The Messy Creative Stage: Why Business Students Should Start Building Now (And Why Perfect Plans Are the Enemy)

Three industry legends just revealed why this chaotic moment in AI is actually the perfect time for students and new grads to start building—even if their ideas aren't fully baked yet.

The Question Every Business Student Is Asking

“Should I learn to code or double down on business strategy?”
“Will AI create jobs or replace them?”
“Is it too late to build something new?”

If you're a business student, recent grad, or just starting your career, you’ve probably asked at least one of these. And the advice out there? Confusing at best.

But after watching three standout talks from Y Combinator’s AI Startup School — featuring Andrew Ng (AI Fund), Kirsten Green (Forerunner Ventures), and Aravind Srinivas (Perplexity AI) — one thing became clear:

👉 This uncertain, chaotic moment in AI isn’t a reason to wait — it’s a rare window to start.

Kirsten calls it the "messy creative stage." Andrew and Aravind agree: the biggest opportunities belong to those who build before the dust settles.

The Andrew Ng Revelation: Concrete Ideas Beat Smart-Sounding Ones

Andrew Ng, the Stanford AI legend who helped build Google Brain and founded Coursera, dropped a truth bomb that should reshape how every business student thinks about innovation:

"Vague ideas tend to get a lot of kudos. If you tell your friends 'we should use AI to optimize healthcare assets,' everyone will say that's a great idea. But it's actually not a great idea—at least not in the sense of being something you can build."

This hit me hard because it describes exactly how most business school discussions about AI go. We love big, visionary statements that sound intelligent but provide zero direction for actual execution.

Ng's alternative? Concrete ideas.

Instead of "AI for healthcare," try "software to let patients book MRI machine slots online to optimize usage." Instead of "AI for email productivity," try "Gmail integration that automatically filters routine emails into specific folders."

The difference isn't just semantic—it's operational. A concrete idea gives clear direction. A team can run fast to build it, validate it, or discover it doesn't work. Either outcome provides valuable learning.

Why This Matters for Your Career

The Speed Advantage: Ng revealed that AI coding assistants have made building prototypes not just 50% faster, but 10 times faster. This isn't a gradual improvement—it's a fundamental shift in what's possible.

The New Bottleneck: "The real bottleneck isn't technology anymore," Ng explained. "It's deciding what to build and getting user feedback quickly."

This creates a massive opportunity for students who can think concretely and move fast. While others debate the theoretical implications of AGI, you can build and test real solutions to specific problems.

The Surprising Career Advice: Ng's most controversial suggestion? Everyone should learn to code—including marketers, finance professionals, and operations people.

"My CFO codes. My head of talent codes. My recruiters code. All of them perform better at their job functions because they can code," he shared.

This isn't about becoming a software engineer. It's about becoming fluent in the language of the future—the ability to tell computers exactly what you want them to do.

The Kirsten Green Framework: From Attention to Affection

Kirsten Green, founder of Forerunner Ventures and the investor behind iconic brands like Warby Parker, Chime, and Dollar Shave Club, offered a completely different lens for understanding the AI opportunity.

Her insight: "For the last two cycles—the internet and mobile—we moved from outcomes and attention into an area of relationships and affection."

Think about the difference between googling "best restaurants in SF" versus having a conversation with someone who knows your dietary preferences, your budget, your past dining experiences, and your current mood. The first is a transaction. The second is a relationship.

The Memory Revolution

Green highlighted something that's easy to miss in all the AI hype: the ability to build context over time.

"The real crazy unlocks come from the ability to take memory that builds over time and intuit it forward," she explained. "You and I sit down today, have this conversation. I go have conversations with three other people about a similar topic. We come back six months later, and the AI can bring all of those things forward and draw conclusions from that body of work."

This isn't just better search—it's a fundamentally different relationship with technology.

Voice Changes Everything

Green also emphasized how voice interfaces unlock richer data: "As soon as you get to have voice, people take a lot more liberty with how they talk. You almost let ideas unfold and peel back the onion. Now we've got voice that talks back to us digitally to prompt conversation."

For business students, this suggests huge opportunities in any field where relationships matter—healthcare, education, finance, career development, even B2B sales.

The Brand and Distribution Lesson

Green's experience backing consumer brands revealed another crucial insight: "You can't market products that are bad."

She shared a startling statistic from a major CPG company: Five years ago, 4% of their sales came from organic word-of-mouth. Today? 48%.

