The AI builder's edge

This week I experimented with building agents on a couple of platforms for a prototype we're investigating at Flintolabs. It quickly became clear that agents are often overkill; simple automation would accomplish most of what we needed.

It made me wonder: has AI made some solutions unnecessarily complex?

I came across an article in New York Magazine, "What Is College For in the Age of AI?" that highlighted something striking: Gen-Z graduates are getting hit hardest by AI. Employment in AI-affected jobs dropped 16% since 2022 for recent grads, while senior workers remain unaffected. The reason? Senior workers have "tacit knowledge" and practical judgment that only comes from real experience.

The experience Catch-22 just got worse. Employers question the ROI of training junior talent, and most internships don't develop real-world skills. Only a handful of schools integrate actual workplace experience into curriculum.

Here's what I’ve noticed with our students: the most successful projects aren't the ones with sophisticated agent architectures. They're the ones that solve a real problem with just enough technology to make it work.

They start with "how do I fix this?" rather than "how do I use this cool new capability?" That problem-first instinct and the practical judgment that comes from actually building things might be exactly what the job market is missing.

-Janani

Applications Open Now

What: An entrepreneurship pitch competition designed exclusively for high school students, where you present innovative business ideas and compete for resources to launch your venture.

Who: All Colorado high school students with entrepreneurial ideas and passion for solving problems.

Format: Teams develop business concepts, create pitch decks, and present to panels of experienced entrepreneurs, investors, and business leaders. Finalists present at CU Boulder's campus.

What Makes It Special: Direct mentorship from successful entrepreneurs, access to CU Boulder's entrepreneurship resources, and the chance to turn your idea into reality with seed funding and support.

Perfect for: Students who see problems others don't and want to build solutions, especially those developing AI-powered applications or tech ventures.

Flintolabs Connection: We're proud to be featured as a resource for the High School New Venture Challenge, helping students develop the technical skills to build their entrepreneurial visions.

The World Economic Forum in Davos delivered a clear message from tech's biggest names: AI's hype phase is over, and the hard work begins now. Microsoft CEO Satya Nadella warned that AI risks becoming an economic bubble if benefits stay trapped in tech companies rather than transforming actual work. The telltale sign of a bubble? "If all we're talking about are the tech firms," Nadella said, emphasizing that real value comes from companies using AI to drive their own productivity.

Nvidia's Jensen Huang reframed the conversation entirely, declaring AI is now infrastructure like electricity or roads. The constraint isn't better algorithms - it's power, compute capacity, and skilled workers who can deploy it. Meanwhile, Google DeepMind's Demis Hassabis cautioned that parts of the investment boom are disconnected from real deployment, with "multibillion-dollar seed rounds in new startups that don't have a product" looking unsustainable.

For high school students learning AI now, the message is electric: the winners won't be those who just use AI tools, but those who can actually build, deploy, and create value with them. As Microsoft's hiring data shows, technical skills matter more than credentials and the gap between AI consumers and AI builders is becoming the defining advantage.

Read the full analysis: What AI Leaders Were Saying at Davos 2026

For years, Gary Marcus was tech's most dismissed critic. OpenAI created an emoji mocking him. Sam Altman called him a troll. But now, as GPT-5 disappoints and scaling hits diminishing returns, Silicon Valley is quietly admitting: Marcus was right.

His core argument? Large language models are glorified memorization machines that break things into little bits and reassemble them statistically—sometimes incorrectly. They hallucinate because they don't actually understand; they can't build "world models" that represent how things actually work. When ChatGPT confidently states that actor Harry Shearer was born in London (he's from Los Angeles), it's not a random error, it's the fundamental limitation of systems that predict words rather than comprehend meaning.

The turning point came in August 2025 when GPT-5 launched. Sam Altman had implied it would approach artificial general intelligence. Instead, users found the same hallucinations and reasoning failures within hours. For students building with AI, Marcus's critique matters: these tools are powerful for pattern recognition and statistical tasks, but they're not magic. Understanding their limitations, and learning to combine them with classical programming approaches, is what separates AI users from AI builders.

Read our deep dive: Understanding AI's Limitations: Why It Matters for Student Builders

This week, we're highlighting Sanaya Verma, a high schooler from Seattle who turned a complex physics concept into an engaging game simulation after just 2 weeks in our Foundations program.

Sanaya was fascinated by space exploration but noticed something: escape velocity is taught as an abstract formula in textbooks, making it hard to grasp why different planets need different speeds to break free from gravity. What if you could actually experience the concept by launching rockets and seeing what happens?

So she built a rocket launch simulator where users choose the number of engines for their spacecraft and attempt to achieve escape velocity from different planets. Too few engines? Your rocket falls back to the surface. Just right? You break free into space. The simulation demonstrates escape velocity across multiple planets in our solar system, making an intimidating physics equation into an intuitive, hands-on experience.

This is exactly what happens when students have the skills to build: they see solutions that textbooks miss.

This type of early achievement exemplifies what's possible when students learn to build rather than just consume AI technology.

Use code NEWYEAR50 - offer expires January 31st!
February cohort starts February 7, 2026

Only days left to start 2026 with real AI skills at a discounted rate!

Here's what you get with just one hour per week for 6 months:

✅ Hands-on AI skills through building real applications - not watching lectures
✅ 3 transferable college credits from University of Colorado Denver
✅ Portfolio of real work that demonstrates capability to colleges and employers
✅ Small class sizes (capped at 20 students) ensuring personalized attention
✅ Advanced concepts like OpenCV, Minimax algorithms, computer vision, and more
✅ The critical thinking and problem-solving skills employers desperately need

While college grads struggle with workplace readiness and entry-level jobs disappear to automation, Flintolabs students are building portfolios of real work that demonstrate genuine capability.

Our students aren't just learning about AI - they're building games, creating tools, solving real problems, and developing the judgment that comes from hands-on creation.

Enroll now with code NEWYEAR50 →

Our program has a 5-star rating with reviews from both students and parents.
Questions? Email us at [email protected]

View all reviews on Google

Found this valuable? Forward this newsletter to other high schoolers and parents who want to be informed about AI trends and what is needed to prepare for an AI-driven future. Every student deserves the chance to build real skills before college.

You're receiving this newsletter because you expressed interest in Flintolabs or crossed paths with our community. If you believe you received this in error, please feel free to unsubscribe using the link in this mail.