Today’s Highlights

• AI: Google DeepMind’s New Reasoning AI Model To Solve The Most Tangled Problems

• Design: Odyssey's AI Tool: Turning Text and Images into Photorealistic 3D Worlds

• Science & Tech: US to Build First Grid-Scale Nuclear Fusion Power Plant

• Founding: Brittle Points in Your Business: Specific Weak Spots At Which the Company Could Break 

• Product: 3 Growth Engines To Metamorphose Your Product

• Today’s AI image: If Einstein Was Alive…

• Quote for the day: Serendipity

AI

Google DeepMind’s New Reasoning AI Model To Solve The Most Tangled Problems

Google’s new AI model, Gemini 2.0 Flash, takes a moment to “think” through tricky problems, just like OpenAI’s o1, but it’s quicker and free to use.

How It Works

The model displays its thought process when solving problems, similar to reasoning models like OpenAI’s o1, making its logic transparent.

Early users report that it outperforms other reasoning models, offering faster results without compromising quality.

More Thought, More Accuracy

By increasing computation time, Gemini 2.0 Flash delivers longer, more accurate responses, though they might take a bit more time to generate.

Ranked #1 in all categories on the Chatbot Arena, Gemini 2.0 Flash is available for free via AI Studio, Gemini API, and Vertex AI.

Why is this a big deal?

As the competition for AI reasoning heats up, Google’s approach stands out. While OpenAI hikes up prices, Google keeps its top-tier AI free, pushing boundaries without the hefty price tag.

Read More

Also in AI

• ChatGPT Expands App Integrations and Voice Features: Now integrates with coding and writing apps, plus adds voice capabilities

• Perplexity Acquires Carbon: Acquires Carbon to enhance AI file contextualisation

• Genesis: AI Physics Simulator: Introducing the first generative AI physics simulator

• Google Partners with Apptronik: Google collaborates with Apptronik to advance humanoid robot development

• Alec Radford Departs OpenAI: Alec Radford leaves OpenAI to pursue independent research.

• Meta Teases Llama 4: Meta hints at advanced reasoning and speech features in Llama 4

• Microsoft Copilot Vision: Copilot Vision now available to U.S. Copilot Pro users

Design

Odyssey's AI Tool: Turning Text and Images into Photorealistic 3D Worlds

Odyssey, an AI startup founded by ex-self-driving car experts Oliver Cameron and Jeff Hawke, has launched Explorer – a tool that can turn text or images into photorealistic 3D worlds.

How It Works

Much like DeepMind's world models, Explorer can generate interactive scenes from simple captions like “a Japanese garden with lush greenery.”

The tool uses Gaussian splats, a classic rendering technique, to create realistic environments.

Trained on Real Landscapes

The AI behind Explorer has been trained on real-world landscapes, captured with Odyssey's custom 360-degree, backpack-mounted camera. The result? Highly detailed, photorealistic scenes.

Explorer’s generated scenes can be imported into popular creative software like Unreal Engine, Blender, and Adobe After Effects for further editing.

It’s all about giving creators control over the final product.

The Catch

Right now, Explorer’s scenes take about 10 minutes to generate, and while the results are impressive, they’re not without flaws—think low resolution and the occasional visual glitch.

The Big Picture

Odyssey’s ultimate goal? To create worlds that practically build themselves—immersive, realistic, and interactive. If they can make films and games better along the way, that’s a bonus.

Cameron and Hawke’s startup has raised $27 million from investors like EQT Ventures, GV, and Air Street Capital. And with their background in self-driving tech, Odyssey appears set to push the boundaries of AI-powered creation.

Read More

Also in Design

• Meta Integrates AI into Ray-Ban Smart Glasses: Real-time language translation, Shazam song identification, and continuous chatbot conversations

• Leaked iPhone 17 Pro Redesign: Leaked images suggest the iPhone 17 Pro will feature a horizontal camera layout designed for internal engineering reasons, rather than aesthetics

• Product Psychology to Prevent Uber Driver Issues: Strengthens trust, reduce disputes, and improve user experiences

• The Importance of Asking Questions as a Designer: Helps designers clarify needs, uncover issues, and avoid assumptions

• Andrei Herasimchuk, Adobe’s first interface designer and an early Figma contributor, envisions the future of design tools

• Thinking Aloud for Usability: Captures User Thoughts to Find Hidden Issues, but Don’t Expect Instant Clarity

• Apple App Store Award Winners 2024: Seventeen apps and games were honoured for innovation, user experience, and cultural impact

Science & Tech

US to Build First Grid-Scale Nuclear Fusion Power Plant

The US is set to host the world’s first grid-scale nuclear fusion power plant, thanks to Commonwealth Fusion Systems (CFS), a startup aiming to bring this futuristic clean energy to life by the early 2030s.

Big Investment, Big Potential

The facility will be built near Richmond, Virginia, with a multibillion-dollar investment.

Once operational, it will produce 400 megawatts, enough to power 150,000 homes. This marks the first time fusion power will be available on a grid scale.

Fusion: The Holy Grail of Clean Energy

Nuclear fusion, the process that powers stars, fuses atoms (typically hydrogen) to release vast amounts of energy.

Unlike nuclear fission, fusion doesn’t produce harmful waste or emissions, making it a clean, nearly limitless energy source.

Challenges Ahead

While the plant promises a revolutionary energy source, the path to commercial fusion has been notoriously difficult.

