech thriller unfold in real-time, where the only clues are emoji breadcrumbs and exceptional AI performance.

For now, the mystery remains, but the consensus is clear: a new, powerful force has entered the arena, and its impact is already being felt across the industry. Whether Google confirms its ownership or not, the nano-banana model is setting a new benchmark for what's possible in AI image generation while keeping everyone guessing who's pulling the strings behind the curtain.

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 👋