Bubble Talk, AI Flop Rates, and the Billion-Dollar Owl

Duolingo cofounder and CEO shares how they’re successfully scaling AI, and it has nothing to do with the usual enterprise playbook.

While CFOs are pulling the plug on failed pilots, and Sam Altman is busy calling the AI market a bubble, Duolingo is building one of the most successful consumer AI businesses of the decade.

Let’s dig into how they did it, and what enterprise leaders can learn from the owl.

Download our guide on AI-ready training data.

AI teams need more than big data—they need the right data. This guide breaks down what makes training datasets high-performing: real-world behavior signals, semantic scoring, clustering methods, and licensed assets. Learn to avoid scraped content, balance quality and diversity, and evaluate outputs using human-centric signals for scalable deployment.

Download the guide now

The $11 Billion Owl Flies Differently

Luis von Ahn, Duolingo's cofounder and CEO, is one of the rare technologists who's not only early to AI, but also effective. As the New York Times spotlighted this weekend, von Ahn built Duolingo’s massive growth around a deceptively simple idea: make learning a language addictive while educational.

The secret is a deep, deliberate product integration of AI. Not just generative razzle-dazzle, but behind-the-scenes optimization and iteration rooted in user behavior. AI enhances content, adjusts lesson difficulty in real time, and supports hyper-personalization. Users don’t care that it’s “AI-powered.” They just keep coming back.

Meanwhile, back in enterprise land:

• 95% of GenAI pilots are still failing, according to a new MIT & Fortune CFO survey.

• Most corporate projects are too theoretical, too bloated, or too vendor-led to deliver real value.

Enterprise takeaway:
We can’t all be Duolingo. But we can absolutely borrow their model:

• Use AI to make the product better, not just smarter.

• Focus on compounding UX advantages, not tech-stack bragging rights.

• Invest in invisible AI—tools that quietly power better outcomes, not dashboards for their own sake.

Don’t just “do AI.” Make AI part of something people want to use.

News Roundup

1. Sam Altman Says the Quiet Part Out Loud
   OpenAI’s CEO admits the AI market is in a bubble, but insists OpenAI’s investments will still pay off. This is the clearest signal yet that the frothy phase is cooling.
   Read more →

2. MIT Finds 95% of AI Pilots Fail
   A sweeping CFO survey reveals most generative AI projects are flopping due to poor integration, unclear ROI, and lack of alignment with actual business needs.
   Read more →

3. HHS Goes All-In on AI
   The Department of Health and Human Services has launched a new strategy to embed AI into public health, analytics, and operations. It’s one of the largest federal commitments to AI integration to date.
   Read more →

TL;DR:

• Duolingo’s success is about product-led integration that actually helps users.

• Sam Altman is calling out the AI bubble. Expect investor sentiment to shift.

• 95% of enterprise GenAI pilots are missing the mark (and CFOs are noticing).

• The U.S. government is doubling down on AI, starting with healthcare.

• Enterprises that treat AI like seasoning and not the main course are getting the best results.

Today’s takeaway: Want to Avoid a 95% Failure Rate? Take a page from Duolingo and build AI around real behavior, not hype.

Stay sharp,

Cat Valverde
Founder, Enterprise AI Solutions
Navigating Tomorrow’s Tech Landscape Together