Moritz Sudhof
speaker
399 appearances
1 recordings
1 series
first heard Jul 2026
last heard 8 Jul
Moritz Sudhof’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsRecordings per month over the last 12 months — 1 in all, peaking in Jul 2026 with 1.
Appearances
Where is the thing that compounds?
And I think a big part of that is in the interaction.
And it's instrumenting the interaction so that your product is learning from it.
It's learning from the users.
Ultimately, those are the people who you're who you care the most about, who you're trying to have succeed, who you're trying to become the biggest advocates for your product and your brand.
They're also the people who in these sessions with the AI, they are teaching it.
They are especially your expert users.
They are getting it to succeed at things that by itself it doesn't, or with a with a lower fluency user, it doesn't.
If you can instrument your interactions to actually have your AI system and your product team learn from what those users are already teaching the AI, have that learning endure beyond just that's that ephemeral session.
That is how you can build a truly great product with an actual proprietary data set of feedback loops from your actual users about how do you get one of these commoditized models to actually deliver an interaction.
and deliver work output that really satisfies your users and gives them the ROI that they want.
I think there's a such an important role.
Um, probably the most important role, certainly in my own product development, it was the non-technical stakeholders, it was the designers, the UX researchers, the people who were closest to the users, who understood the business context.
They were the ones who invariably had the most actionable recommendations for how the AI system needed to.
change or how it needed to evolve.
Because those are the people who actually understand the user, the problem, the user experience.
And that's the precisely the the often the gap between these highly sophisticated technical systems and like a product that will actually satisfy a user.
And so the the number one recommendation I have is um it's just read read real transcripts between
Read real sessions between users and the AI, or look at your own sessions that you've been dog fooding.
And identify from those.
Showing 361–380 of 399 · page 19 of 20
← Previous
Next →