"People are savvy and see through inauthentic marketing," she noted. "The first tactic I know everybody likes to go to is product marketing—what about your product gets better because you bring other people into the experience?"

Translation for students: Stop looking for growth hacks. Focus on building something so genuinely useful that people can't help but recommend it.

The Aravind Srinivas Mindset: Competing Against Giants

Perhaps the most inspiring perspective came from Aravind Srinivas, founder and CEO of Perplexity AI, who's building a search engine while competing directly with Google.

His approach to handling competition? "I read all the Twitter comments every time. Google IO last year was 'AI overview and Perplexity is dead.' This year was 'AI mode and Perplexity is dead.' I read all of that too, and it's always fun. I love it actually."

This isn't just bravado—it's a strategic mindset that every aspiring entrepreneur should adopt.

The Application Layer Opportunity

Srinivas offered a crucial insight about where the real opportunities lie: "Almost by definition, the biggest opportunities have to be at the application layer because we actually need the applications to generate revenue so they can afford to pay the foundation cloud and semiconductor technology layers."

While media focuses on foundation models and infrastructure, the real money is in solving specific user problems at the application level.

The Speed-Only Strategy

When asked how small startups can compete with tech giants, Srinivas was blunt: "The only mode you have is speed. You have to innovate. You have to move faster than everybody else. It's like running a marathon but at an extremely high velocity."

But here's the counterintuitive part—this speed advantage is more sustainable than it appears. Large companies have structural disadvantages:

• The innovator's dilemma: Google can't cannibalize their advertising business to build better search

• Risk aversion: "For Google, one mistake tanks their stock. For us, we can make a lot of mistakes and it's fine"

• Model access: "As a startup outside Google, you had access to AI that was better than what Google internally had, which was unprecedented"

The Browser Vision

Srinivas shared Perplexity's next big bet: building a browser that functions as a "cognitive operating system" where you can launch multiple AI tasks in parallel.

"You go and do research on real estate, the markets, and these are all just processes running on your browser. That's never been possible before," he explained.

This vision points to a broader opportunity: reimagining fundamental digital experiences rather than just adding AI features to existing products.

The Convergent Insight: Why This Moment Is Special

Despite their different backgrounds and focus areas, all three speakers converged on several key insights:

1. The Application Layer Is Where Students Should Focus

While everyone debates foundation models, the real opportunities are in building specific solutions for real users. This is perfect for students who might not have the resources to train large models but can identify concrete problems worth solving.

2. Speed Trumps Perfection

The ability to build and test quickly is more valuable than having a perfect plan. This favors scrappy students over established players with complex approval processes.

3. Concrete Beats Clever

The ideas that succeed won't be the ones that sound impressive in pitch decks—they'll be the ones that solve specific problems for specific users in measurable ways.

4. The "Messy Creative Stage" Won't Last Forever

As Green noted, "There's something to be said for being first, being early. You have more of a chance of surprising and delighting somebody. Once some version of that has been demonstrated in a dozen ways, it loses some of its luster."

The window for novelty advantage is open—but closing.

What This Means for Your Career Strategy

If you're a business student or recent grad, here's how to think about positioning yourself in this AI-transformed landscape:

1. Develop AI Fluency (Not Just Business Fundamentals)

Traditional business education focuses on analyzing existing markets and optimizing known processes. But the AI revolution requires a different skill set:

• Understanding what AI can and cannot do accurately

• Knowing when to use different AI approaches (prompting vs. fine-tuning vs. agentic workflows)

• Making technical architecture decisions that can save months of work

• Communicating complex intent to AI systems effectively

As Ng noted, "There are a lot of things that humans can do that AI cannot, but people that know how to use AI to get computers to do what you want will be much more powerful than people that don't."

2. Choose Your Domain Expertise Carefully

All three speakers emphasized the importance of deep subject matter expertise. But choose wisely:

• Look for industries ripe for emotional relationships (healthcare, education, finance)

• Focus on sectors with clear pain points that AI could address

• Consider areas where you already have insights from personal experience

The goal isn't to become a generalist AI consultant—it's to become the person who understands both AI capabilities and a specific domain deeply enough to spot concrete opportunities others miss.

