EO StudioHow Top 1% AI-Native Organizations Actually Make Money | Harvard Business School, Rem Koning
At a glance
WHAT IT’S REALLY ABOUT
AI-native advantage comes from allocating intelligence and embedding it profitably
- A field experiment with WhatsApp-delivered ChatGPT for Kenyan entrepreneurs found AI can reduce profits for low performers while boosting results for high performers, because outcomes depend on users’ judgment in filtering advice.
- AI-native advantage is less about using AI to speed internal tasks and more about embedding AI into the product so customers get work done directly, enabling scale via compute instead of adding staff.
- The next competitive edge shifts from allocating capital or talent to allocating intelligence—deciding which models do which tasks and what humans should uniquely own to create differentiation.
- AI acts as both equalizer and amplifier: it lifts baseline productivity for many tasks but disproportionately rewards founders with strong mental models, context, and ability to design effective AI-enabled systems.
- The coming wave of “agentic” AI and lower inference costs could unlock new, bootstrap-friendly businesses globally, but raises questions about wealth concentration and the need for policy vigilance.
IDEAS WORTH REMEMBERING
5 ideasAI can hurt performance when judgment is weak.
In the Kenya study, struggling entrepreneurs used AI heavily but couldn’t reliably distinguish good advice from plausible-sounding bad advice, leading to ~10% declines in profits/revenue; stronger entrepreneurs improved by selectively applying what worked.
The durable advantage is “allocating intelligence,” not just adopting AI tools.
Firms gain edge by orchestrating which models handle which tasks (and how they’re combined) and by deliberately assigning humans to the work where they add differentiated value beyond model outputs.
AI-native means embedding AI into the product experience.
Using AI for internal speedups helps, but the step-change comes when AI directly performs work with/for customers, removing the human bottleneck and enabling scale without proportional hiring.
Chatbots are often the wrong interface for real business impact.
Advice-only chat can stall when users lack time/skills to execute (e.g., “update your website”); agentic systems that can take actions—build pages, run campaigns, ship workflows—better address real constraints.
Context is the startup’s wedge against frontier model labs.
Startups likely won’t outbuild OpenAI/Anthropic on base models, but can win by encoding workflow-specific or local-market context (e.g., reusable “skills”/procedures) that makes generic models perform like specialists.
WORDS WORTH SAVING
5 quotesI think we're in a world where increasingly what matters is your ability to allocate intelligence.
— Rem Koning
The key for AI native is that you're not just using it to do the work, you're embedding it in the product so that the AI can directly do the work with the customer.
— Rem Koning
It would've been better had we never given them the AI.
— Rem Koning
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.
— Rem Koning
The most dangerous assumption they make is that by building with AI, they have made something people want, and that is just not true.
— Rem Koning
High quality AI-generated summary created from speaker-labeled transcript.
