July 6, 2026 · 57 min
Why Companies Aren't Getting ROI on AI (with Brenn Hill)
Everyone's spending big on AI coding tools — but is anyone actually measuring if it's working?
In this episode, we're joined by Brenn Hill, a Berlin-based engineering leader and author of "The Delivery Gap," to break down the AI ROI myth: why token-based pricing is nearly useless for real business planning, why companies like Uber and Microsoft have pulled back on major AI investments, and the three-part framework (cost, verification, intent clarity) Brenn uses to actually measure whether AI is paying off.
We also talk about:
- Why AI gets you "80% of the way there" — and why that's not the same as production-ready
- Machine catch rate vs. change failure rate, and why they matter more than token counts
- Why tokens can't be compared across AI providers or even across tasks
- What it's like navigating Berlin's housing crisis as an American expat
- Entrepreneurial culture differences between the US, Vietnam, and Australia
Brenn's book, "The Delivery Gap," lays out the research behind what's actually working (and mostly not) in AI-assisted software delivery — plus practical roadmaps for engineering leaders and individual contributors navigating this shift.
📖 Get the book: "The Delivery Gap" — available on Amazon
🔗 Connect with Brenn Hill on LinkedIn
🎥 Watch on YouTube
If you enjoyed this conversation, subscribe for more casual, honest conversations at the intersection of tech, careers, and the future of work.
Timestamps:
- 00:00 – Intro & catching up with Brenn in Berlin
- 08:00 – The AI ROI myth
- 12:00 – Why tokens are a useless cost metric
- 17:00 – The cost/verification/intent-clarity framework
- 24:00 – Machine catch rate & change failure rate
- 52:00 – About "The Delivery Gap" and where to find it
Find The Sunday Draft on YouTube, Spotify & Apple Podcasts →