3. Build Your "Prototype Portfolio"

With AI coding assistants making development 10x faster, you can afford to test many more ideas. Start building a portfolio of quick experiments:

• Build 20 prototypes to see what works

• Focus on "quick and dirty" solutions initially

• Test with real users immediately

• Be prepared to completely rebuild based on feedback

As Ng explained, "The cost of a proof of concept is low enough that it's actually fine if lots of proof of concepts don't see the light of day."

4. Master the Art of Concrete Thinking

Practice turning vague ideas into specific, buildable concepts:

• Bad: "AI for student productivity"
  
  

• Good: "Slack bot that automatically schedules study groups based on calendar availability and course enrollment"
  
  

• Bad: "AI for healthcare"
  
  

• Good: "App that reminds patients to take medications and tracks side effects for doctor visits"
  
  

The more concrete you can be, the faster you can move from idea to validation.

5. Embrace the "Move Fast and Be Responsible" Philosophy

Ng's updated version of "move fast and break things" is perfect for students: "Move fast and be responsible."

This means:

• Quick iterations with real user feedback

• Ethical considerations built into your process from day one

• Responsible experimentation that doesn't cause harm

• Transparency about limitations and potential risks

The Uncomfortable Truth About Timing

Here's what nobody wants to tell business students: This window won't stay open long.

Green's "messy creative stage" is temporary. As AI capabilities become more understood and best practices emerge, the advantage will shift from scrappy experimenters to well-funded incumbents.

Srinivas's speed advantage at Perplexity works because AI is still new enough that small teams can outmaneuver large organizations. But this won't last forever.

Ng's emphasis on concrete ideas matters because the obvious applications will get built quickly, leaving only the more nuanced opportunities for later entrants.

The implication: If you're going to build something, start now. Don't wait for the perfect idea, the perfect team, or the perfect market timing.

Your Next Steps: From Analysis to Action

If this resonates with you, here's your concrete action plan:

Week 1: Learn the Fundamentals

• Take Andrew Ng's AI courses on Coursera or DeepLearning.AI

• Experiment with AI coding assistants (Cursor, Claude, GitHub Copilot)

• Pick one domain where you already have knowledge or interest

Week 2: Find Your Concrete Idea

• List 10 specific problems you've personally experienced in your chosen domain

• For each problem, write a one-sentence description of a concrete solution

• Pick the one you could test with 3 real users this week

Week 3: Build and Test

• Create a quick prototype (remember: 10x faster with AI tools)

• Test with real users immediately

• Document what you learn, even if the idea doesn't work

Week 4: Iterate or Pivot

• If users love it, iterate based on feedback

• If users don't care, pick a different concrete idea and repeat

• Either way, you're now ahead of students still debating whether to start

Ongoing: Build Your Network

• Share your experiments publicly (LinkedIn, Twitter, personal blog)

• Connect with other builders in your space

• Attend AI meetups and startup events in your area

The Meta-Lesson: Bias Toward Action in Uncertain Times

The deeper insight from all three speakers isn't really about AI—it's about how to make decisions when the future is uncertain.

Traditional business education teaches you to analyze markets, develop comprehensive strategies, and minimize risks. These skills remain valuable, but they're insufficient in rapidly changing environments.

The new skill is intelligent experimentation—the ability to form hypotheses quickly, test them cheaply, and iterate based on real-world feedback.

This applies whether you're building AI products, choosing a career path, or deciding which skills to develop. In uncertain times, the biggest risk isn't making the wrong choice—it's making no choice at all.

Conclusion: Your Competitive Advantage Is Starting Now

While your peers debate whether AI will steal their jobs, you can be building the solutions that create new opportunities. While others wait for the market to stabilize, you can be experimenting with what's possible in the messy creative stage.

The three insights from Y Combinator's AI Startup School point to the same conclusion: This moment favors action over analysis, concrete solutions over clever theories, and speed over perfection.

As Kirsten Green noted, "We're in that time period where we have new technology to play with, and you just ask: what will the next iteration of experiences look like? The only way to really get to that big idea is to try a bunch of different things."

Your advantage as a student isn't that you have all the answers—it's that you can afford to try many different approaches until you find what works.

The question isn't whether you're ready to start building.

The question is: What will you test with real users this week?

Ready to start building? Share your concrete idea in the comments below—you might find your first user, your next collaborator, or just the encouragement you need to take that first step.

Image credit: Custom watercolor-style illustration created with the help of ChatGPT (OpenAI).