Particle's AI news app scans podcasts for interesting clips, so you don't have to

Particle, an AI news app built by former Twitter engineers, now keeps up with breaking news from podcasts as well as the web.

Shortly before its Android launch, Particle introduced Podcast Clips, a feature that identifies the most relevant and interesting moments from a wide range of podcasts, then includes them alongside related news stories in the app's feed.

Instead of sitting through a long podcast to hear a 45-second interesting comment, you can play the clip while reading news on Particle. You can also read the transcript, with words highlighted as they're spoken.

“We’ve done that for virtually any news story — if a podcast is discussing it or even remotely relevant, we have those clips,” Particle CEO Sara Beykpour, formerly Senior Director of Product Management at Twitter, told TechCrunch. “It’s a great way, when reading or learning about a story, to get a sense of what people are saying about it — what’s the commentary?”

This addition reflects a shift in the news ecosystem that has been building for years. More people are not only getting their news from podcasts and trusting them as reliable sources, but the medium is also becoming a go-to platform for breaking news and major announcements from public figures.

Tech CEOs, in particular, are now turning to favorable podcast hosts to share their talking points rather than engaging with traditional media, as Bloomberg reported in 2024.

This makes following podcasts even more essential for staying up to date with the news.

Beykpour explains that Particle uses embedding models to determine when a podcast relates to a specific news story. These models come from the same companies that provide LLMs, but they are not generative AI technologies, she notes.

“We use vector embeddings to understand how different segments of a podcast relate to different stories,” Beykpour notes. “A single podcast might cover 10 or 20 stories, so we use AI to parse that. We also use AI for the clipping logic — determining when to start and end a clip.”

The company uses ElevenLabs technology for transcription. However, the precise audio-clipping identification is part of Particle’s proprietary approach.

The concept of leveraging podcasts to better grasp the conversation around news is also gaining traction in newsrooms. As Nieman Lab reported this month, The New York Times has been using a custom AI tool that employs LLMs to transcribe and summarize new episodes from dozens of right-wing and conservative podcasts, aiming to understand what influencers on that side are saying about the news.

Particle’s Podcast Clips feature isn’t limited to news stories. Since the app already recognizes entities — such as people, places, or things — you can visit the page for a notable figure, like OpenAI CEO Sam Altman, to see all of his podcast appearances displayed as a feed.

Particle has also been developing other features. The company has taken its first step toward monetization with Particle+, an optional $2.99/month (or $29.99/year) subscription that unlocks premium features. These include: using natural language to get news summaries in your preferred style; selecting different voices for the personalized audio feed; “Listen to the News”; unlimited crossword puzzles; private questions with its AI chatbot, and more.

The Android release also introduces several other notable changes. The browse tab now features timely stories, such as the 2026 Winter Olympics, alongside standard sections like politics, tech, or entertainment. Additionally, tapping on an entity brings up a new page with its definition, stories, articles, related entities, and related topics.

Particle isn’t disclosing user activity or conversion rates, but Beykpour highlighted the app’s international audience before the Android launch. On a weekly basis, 55% of Particle’s users are outside the U.S., with India (15%) as its largest market after the U.S.