Harvard Business School professor Rem Koning shares research on how AI-native companies outperform traditional ones, why the ability to allocate intelligence is becoming the new competitive edge, and how AI simultaneously equalizes and amplifies outcomes depending on founder judgment.
The AI Founder Sprint: Global Evidence
Koning led the AI Founder Sprint through INSEAD, tracking over 500 entrepreneurs across Africa, Asia, the Americas, and Europe as they adopted AI-native practices.
- Founders taught to be AI-native accomplished roughly 20% more per week
- They were more likely to get customers, launch products, and generate revenue
- Despite growing faster, these founders wanted to raise $250,000 less in capital
- Entrepreneurs reimagined workflows by building custom AI agents instead of hiring headcount, fundamentally changing business economics
What AI-Native Actually Means
Being AI-native goes beyond using AI to speed up internal processes. The real value comes from embedding AI directly into the product.
- Process AI: Using AI for coding, customer support, internal workflows — helpful but not transformative
- Product AI: Embedding AI so it works directly with the customer, removing humans from the loop
- Gamma as an example: instead of hiring tens of thousands of graphic designers, they scale with compute by letting AI generate presentations from a few sentences
Where are those places you can create loops where the AI is working with a user or another AI or something on the website where your team doesn't even need to be involved? That's the key to building AI native organizations.
Allocating Intelligence as Competitive Advantage
Koning draws a parallel to historical sources of business advantage: capital allocation (Warren Buffett) and talent allocation (McKinsey). The new frontier is intelligence allocation.
- Deciding what to assign to Claude, Lovable, Grok, Deepseek, or other models
- Determining what remains with humans — not because humans are faster, but because they think differently
- Strategic advantage comes from doing something different, not just doing something better
If you can work out how you bring your human intelligence and you allocate jobs in the company to humans and the places where they can add value over and above the models... I think that's a place where we're going to see a source of advantage moving forward.
AI as Equalizer vs. Amplifier
AI acts as both, but the distinction matters enormously.
- Equalizer: Everyone can now code with Lovable, build decks with Gamma, and eliminate typos with Claude — the baseline rises for everyone
- Amplifier: When it comes to building new businesses and reimagining workflows, returns disproportionately favor those with existing judgment and experience
The Kenya WhatsApp Study
A pivotal research finding that illustrates the amplifier effect:
- ChatGPT delivered over WhatsApp to Kenyan small business owners
- Struggling entrepreneurs saw a 10% decline in profits — they couldn't distinguish good advice from bad
- Already-successful entrepreneurs improved — they had the judgment to follow the right advice
- Koning believes even with Claude Opus behind WhatsApp today, the same pattern would hold — the bottleneck is human judgment, not model quality
Unless you've developed the judgment, the mental models to actually know where to apply it, it can lead you down a road of slop. And that slop can actually lead you to make less money.
Beyond Chatbots: The Agentic Shift
Koning argues the industry is stuck in a chatbot mindset from the ChatGPT era.
- Chatbots give advice but can't execute — telling a Kenyan entrepreneur to "update your website" is useless if they can't code
- The real opportunity is agentic AI: virtual employees that take actions, build websites, launch marketing campaigns
- Context is the startup moat — you'll never build better foundation models than OpenAI or Anthropic, but you can have better contextual knowledge of specific workflows and markets
AI and Emerging Markets
- Developing markets could leapfrog like they did with fintech (India's UPI payments infrastructure)
- Plummeting inference costs will make AI practically free, opening markets where people have less money to spend
- AI can provide virtual expert labor in places where specialists simply don't exist — a marketer-quality agent in Nairobi equivalent to one in New York
- Software will proliferate to solve niche problems: a CRM for Thai restaurant owners, tools for specific local markets
The Biggest Mistake Founders Make
The most dangerous assumption they make is that by building with AI, they have made something people want. And that is just not true.
- AI tools are so fun that founders get trapped in an over-engineering loop — building features for months without validating demand
- A little AI goes a long way: Gamma's core innovation was just generating a deck from a few sentences; everything else was traditional software
- Find the smallest point in a workflow where a drop of AI unlocks real value
- Earned insight, judgment, and taste matter more than ever — use last generation's model if needed, but know where and how to apply it
Looking Ahead
- More people will become entrepreneurs as building becomes accessible to everyone
- Vibe coding may lead to a world of small SaaS applications and bootstrapped businesses rather than network-effects monopolies
- Policy questions remain about wealth concentration as more of the economy starts to resemble software
- An eventual inversion of control may occur where AI handles high-level strategic thinking and delegates to humans for areas requiring taste and agency