Aaron Levie

speaker
937 appearances 2 recordings 1 series first heard Jul 2025 last heard 5 Sep

Aaron Levie’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months
1 · Sep OctJan 26AprJulnow

Recordings per month over the last 12 months — 1 in all, peaking in Sep 2026 with 1.

Appearances

newest first · ▶ plays the moment
So uh I think it's it's uh it's definitely trained to be highly uh highly conservative on uh project timelines.
Same with Matt.
Yeah.
Um I I mean kind of that's that's uh that's my conclusion, but I'm also extremely biased in in that conclusion.
Like like like the the more that you need multiple models to to do a a task or a set of tasks, the more value accrues to the layer that can understand the task and get access to the data and handle the workflow, which obviously is is a better outcome for the, let's say, applied AI layer.
Uh
uh which is it which is where we tend to sit.
Um and uh and that that's some that's a that's a future that obviously a lot of companies are are aligned to and and you know strategically and and kind of existentially in some cases.
So if you're cognition and factory and cursor at you know and and replit, et cetera, obviously the outcome that that you want to have happen is that you actually have many models.
They're all good on different axes.
Some are like really like the cost tuned, you know, kind of workhorse models, and some are like the super frontier um, you know, orchestrators.
And you want to have an outcome where you actually need multiple of those m models uh to to uh to be able to complete the task effectively or at least cost effectively.
That's actually the that that appears to be the timeline that we're on.
Uh and uh and now again, I think you have basically five credible US players um in uh
uh in in in model development between SpaceX, um uh Google, um uh uh anthropic OpenAI and uh and and um uh who did I forget Meta.
And so the like those five players are all on a on a war path for both driving down the cost of intelligence and improving the frontier at the same time.
So having a model router that can kind of be above that and
Again, pick and choose at different points, you know, which model to use, uh, I think creates a lot of value for enterprises.
Um, and actually helps with interestingly, it actually helps with the diffusion of AI.
Um, you know, one one of the the the challenges that enterprises have is uh is almost kind of like um analysis paralysis on on how if you have so many models that are that are emerging and they're all constantly leapfrogging each other, you actually
Showing 241–260 of 937 · page 13 of 47 ← Previous Next →