Pinned Tweet
I'm a Product Leader @ Amazon (AWS), with a background in psychology & cognitive neuroscience.
I’ve spent a decade combining those worlds to build software used by millions.
But I see a lot of noise. Vanity metrics. Feature factories.
I started The B2B Signal to fix that. 🧵👇
I tested 10 AI models on 11 real product decisions.
I wrote up how I built the eval, including test cases, scoring rubrics, AI judges, and lessons learned.
Code is on GitHub if you want to build your own.
overcorrected.substack.com/p…
I tested 10 AI models on 11 real product decisions I’ve made as a PM. The bigger takeaway is that you should stop asking which model is best. Ask which model is best for your specific work and use case. overcorrected.substack.com/p…
In 1990, a Stanford study asked "Tappers" to tap a song for "Listeners" to guess.
Prediction: 50% success. Reality: 2.5%.
Tappers hear the melody. Listeners just hear noise.
This "Curse of Knowledge" is ruining your customer interviews.
The fix: theb2bsignal.substack.com/p/…
I write about 3 things:
1/ The Psychology: Why your brain (and your customer's brain) tricks you.
2/ The Frameworks: The mechanisms we use at Amazon to kill bad ideas before they become code.
3/ The Metrics: How to measure "Outcomes" instead of just "Shipping."
If you want to decode the noise and build software that earns its place in the budget:
Follow me @tunvall for insights.
Join the newsletter (it’s free): theb2bsignal.substack.com/
Let’s build the signal. 📡



