• Charbel X
  • Posts
  • OpenAI's 'Deep Research' Feature, How to Fix Friction During Big Projects, 3 Strategies to Deal With Nasty Competitors and more

OpenAI's 'Deep Research' Feature, How to Fix Friction During Big Projects, 3 Strategies to Deal With Nasty Competitors and more

A city named "Data Centre". Soon we'll have a 100.

Author

Charbel Zeaiter
February 04, 2025

Good morning!

For anyone on the $200/month ChatGPT plan, research agents have been released. It’s a semi-autonomous model built on a refined o3 model. Also, one person’s UX trend predictions for 2025 offer a sobering reality about the rise of end to end product agents and what it means for UXers (hint: please add a few grains of salt when reading this).

So much in today’s newsletter.

Have a wonderful day 🙏🏽

Yours in Wonder,
Charbel
Founder of Velvet Onion, Faster Zebra and more to come …

Today’s Highlights

  • AI: OpenAI Introduces ‘Deep Research’: Your AI Research Assistant

  • Design: Why Big Ideas Get Stuck (And How to Fix It)

  • Science & Tech: AI Data Centres As Large As a City

  • Founding: Liar, Liar: 3 Strategies for When Competitors Play Dirty

  • Product: Build for Buyers VS Build for Users? Bulls & Bears in Enterprise SaaS

  • Today’s AI image: An Artificially Intelligent New York City

  • Quote for the day: Practically Stoic

AI

OpenAI Introduces ‘Deep Research’: Your AI Research Assistant

OpenAI just rolled out Deep Research, a fresh ChatGPT feature designed to tackle deep-dive investigations.

Instead of spitting out instant answers, this tool combs through multiple sources, analyses text, images, and PDFs, and delivers a polished research report—all in under 30 minutes.

All You Need To Know

  • Runs on a tweaked version of o3, giving it sharper analytical skills for digesting various media types.

  • Currently locked behind a $200/month Pro subscription, with 100 research queries per month. If it proves stable, Plus and Team users may get access soon.

  • Users kick off the process by answering a few clarifying questions, then get notified when the final report is ready.

  • Completion time varies from a few minutes to half an hour, depending on the complexity of the request.

  • Scored a 26.6% on Humanity’s Last Exam, obliterating rivals like Gemini Thinking (6.2%) and GPT-4o (3.3%).

Why Is This a Big Deal?

ChatGPT has always been great for quickfire answers, but Deep Research signals a shift towards more thoughtful, autonomous AI-assisted work.

Instead of a chatbot that merely reacts, OpenAI is nudging AI into the role of an actual researcher—handling tasks that would take humans days, if not weeks.

With this and the recent launch of Operator, AI isn’t just responding anymore. It’s thinking ahead.

Also in AI

Design

Why Big Ideas Get Stuck (And How to Fix It)

Ever feel like your organisation is great at ticking off the little things but somehow, the major strategic projects never quite get off the ground? You’re not alone.

Many CEOs face this challenge, and the culprit is usually a lack of focus on the big rocks.

The Productivity Jar: Rocks, Pebbles, and Sand

Think of your organisation’s capacity as a glass jar, filled with three types of priorities:

  • Sand – The everyday operational tasks: answering emails, fixing bugs, handling customer queries.

  • Pebbles – Small but valuable improvements: process tweaks, minor feature releases, efficiency boosts.

  • Rocks – The major strategic moves: launching a new product, restructuring the business, expanding into new markets.

Most companies fill their jar with sand and pebbles first, leaving no space for the rocks. Sound familiar?

The Fix: Put the Rocks First

To truly move the needle, big goals must come first. That means:

  • Pick your rocks wisely – Identify one to three major strategic goals for the quarter or year.

  • Make them non-negotiable – Ensure your team understands these priorities and doesn’t let them get buried under day-to-day noise.

  • Trust your people – Let them handle the sand and pebbles while you focus on leading with the rocks.

  • Keep the momentum going – Regularly reinforce the importance of these goals so they don’t get sidelined.

When you put the rocks in the jar first, the pebbles and sand naturally fill the gaps—not the other way around.

Why This Matters

Too often, organisations chase endless small wins while the big, transformative work stalls.

By shifting focus to strategic priorities, CEOs can create real progress rather than just busywork.

Also in Design

Science & Tech

AI Data Centres As Large As a City

As AI continues to grow in capability, the need for increasingly massive data centres to train future models has become undeniable.

We're already pushing the limits of current facilities, and experts predict the future requires something much larger—if the industry can even support such expansion.

The State of AI’s “Home” – The Data Centre

  • Where does AI live? Despite the internet's cloud-like feel, AI actually resides in massive data centres. Think large warehouses stacked with computers humming away.