The technology has yet to be proven viable, and the industry has long joked that fusion is “just decades away.”

Why Virginia Though?

The James River Industrial Center in Virginia was selected for its growing economy, skilled workforce, and clean energy focus. The plant will also connect to the grid after a nearby coal plant is retired.

While other startups are aiming for fusion in the next decade, experts warn that there’s a big difference between achieving fusion and having a fully operational, safe, and licensed system.

Read More

Also in Science & Tech

• Starlink Launches Satellite Texting in New Zealand

• Hydroxychloroquine Study Retracted: The COVID-19 Treatment Claims That Didn’t Hold Up

• Bile Acid’s Secret: Found to Mimic the Anti-Aging Benefits of Cutting Calories

• Blue Origin Prepares to Test New Glenn Rocket’s Engines: Set to ignite all seven BE-7 engines of the rocket in a final test before its debut flight

• Another Black Hole: The James Webb Space Telescope detects a dormant supermassive black hole dating back to 800 million years after the universe's start

Founding

Brittle Points in Your Business: Specific Weak Spots At Which the Company Could Break 

Brittle points are like that one dodgy plank in a rickety bridge—if it snaps, the whole structure comes tumbling down. These are the single points of failure that can unravel an otherwise healthy company.

Here’s how they tend to manifest:

1. Platform Betrayals: Your product thrives on a platform, but then the platform changes the rules, pulls the plug, or outright copies your idea. Poof—your app’s downloads plummet to zero.

2. Marketing Meltdowns: Your golden marketing channel dries up—Google tweaks its algorithm, costs per click triple, or your go-to event gets cancelled. Suddenly, your growth engine stalls.

3. Customer Overload: One giant customer, paying most of your bills, decides to leave. Sure, they funded your company, but now you’re left scrambling.

4. Key Employee Exodus: That superstar employee—the one who practically is the company—leaves, and replacing them is like finding a needle in a haystack… while blindfolded.

When Good Companies Go Bad

It’s one thing when a business fails because the product stinks or no one wants it. But what about when everything’s humming along—sales climbing, customers happy—and then, out of nowhere, disaster strikes?

Here’s what happens:

• Your app gets buried because Apple bakes your feature into iOS.

• Your SEO traffic tanks after Google’s latest update.

• Your biggest client packs up, leaving you clutching invoices that’ll never be paid.

• Or that one irreplaceable genius on your team decides they’ve had enough of ramen and 70-hour weeks.

It’s not bad luck—it’s brittleness.

Scaling Brittleness: Lessons from HubSpot

Even big companies aren’t immune. In 2016, HubSpot was raking in $260M annually, growing at 60% per year, and had 1,000 employees. Yet, they realised they were a one-trick pony, relying solely on inbound marketing software.

Their solution? Become a multi-product powerhouse. They didn’t just slap together a second product; they ensured it could scale as well as the first. Fast forward to 2024, and HubSpot’s success speaks for itself.

Building a robust company isn’t about eliminating every risk—it’s about managing them thoughtfully.

Like fixing that rickety bridge, you don’t need to replace every plank at once. Just make sure the next one you step on doesn’t send you plunging into the river below.

Read more

Also in Founding

• Predicting Startup Success with Data: Spotting the next big startup isn’t just about data; it’s about knowing which data matters

• AI Breakthroughs: What Just Happened

• Channel-Startup Fit: Putting your “growth” eggs into different “channel” baskets

• Hershey Sells More Than Chocolate – They Sell Happiness

Product

3 Growth Engines To Metamorphose Your Product

1. Growth Loops

The Growth Loops framework is a game-changer for anyone stuck in the linear mindset of traditional funnels.

While funnels focus on moving users through stages—awareness, consideration, conversion—loops create a self-sustaining cycle of growth. Unlike a funnel that needs constant refilling at the top, a loop continuously feeds itself, creating compounding growth.

For instance, Dropbox’s growth loop is a prime example: a new user adds content to the platform, shares it with others, and the recipient becomes a new user, repeating the cycle.

2. Racecar Framework

The Racecar Framework takes the idea of growth loops and builds on it by categorising all growth activities into four parts: the engine (growth loops), lubricants (optimisations), turbo boosts (one-off tactics), and fuel (funnels).

This metaphor helps teams understand where their efforts are going and whether they’re striking the right balance.

For example, a team may spend too much time optimising lubricants—tiny tweaks that reduce friction—without investing enough in the engine itself.

Conversely, over-relying on turbo boosts, like viral campaigns, can lead to short-term spikes but no long-term sustainability.

This framework is especially useful for avoiding these pitfalls and ensuring that every growth initiative contributes to the broader strategy.

3. Adjacent User Theory

Every product eventually reaches a point where its core audience is saturated, and growth slows. This is where the Adjacent User Theory comes in.

The idea is to expand beyond your initial product-market fit by identifying “adjacent users.” These are individuals or groups who are not your primary target audience but are close enough to benefit from your product.

For example, Instagram used this approach to grow from 400 million to 1 billion users by identifying new personas and use cases. This framework helps companies re-accelerate growth by broadening their appeal without losing focus on their core strengths.

Read more

Today’s AI Image

If Einstein Was Alive…

Quote of the Day

Serendipity

"I find that the harder I work, the more luck I seem to have."

Thomas Jefferson-

What we’re working on

🧞Your wish is my command