Hi! Welcome to the 27th 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.

_{~9 mins read}

/OpenAI /Benchmark /Enterprise

Early December was not a good month for Sam Altman. Google's Gemini 3 had just topped the AI leaderboards, ChatGPT traffic was declining, and whispers spread that OpenAI was losing its edge. The response? A dramatic internal "code red" memo that marshalled resources toward one mission: fix ChatGPT, and fix it fast.

Less than two weeks later, GPT-5.2 landed - the fastest major model iteration OpenAI has ever shipped. Where GPT-5.1 took three months to follow GPT-5, competitive pressure compressed the timeline to under a month. The urgency was palpable, and the stakes were enormous for a company with a $500 billion valuation and over $1.4 trillion in planned infrastructure spending.

So, did they deliver? According to OpenAI's benchmarks, yes. GPT-5.2 Thinking mode edges out both Gemini 3 and Claude Opus 4.5 across nearly every reasoning test - from real-world software engineering tasks (SWE-Bench Pro at 55.6%) to doctoral-level science knowledge (GPQA Diamond) and the notoriously difficult ARC-AGI abstract reasoning benchmark.

But here's what really matters for everyday users: GPT-5.2 produces 38% fewer errors than its predecessor and costs significantly less to run. On OpenAI's GDPval benchmark (measuring professional knowledge work across 44 occupations), it beats or ties top industry professionals 70.9% of the time - at 11 times the speed and less than 1% of the cost.

The model comes in three flavours: Instant for quick queries, Thinking for complex structured work like coding and analysis, and Pro for maximum accuracy on difficult problems. All are available now to paid ChatGPT users and via API.

The bottom line: The AI arms race is accelerating. Where model leads once lasted quarters, they now last weeks. For businesses, this means frontier AI capabilities are becoming both more powerful and more affordable - making this the perfect time to experiment with AI in your workflows.

/Google /Browser /Productivity

If you're anything like us, your browser currently has approximately 47 tabs open across three windows, and you've genuinely forgotten what half of them are for. Google's latest experiment, Disco, thinks it has the answer to our collective tab chaos.

Disco is Google Labs' newest "discovery vehicle" for reimagining web browsing, and its flagship feature - GenTabs - is genuinely impressive. Using Gemini 3, it analyses your open tabs and chat history to understand what you're trying to accomplish, then generates custom interactive web applications to help you complete those tasks.

Planning a trip to Japan? Instead of juggling tabs for flights, accommodation, itineraries, and cherry blossom forecasts, GenTabs creates a unified travel planning app with maps, calendars, timelines, and crowd-level predictions - all tied back to your original sources. Need a meal plan for the week? It'll generate an interactive planner with recipes, shopping lists, and nutrition info. Helping your kid learn about the solar system? It creates a 3D interactive model they can explore.

The magic is that you never write a line of code. Just describe what you need in natural language, and GenTabs builds it. The AI even proactively suggests tools you hadn't thought of based on your browsing patterns.

This feels like Google's strategic counter to AI browsers like Perplexity's Comet and ChatGPT Atlas - but instead of building a separate browser, they're experimenting with fundamentally rethinking what a browser can do.

The catch? It's currently waitlist-only, macOS-exclusive, and limited to US users with personal Google accounts. Classic Google Labs move - test small, learn fast, maybe integrate into Chrome later. The bigger picture: We're watching AI labs use these experiments to see what actually sticks with users. Not every Google Labs project survives (RIP to the dozens that didn't), but the ones that do often reshape entire product categories.

/Coding /Design /UI

For years, the handoff between designers and developers has been... let's call it "fraught." Designers live in Figma's world of pixels and frames. Developers translate that into code. Miscommunication ensues. Products ship looking slightly off. Teams blame each other over Slack.

Cursor's new Visual Editor aims to end this cycle by merging design and code into a single interface powered by AI.

Here's how it works: Inside Cursor's browser environment, you can now drag and drop elements, inspect components and props, adjust styles with sliders and colour pickers, and describe changes in plain English - all while the AI writes the actual CSS and React code in real time. It's like having Chrome DevTools, Figma, and an AI coding assistant in one unified workspace.

The killer feature? You can point at any element and say "make this bigger," "turn this red," or "swap their order," and multiple AI agents work in parallel to execute your changes within seconds. Every control maps directly to real CSS, so you're not working with symbolic approximations - you're manipulating the actual code that ships to users.

