Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


Fathom Applies AI to Podcast Listening

Company wants to supplant big names as podcast lovers’ go-to player

Finding interesting podcasts can be like shopping in an old-fashioned record store. There are millions of albums grouped by genre, but that’s it. Unless you know specifically what you’re looking for, you’re on your own.

The Fathom podcast platform intends to change this. That’s because the Fathom podcast player uses artificial intelligence to help listeners find podcasts that really resonate with them.

“Most other podcast players treat podcasts more like music — a large, long audio file that’s played, paused or skipped,” explained Tyler Termini, Fathom’s product manager. 

“Fathom uses AI to comprehend the ideas within podcasts, so listeners can confidently decide what to listen to next using personalized audio clips, use ‘x-ray vision’ for podcasts by being able to navigate chapters and transcripts derived from them, and search podcast content in detail.

“One of our fans described Fathom like ‘Netflix, but for podcasts.’ We appreciate this comparison, but it would only be 100% true if Netflix also provided users the ability to ‘search within’ movies like Fathom can search within podcasts, or let viewers create their own movie clips to share with others.”


Fathom users can listen to a curated feed of highlights from podcasts they follow.

Like Netflix, the Fathom podcast platform pays attention to the initial content its users select, using it and further selections to make podcast recommendations. However, Termini said, the AI-engine driving Fathom allows it to provide much more focused recommendations, along with empowering listeners to delve more deeply in their own searches.

“Due to the accuracy/precision of Fathom’s assessment of what a listener actually listens to and likes, it’s able to surface the most relevant and interesting ‘clips’ from a podcast for the listener. This makes the podcast discovery process far more relevant to users. Listening to a preview from the latest podcast you follow is a lot more interesting than just searching by name, as is finding relevant clips via search.”

Further, because Fathom is transcribing and “comprehending” podcasts, it knows a lot more than the broad ‘categories’ that a podcaster might have listed their podcast under. 

“These are often inaccurate, since podcasters will leverage adding several superficial categories for marketing purposes.”


In today’s world, AI is becoming an integral element of data processing. In aviation, AI is being used to create digital twins of jet engines, with every change detected in the real engine being added to its virtual sibling. In this way, mechanics can troubleshoot and remedy anomalies in the real engine as soon as they occur, by tinkering with the digital twin to see what works and what doesn’t.

AI is being used by law enforcement to search and analyze everything from surveillance video to vehicle license plates caught by patrol car cameras. In this latter instance, the AI compares the license plate numbers to the department’s database of stolen and wanted vehicles. In seconds, the system can tell patrol officers if the car ahead of them is indeed wanted, and direct them to apprehend it. This is proving to be a much more effective and efficient way to detect and stop stolen/wanted vehicles than by having officers do it on their own.

Given how well AI is performing in other industries, it makes sense to apply it to a data-based application like podcast selection. 

“Using AI to comprehend the content within podcasts for recommendation and curation allows listeners to more accurately target specific ideas or conversation topics, as opposed to just listening to ‘what others are listening to,’ although this will always play some role, including within Fathom,” Termini said. “This makes podcast recommendations in Fathom far more relevant than any other podcast experiences.”


The AI-enabled Fathom podcast platform has big ambitions.

Using natural language, the user can search within a specific podcast, or “across the podcast universe.”

“Fathom seeks to fully replace Apple Podcasts, Spotify, YouTube and others as a podcast lover’s go-to player,” said Termini.

Is this hubris? By using AI at the heart of its podcast recommendation, search and selection process, Fathom is doing far more than other platforms that simply view podcasts as another form of music-like content. If the company gets a break and podcasters and fans discover its enhanced capabilities en masse, maybe Termini’s words will bome true.

After all, doing something better is what helped Netflix’s on demand/online videos replace Blockbuster’s videotape/DVD rentals, and Apple’s touchscreen iPhones to dethrone BlackBerry’s smartphones with physical keyboards. (This writer makes that last statement with a pang of regret, since he owned every generation of BlackBerry from the initial two-line pager to the final BlackBerry Motion touchscreen, which he still uses.) 

So maybe the possibility of Fathom overthrowing today’s most popular podcast platforms should not be dismissed.

“AI will be the primary tool by which listeners hone in on the podcasts that matter the most to them, and that are of the highest quality,” Termini said. 

“Until these AI tools and infrastructure become as accessible as a weekend carpenter’s tools, solutions like Fathom will be the go-to solution for the everyday podcast listener.”

In the meantime, the company is working to make its AI-enabled podcast platform more powerful and appealing. “Fathom 2.0 has already been launched and is available on the Apple App Store as ‘Fathom Podcast Player,’” Termini said. 

[Read More Radio World Stories About Artificial Intelligence]