Amazon Music turns to AI to chop up podcasts

Amazon Music wants to fix the biggest annoyance in podcast listening: the endless, blind scrub through a two-hour progress bar just to find one ten-minute topic.

According to a report from CNET, Amazon Music Podcasts will soon roll out AI-generated episode highlights and automatic chapter markers.

The feature is designed to break long-form episodes into structured segments so listeners can jump straight to the parts they actually care about.

The Snackable AI takeover

This update is the direct result of Amazon quietly acquiring Snackable AI, an audio intelligence startup, late last year.

Snackable AI built its reputation on software that automatically generates transcripts, key highlights, and chapters from long audio and video files. Its website now officially redirects to Amazon Music.

As part of the deal, Snackable AI founder and CEO Mari Joller has taken over as the AI and machine learning product leader for Amazon Music Podcasts.

Instead of treating podcast files like locked black boxes, Amazon is using machine learning to turn hours of spoken audio into skimmable, searchable chapters.

Why it matters

For listeners, this solves a genuine friction point. Podcasts are getting longer, and unless creators manually spend hours mapping timestamps, finding a single insight is nearly impossible. Auto-generated chapters make dense shows browseable.

For creators, chaptering and extracting soundbites is tedious administrative work. Offloading that task to on-platform AI is a win for workflow efficiency.

From a platform perspective, competing with Spotify and Apple Podcasts on exclusive content is notoriously expensive. We think Amazon's move to differentiate through playback intelligence and UX improvements is a much sharper strategy than overpaying for celebrity exclusives.

That said, spoken conversation is packed with sarcasm, tangents, and nuanced context. If the AI chops up episodes without understanding narrative flow, it risks pulling out misleading soundbites. The real test will be whether the AI understands conversational nuance or just hunts for keywords.