Shane Wilson's Blog

When you’re in the second year of your MBA program, you go to a lot of conferences.  Though the city and conference hosts change from week to week, the experiences start to converge.  They start with an early morning train ride balancing a cup of Dunkin Donuts coffee on one knee and a laptop on another- doing what you can to chip away at work to create some value out of the travel time.  When you detrain in Boston, New York, or DC you find the hotel or event space, collect your name tag, and mosey your way to the refreshments.  A cup of weak conference coffee is obligatory, perhaps a pastry to join it if they pass the eye test- extra points for a big jug of resort-style ice water.  Make small talk, find a seat, sit down and hope for a good discussion and get ready to fill out your bingo card: 

• At least one person from the panel description in the brochure has no-showed, sometimes so soon the event staff forgot to subtract a seat
• Screeching microphones, rectified by an 80% reduction in volume for the remainder of the panel
  An excessively long question by an over-zealous attendee trying to demonstrate employability with the thoughtfulness of a question panelists inevitably have trouble hearing
• A moderator referencing a room full of aspiring professionals, asking panelists for practical career advice applicable to 200-500 people
• 50% of panelists answering said question on careers with some derivation of ‘follow your passion’
• A bumrush of desperately unemployed students swarming speakers in hopes of asking 4 very specific, annoying questions that none of the other people huddling around even bother listening to
• And the center square of every conference Bingo since 2022: AI. 

By this point in the conference tour circuit, I’m only slightly jaded and a little road-weary from the whirlwind of panels and travel. Still, there were real bright spots—moments of clarity and insight that cut through the noise. Two events in particular, both hosted by Yale and held back-to-back, stood out as genuinely worthwhile.

The first was the 2025 Yale School of Management Private Equity & Venture Capital Symposium, held on Friday, February 21, in New York City. Throughout the day, panel discussions covered topics such as the integration of artificial intelligence in investment strategies, the current status of fintech, and where climate tech and sustainability in private equity and venture capital fit in the landscape of the second Trump Administration. 

The following day, Saturday, February 22, I returned to New Haven for the Yale Alumni in AI & Innovation Symposium. Organized by the Yale Alumni Association’s Careers, Life, and Yale program, the event brought together students and alumni for a day of panel discussions and networking focused on artificial intelligence and emerging technologies. The programming highlighted the intersection of AI with entrepreneurship, ethics, public policy, and investment—and provided a welcome opportunity to hear from Yale alums working on the front lines of innovation across industries. 

Here are some of the best ideas from the 2 events:

1. AI is coming for your job, people are now willing to admit it

One of the most significant takeaways from these discussions was the growing role of AI in human capital management. Companies are increasingly leveraging AI as a direct substitute for workforce costs. I had an after-stage conversation with a VC focused on software solutions for the built environment, and had the opportunity to ask how startups are growing from software budget lines to sell into human capital budgets. He explained that startups are effectively building value by calculating:

(average human time per task performed by AI)

*(# of tasks performed by AI)

*(fair wage for time) 

---

= price anchor for AI solutions.

This feels like a significant shift—for the first few years A.G.P.T. business leaders and product developers were careful to categorize AI solutions as compliments to human workers, tools that remove drudgery from the modern working experience to allow humans to be more creative, relaxed, and productive. It’s refreshing that speakers have dropped this charade; now we can all agree what we’re looking at: AI is a value creator and job destroyer.

2. Startups are being built differently: leaner, faster, cheaper—with AI at the core

Founders used to need years and millions to ship enterprise-grade products. Now, companies are being built around a few people and some strategic manipulation, repackaging, and prompting of LLMs.

To illustrate this point, let’s look at Mistral. Mistral was founded in 2023, and hit a $2 billion valuation after a €385 million Series A round in late 2023, led by Andreessen Horowitz and Lightspeed Venture Partners. At the time, Mistral was less than 1 year old and had about 55 employees. By contrast, at the two-year mark Facebook was valued at $500 million with 155 employees. Mistral has 4x the valuation, a third of the headcount, and less half of the time—proof that AI-native companies are rewriting the startup growth playbook.

3. The AI engine race mirrors the horsepower wars of the 20th Century

A panelist made a clever analogy that stuck with me: today’s foundational models can be compared to car engines in the 20th century. Starting in the 1960s, American automakers were obsessed with bigger engines—more displacement, more cylinders, more horsepower. Then Japanese manufacturers came in with inline-4s boosted by turbochargers, offering similar performance with less weight, greater efficiency, and smarter engineering. That same dynamic is playing out now in AI. 

AI to Motors Analogy

GPT-4	Your classic V8—massive, powerful, and general-purpose
DeepSeek	Like the turbocharged inline-4: lighter, more specialized, and competitive in the right application.
Small Language Models	Like EVs engineered for a narrow use case with maximum efficiency.

In the post-DeepSeek era, performance doesn’t just come from size. It comes from tuning, architecture, and fit for purpose.  And comparing these to motors may help start a conversation with my Dad where he actually knows what’s going on- we can hope!

4. AI can write your code, draft your content—but it still can’t close

AI has become a powerful tool for modern startups. Developers are using it to generate production-grade code, debug in real-time, and ship features that once required entire teams. Founders are leaning on AI to spin up product demos, design brand assets, and create everything from pitch decks to video content. 

But when it comes to actually selling—to navigating objections, reading the room, building trust, and driving urgency—AI falls flat. Founders shared stories of ChatGPT writing excellent outbound emails that got zero replies. Investors nodded knowingly at the mention of AI-generated demos that never converted. Everyone agreed: AI can get you to the door, but it can’t seal the deal. That still takes a human. For now.

Final Thought

Two conferences, four takeaways, and a lot to chew on. The hype may be tiring—but the stakes are real, and the smart conversations are just getting started.  See you out there on the conference circuit.