I was early to AI in 2018—here's what I learned
Reflections on shipping AI-generated audio news at NU.nl before the current AI boom.
In 2018, I shipped an AI-generated audio news product at NU.nl that reached 90,000 daily listeners. This was before GPT-3, before the AI hype cycle. Here's what that experience taught me about adopting emerging technology.
The Project
NU.nl is the largest news platform in the Netherlands with 8M+ monthly users. The challenge: how do we reach users who don't have time to read?
The solution: AI-generated audio summaries of news articles, delivered as a daily podcast.
Why It Worked
1. We Solved a Real Problem
People wanted news during commutes but couldn't read while driving. Audio was the answer, but producing traditional podcasts didn't scale.
AI let us generate audio for dozens of articles daily—impossible with human narrators.
2. We Set Appropriate Expectations
We never claimed it was human-quality. The product was positioned as "AI-generated summaries"—fast, convenient, good enough.
Users accepted the trade-off because the value proposition was clear.
3. We Iterated on Feedback
Early versions were rough. The voice was robotic, pacing was off, some pronunciations were wrong.
We built feedback loops:
- User ratings on each episode
- Skip patterns to identify boring content
- Comments and support tickets
Each week, we improved based on data.
What I Got Wrong
Overestimating Technology, Underestimating Content
The AI could generate audio, but choosing what to narrate mattered more. We spent too much time on voice quality and not enough on content curation.
Thinking Technology Alone Was the Product
Users didn't care about AI. They cared about getting news efficiently. The technology was a means, not an end.
Not Building for Mobile First
Our initial player was desktop-optimized. But people listened during commutes—on mobile. We lost early adopters to poor UX.
Lessons for Today's AI Products
Having shipped AI in 2018 and now in 2024, here's what I've learned:
1. AI is Infrastructure, Not Product
The best AI products don't lead with "AI-powered." They lead with the outcome. "Remove your background in seconds" beats "AI-powered background removal."
2. Start with 10x Improvement, Not 10%
AI products need to be dramatically better than alternatives to overcome adoption friction. Incremental improvements don't justify learning new tools.
3. Human-in-the-Loop Is Underrated
Pure automation is tempting but often worse than human-AI collaboration. The best results come from AI doing 80% and humans refining the rest.
4. Technical Capability ≠ Product-Market Fit
Just because you can build something doesn't mean you should. The graveyard of AI startups is full of impressive technology solving problems nobody has.
5. Timing Matters More Than You Think
In 2018, AI audio was novel enough to get press but not good enough for mainstream adoption. The same product in 2023 would face different challenges—more competition, higher expectations.
What's Different Now
The AI landscape in 2024 vs 2018:
| Aspect | 2018 | 2024 |
|---|---|---|
| Model quality | Limited, task-specific | General-purpose, high quality |
| User expectations | Low, forgiving | High, demanding |
| Competition | Minimal | Intense |
| Infrastructure | Build everything | APIs and platforms available |
| Hype | Niche interest | Mainstream attention |
The Constant
Despite all changes, one thing remains true: users don't care about your technology. They care about their problems.
Solve real problems. Deliver clear value. Let the technology be invisible.
Building AI products? I've learned a lot from mistakes. Happy to share more—reach out on LinkedIn.