Today’s Highlights

• AI: More Organised ChatGPT: The “Projects” Feature

• Design: Adobe’s 6 Types of Design Leaders

• Science & Tech: A Hyperloop Tunnel From New York To London: A 1-Hour Journey

• Founding: Hire Folks With “Chips On Their Shoulders”: Perplexity CEO's Advice to Startups 

• Product: 4 Tiers Of Working With Data: Business Models To AI-Supported Data Manoeuvres

• Today’s AI image: “Just a Tunnel Away”: Transatlantic Tunnel

• Quote for the day: From Babe Ruth

AI

More Organised ChatGPT: The “Projects” Feature

ChatGPT has just launched Projects, an excellent way to organise your chats, uploaded files, and custom instructions in one spot. It’s a great tool for keeping your ongoing work in check, making it easier to pick up where you left off.

Projects support web searches and Canvas.

Availability

Currently, Projects is only available on the web version and Windows desktop app of ChatGPT. If you’re a ChatGPT Plus, Pro, or Team user, you’re in luck. Everyone else (Enterprise, Edu, and Free users) will have to hang tight.

Demos

You can now store all sorts of handy docs (like fridge manuals or garage instructions) in one project for easy access.

Or keep all your personal website details—like code documentation and author info—neatly organised in a project for smooth iteration.

Why is this a big deal?

People have been asking for this feature forever. No more hunting down that prompt or context from past chats. Plus, creating these “mini-systems” for yourself is a brilliant way to get the most out of AI—whether it’s ChatGPT or Claude.

Read More

Also in AI

• Microsoft Launches Phi-4: A 14B parameter model that rivals larger models like GPT-4o and Gemini Pro 1.5, excelling in mathematical reasoning

• Pika Labs 2.0: AI video upgrade including enhanced realism and customisation for AI-generated videos

• xAI released Grok-2 with faster speeds and better multilingual support

• Meta FAIR: New AI projects include Meta Motivo (agent control) and Video Seal (watermarking)

• OpenAI: Shared a timeline of Elon Musk's involvement and ongoing lawsuits

• Google launched NotebookLM Plus with interactive audio features

• DeepSeek released VL2, a compact vision-language model

Design

Adobe’s 6 Types of Design Leaders

Designers at Adobe can pursue two paths: management or Individual Contributor (IC).

The Principal Designer is the highest role on the IC ladder, focusing on leadership without managerial duties.

Adobe has identified six key archetypes that represent the skills needed for success at the principal level.

These archetypes highlight the varied leadership styles and strengths of principal designers. Each designer brings a unique blend of these archetypes to shape their leadership approach.

• Visionary: Foresees industry trends and innovates accordingly.

• Evangelist: Champions new ideas, ensuring they gain visibility within and outside the company.

• Researcher: Relies on data and research to inform design decisions.

• Innovator: Identifies market gaps and creates solutions to drive business results.

• Strategist: Focuses on strategic thinking, guiding product launches and aligning teams.

• Subject Matter Expert: Becomes the go-to authority in a specific area, such as video or emerging tech.

Navigating Your Own Path:

• Find Your Specialisation: Principal designers are known for their expertise in specific areas. Discover your passion within design and focus on honing that skill.

• Build Visibility: Gain exposure within your team and company, ensuring your contributions are recognised.

• Collaborate Across Boundaries: Work with cross-functional teams to extend your influence beyond your immediate group.

• Measure Your Impact: Demonstrate how your work affects the business, especially as you move up the career ladder.

By mastering these archetypes, you’ll carve out your own path to success in the design world.

Read More

Also in Design

• Craft Smooth Flowing Multistep Forms With An Eased User Experience

• Upwork’s New Visual Identity: Fluid grid system, dynamic visuals, and global adaptability

• Apple’s Magic Mouse Redesign: Finally addressing the bottom charging port issue

• Figma’s New Pricing Model: Starting March 2025, Figma will shift to seat-based pricing, bundling features like Slides and FigJam

• Prototype Your Product Ideas with AI: AI-native editor to iterate designs and export to React or Figma

• Garanti BBVA Securities Case Study: Redesigned eTrader platform refurnishes UX for diverse investors with tailored features and seamless design

• Google Maps Redesign: Updates include dark teal accents, black labels for better readability, and a simplified floating action button 

Science & Tech

A Hyperloop Tunnel From New York To London: A 1-Hour Journey

The concept of a Transatlantic Tunnel connecting New York and London has resurfaced, but the proposed $20 trillion cost makes it a highly ambitious project.

Despite this, advances in vacuum tube technology have made the idea more feasible, potentially allowing trains to travel at speeds exceeding 3,000 mph in a vacuum, reducing the travel time between the two cities to just under an hour.

All You Need To Know

• The tunnel would span over 3,000 miles, making it an immense engineering challenge.

• Costs are estimated at $19.8 trillion, comparable to the construction of massive infrastructure projects like the Channel Tunnel.

• New technologies, such as hyperloop systems that reduce air resistance, could make the project more viable by allowing incredibly high-speed travel.

• Several design proposals exist, including tunnels built on stilts or floating structures, but no definitive design has been agreed upon.

• The project would significantly reduce the time it takes to travel between the two cities, making it a more sustainable alternative to flying.