Cursor's head of design, Ryo Lu, told WIRED: "Previously, designers lived in their own world of pixels and frames that couldn't be converted into code. We combined the world of design and the world of programming into one interface with one AI agent."

This positions Cursor to compete not just with other code editors, but potentially with Adobe and Figma in the design space - though Cursor argues the market is large enough for different approaches.

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: Master Google's Nano Banana Pro and create professional-grade visuals 🍌

Remember when AI-generated images looked obviously fake? Those days are over. Google's Nano Banana Pro (the nickname for Gemini 3 Pro Image) is sitting at the top of image generation benchmarks, and for good reason - it's scary good at creating professional-quality visuals that most people can't distinguish from real photos or human-created graphics.

The team at Google AI Studio just dropped a comprehensive guide with 10 advanced techniques for getting the best results. Instead of using generic "tag soup" prompts (you know, "dog, park, 4k, realistic"), you learn to prompt like a creative director.

Here are three powerful use cases we're excited about:

1. Turn Data into Visual Infographics

Nano Banana Pro excels at transforming dense information into clean, professional graphics. Upload a PDF of your company's quarterly earnings, and prompt:

The AI doesn't just make it pretty - it actually reads, analyses, and compresses the information into scannable visual formats. Perfect for presentations, reports, or social media.

2. Solve Equations on a Whiteboard (Yes, Really)

This one blew our minds. Nano Banana Pro can solve complex mathematical problems and show its working on a visualised whiteboard. Try this:

It generates an actual image of a whiteboard with the equation solved step-by-step, using colour-coding to make it easy to follow. Students and educators, this one's for you.

3. Create Viral-Worthy Character Content

Want to create a series of images with consistent characters? Upload reference photos and prompt:

The result? Professional-grade sequential storytelling that maintains character consistency - something that was nearly impossible with earlier AI image models.

Pro tips from the guide:

Pricing in SA context: Nano Banana Pro is available through Google AI Studio with generous free limits for testing. Production use via API costs approximately R0.50 per image (based on current exchange rates) - significantly cheaper than hiring a designer for one-off graphics.

Try it in Google AI Studio →

You spoke, we listened. “AI in Enterprise” is here to stay. In this section, we're spotlighting real businesses using AI to solve actual problems.

OpenAI's State of Enterprise AI Report: What 9,000 Workers Tell Us About AI at Work

This week, OpenAI released its first comprehensive State of Enterprise AI report, and the findings reveal something striking: we're past the experimentation phase. Enterprise AI adoption is accelerating in both breadth (more workers using AI) and depth (existing users going deeper into sophisticated use cases).

The report draws on real-world usage data from OpenAI's enterprise customers plus a survey of 9,000 workers across nearly 100 enterprises. Here's what caught our attention:

Adoption is exploding: Over the past year, weekly messages in ChatGPT Enterprise increased roughly 8 times, and the average worker is sending 30% more messages. But it's not just volume - usage of structured workflows like Projects and Custom GPTs has increased 19 times year-to-date, showing a shift from casual querying to integrated, repeatable business processes.

Workers report real value: 75% of workers say AI has improved either the speed or quality of their output. Time savings are significant - workers report saving 40-60 minutes per day, with heavy users saving over 10 hours per week. Crucially, AI isn't just helping people do the same work faster - 75% of users report being able to complete new tasks they previously couldn't perform.

The gap is widening: Frontier workers (95th percentile) send 6 times more messages than median employees and engage more intensively across advanced capabilities. Frontier firms send 2 times more messages per seat and show deeper integration across teams. This suggests that organisations moving slowly risk falling significantly behind more aggressive adopters.

Industry-specific insights: Technology, healthcare, and manufacturing are the fastest-growing sectors. Professional services, finance, and technology operate at the largest scale. Internationally, Australia, Brazil, the Netherlands, and France are seeing the strongest growth, each exceeding 140% year-over-year.

The bottom line: The primary constraint for organisations is no longer model performance or tooling availability - it's organisational readiness and implementation strategy. Companies that master deployment see compounding benefits as usage deepens and spreads across teams.

For South African businesses watching this trend, the message is clear: AI adoption in enterprise is not a future consideration - it's happening now, at scale, across every industry. The question isn't whether to adopt, but how quickly you can implement effectively.

Read the full report →

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 👋