Traditional SaaS is heading for a cliff, and the cause isn’t pricing pressure or consolidation — it’s that the underlying model of “build an app, then bolt on AI” has stopped making sense. AI-native development flips that assumption: the model becomes the core engine, and what used to be application code is now a generated artifact wrapped around it.

Alan Shimel, broadcasting from RSAC, sits down with Anaconda CEO David DeSanto to dig into what that shift actually looks like for developers. DeSanto’s argument is blunt — if a product’s only differentiation is being a thin AI wrapper, it’s already obsolete, because the same vibe-coding workflow that lets a developer stand up a full application in a few hours can replicate it just as fast.

They get into the harder questions sitting underneath the hype: how to keep AI-native code secure when models are stitching together open source packages on the fly, what trustworthy package supply chains need to look like in that world, and where Python’s ecosystem fits as the default substrate for AI development.

The bigger takeaway is that the next wave of software isn’t going to be defined by who has the slickest UI or the deepest integrations, but by who can ship AI-native applications safely and fast — with the model, the data and the developer experience all designed to work together from day one.