Moritz Sudhof
I'm co-founder and CEO of Bigspin. Teams building AI products have analytics that tell them sessions are happening and evals that catch known failures — but neither tells them what the experience actually is. So they review manually, and that doesn't scale. Bigspin reads every conversation in your traces and forms a view: where users pushed back, where the model lost the thread, where something useful nearly happened and didn't. The interaction layer between human and AI is where success lives, and we make it observable.
I've spent my career at the intersection of language, behavior, and machine learning — first in research, then building products. I co-founded two NLP startups; the first, Motive Software, was acquired by BetterUp in 2021, where I went on to lead AI as VP, building coaching products used by hundreds of organizations. Before that, I studied computer science at Stanford and did research with the Stanford NLP Group on semantics, sentiment, and how social context shapes communication.
I grew up between German and American cultures and have spent most of my career in the Bay Area. My work sits at that intersection: systems thinking with a practical, product-driven bias — curious about ideas, but anchored in whether something actually helps people do better work.
I write about things that confuse me, usually involving AI or parenting.
Research
My research has focused on natural language processing and how language reflects social dynamics — first in academia, now continuing at Bigspin. The frame I keep coming back to: AI isn't a static model problem, it's an ongoing conversation between humans, data, and context. Selected publications:
- Invisible Failures in Human-AI Interactions — On the failures users don't report, and why most AI quality problems go unseen. →
- A Paradox of AI Fluency — On the persistent gap between what AI can do and what it actually delivers for most users, and the user behaviors that close it. →
- Measuring News Sentiment (Journal of Econometrics, with Shapiro & Wilson, Federal Reserve Bank of San Francisco) — Developed the Daily News Sentiment Index, still actively maintained by the Fed. →
- A Computational Approach to Politeness (ACL, with Danescu-Niculescu-Mizil, Jurafsky, Leskovec, Potts) — Built the Stanford Politeness Corpus, still used in NLP and social computing research. →
- Sentiment Expression in Social Communication (KDD, with Potts and colleagues) — Models for understanding emotional expression in text and social contexts. →
Press
Featured in Forbes, Bloomberg, The Wall Street Journal, and Fast Company.
Contact
Email: moritz@bigspin.ai
LinkedIn: linkedin.com/in/sudhof
Twitter: @mmooritz