  • Sheer size: These centres can span the size of multiple football fields and consume as much power as small cities.

  • Capacity issues: Hyperscale data centres, essential for AI training, are already falling short of the required computing power.

  • Energy needs: AI’s growing appetite for power demands gigawatt-scale facilities—as much as the energy required for New York City.

The Roadblocks to Building Gigantic Data Centres

  • Physical space: Larger centres need to be closer together for efficient data syncing, which limits options.

  • Financial strain: The projected costs for 1-5 GW centres may stretch tech giants’ capital expenditures.

    Current estimates show building one could eat up most of a company’s annual budget.

  • Power limitations: Securing enough power for such large facilities will require significant energy production boosts and potentially years of preparation.

Why is This a Big Deal?

AI's rapid evolution demands exponential growth in data infrastructure.

These monstrous data centres are no longer just a distant dream—they are the bottleneck holding back the next level of AI.

But overcoming the hurdles of space, power, and cost will take more than just ambition.

Whether we can truly build AI's future is still an open question.

If we don't, AI training might just be relegated to multiple-year projects instead of real-time breakthroughs.

Also in Science & Tech

Founding

Liar, Liar: 3 Strategies for When Competitors Play Dirty

Some marketers argue that product differentiation is futile since competitors can just replicate your features.

While this might hold water in consumer goods, it’s a weak take in B2B tech. Software architecture isn’t a cut-and-paste job—true replication is far trickier than it sounds.

Why Copy Features When You Can Just Say You Did?

  • Competitors often take a “creative” approach—claiming parity without actually delivering it.

  • Sometimes it’s an outright fib; more often, it’s a strategically vague stretch of the truth.

  • The easiest trick in the book? Slap “we do that too” on the marketing page—no dev work required.

3 Tactics to Expose the Lies and Win the Market

1. Shift the Focus from Features to Value

If a competitor calls their mismatched bundle a “platform,” call it what it is—a suite held together with duct tape.

Skip the jargon battle and showcase real benefits: seamless data, better reporting, and effortless automation.

2. Teach Customers to Spot the Difference

When competitors overpromise, shine a light on the real effort needed to achieve results.

Example: A vendor may claim “real-time insights,” but if setup requires heavy lifting that no customer ever implements, it’s an empty promise.

Smart sales teams coach prospects to probe deeper—“Ask them exactly what’s involved in making this work.”

3. Bring the Receipts (and Make Competitors Do the Same)

Customers trust proof over puffery—so provide hard evidence.

Example: A startup claimed to be the “#1 vendor,” despite having a fraction of the market share.

The real leader countered with exact deployment numbers, analyst reports, and customer case studies—forcing the competitor to quietly drop the claim.

Ever had to outmaneuver a dishonest competitor? I’d love to know.

Also in Founding

Product

Build for Buyers VS Build for Users? Bulls & Bears in Enterprise SaaS

Jira’s sluggish interface, bloated features, and overwhelming menus make even simple tasks feel like piloting a 747.

Mastering it requires hours of soul-draining tutorials, and even then, it’s more patchwork than platform.

No One Wants It—But Everyone Uses It

If given a choice, people would pick Trello, a notepad, or a vinegar bath over Jira.

But the choice isn’t theirs—management, who never has to suffer through Jira’s horrors, buys it. The result? The ones who use it hate it; the ones who don’t, force it on everyone else.

Built for Buyers, Not Users

For years, enterprise software was designed to check IT’s compliance boxes, not to make users’ lives easier.

Long feature lists, security certifications, and schmoozy conference dinners mattered more than usability.

If it made employees want to claw their eyes out, well, that was their problem.

The Revolt: Consumerisation of IT

Then, a shift: startups like Asana bypassed IT gatekeepers, selling directly to the frustrated end-users.

People adopted better tools on their own, forcing IT to follow. Good UX finally started winning over bloated enterprise nightmares.

From Utility to Trend

Jira was a product of an era when IT dictated software choices. That’s changing.

The best tools now spread not just because they work well but because they feel right—sleek, intuitive, and, most importantly, cool.

And in a world of abundance, taste is the ultimate differentiator.

Today’s AI Image

An Artificially Intelligent New York City

Quote of the Day

Practically Stoic

"We cannot direct the wind, but we can adjust the sails."

Dolly Parton

What we’re working on

Velvet Onion & Friends

The new Velvet Onion & Friends will be launched soon. It’s our latest evolution, helping companies build products. It’s more than services.

Faster Zebra

February 2025 - the product and venture school journey begins. Whitepaper launching in January.

🧞Your wish is my command

Charbel

Keeping you updated with daily insights on rapid changes in business, tech, design, product, as well as news about science, environment, and wellness. Join the community today!

www.charbelx.com/subscribe