While the tunnel's construction would take years and face numerous challenges, the potential for reduced travel time and environmental benefits is a major draw.

Critics argue that the cost and technical difficulties make it an impractical venture at this stage.

The mere fact that we're even discussing the possibility of constructing something once only imagined in sci-fi shows just how far we've come.

Read More

Also in Science & Tech

• Google’s Gemini AI Glasses: Real-time AI assistant, language translations, and directions

• Scientists warn about creating mirror-image microbes, which could resist natural enzymes and pose pandemic risks

• Apple’s Foldable iPad: Apple plans to release a foldable 18.8-inch iPad by 2028, with a creaseless design

• Genetic analysis reveals Neanderthals and humans interbred for 7,000 years starting ~50,000 years ago

• Claude: Favoured by tech insiders for its human-like responses and emotional intelligence

Founding

Hire Folks With “Chips On Their Shoulders”: Perplexity CEO's Advice to Startups

An AI-powered search engine with impressive versatility and state-of-the-art features to offer in every sector rose itself to a valuation of $9 billion in 2024. 

Perplexity AI scaled to a huge size despite challenges thrown at it constantly and managed to get head-to-head with market rulers like Google and OpenAI.

Recently in his talk at Stanford, the CEO Aravind Srinivas lighted the path forward for emerging startups on putting together high-ROI teams and becoming a sage yet sharp-witted leader. Here’s his message summarised into three short lessons: 

#1 Your Team Should Excel In What You Suck At: Your co-founder/s and team members should be better than you in their field of work and expertise. When their skills complement yours, all your resources (mainly human resources) are optimally utilised and with least wastage. 

#2 Get Biased For Action: Avoid prolonged stagnation of operations. This commonly happens when decision-making is complexified and thus delayed. Or when there is an obsession with excessive planning.   

So move. Move speedily and uplift your team to do so, too. 

Besides, prepare for when your organisation grows huge. When it’s difficult to move quickly. 

With a complex org structure, you couldn’t simply go like - “All the old outdated operating systems will be replaced with a modern AI tool” - and expect a 100 employees to instantly understand and adapt. 

#3 Take a Chance by Giving a Chance: Provide opportunities for people to prove themselves in new areas rather than always hiring established experts.

Promote a culture of learning by letting employees "figure out how to swim" in uncharted waters.

Read more

Also in Founding

• What Makes a Great Business: Goes Beyond Mere Transactions

• Software Stocks Outlook: Despite flat earnings, software stocks saw growth in 2024, driven by stabilising macro conditions and AI potential

• Onboarding User Profiling: Gather first-party data during onboarding in three steps

• Managing Teams Effectively: Focus on defining clear outcomes rather than micromanaging methods

Product

4 Tiers Of Working With Data: Business Models To AI-Supported Data Manoeuvres 

Data is ultimately just numbers. 

Numbers that often trick human heads. Numbers that some exploit to meet the exact same purpose - tricking human brains. 

Some orgs, though, recruit these numbers and graphs as their compasses and guides for making clever moves along the way. Here’s how they (and any organisation should) climb up the ladder of data proficiency-

Level 1: Business Models

Many unfamiliar startups often over-rely on numerous business metrics (like revenue and costs) without focusing on user experience or product health. 

They end up chasing random metrics without strategic alignment.

Solution? Start with models, not metrics. Build frameworks that clarify cause-and-effect relationships and identify key metrics. Like flywheels, metrics trees and value exchange loops. 

Level 2: Data-Led Decisions

Data is central to decision-making. But more importantly, it should complement human judgement. 

What to do?

Invest in Data Infrastructure. Build pipelines, warehouses, and integrate qualitative research. Start small and expand over time.

Use dashboards to monitor key metrics.

Hold weekly metrics review meetings for collective learning and decision-making.

Level 3: Evidence-Guided Development

Data becomes evidence when it confirms or refutes assumptions, enabling better decisions. 

How to make meaningful use of evidence? 

Validation: Test assumptions with techniques like interviews, surveys, and experiments.

Evaluation: Analyse impact and costs based on evidence to decide on ideas to scale, pivot, or park.

Level 4: AI Deals With the Numeric Mess Firsthand

Let us all admit it - data is a tedious business. Especially analysing it manually number by number turns out to be mundanely repetitive and yet, ineffective. Often dulls our creative and critical nerves. 

Well, now we’ve got AI at our hands that could take up the technical procedures of splitting insightful threads out of yarns of data. Here’s some use cases- 

• Business Modelling: Identify growth drivers and estimate impacts.

• Data Infrastructure: Clean, format, and detect anomalies.

• Data Analysis: Generate insights, visualisations, and summaries.

• Product Discovery: Automate prototype design, analyse experiments, and predict outcomes.

Building data proficiency is an iterative process. Start with models, integrate data into decision-making, and aim for evidence-guided development.

Yes, AI holds promise for reducing friction and improving data use, but human oversight remains critical.

Read more

Today’s AI Image

“Just a Tunnel Away”: Transatlantic Tunnel

Quote of the Day

From Babe Ruth

"Never let the fear of striking out keep you from playing the game."

Babe Ruth

What we’re working on

🧞Your wish is my command