PodSnacks (https://podsnacks.org) — Personalized AI summaries of top podcasts, delivered to your inbox.
I've been building PodSnacks because I found it overwhelming to keep up with podcasts across tech, business, and science. PodSnacks uses LLMs to summarize the most popular episodes from shows like Lex Fridman, Acquired, All-In, Invest Like the Best, and more.
You choose your favorite shows, and we email you short, high-signal summaries — no audio to skim through, no endless backlog guilt.
So far:
126K+ episode summaries generated
92K+ hours of podcasts processed
48–50% open rates
2,900+ early users
Still iterating and adding features like bundling by theme, language translations, and audio feeds.
If you're the kind of person who wants more inputs without more noise — would love for you to check it out.
What are the costs like, for such a service? You have to spend for transcribing (even if you ran something like Whisper) and pay for APIs to summarize, correct?
I’d prefer summaries of my YouTube subscriptions, where I can also add my own “summary template”
In currently have a prompt for it that works for me, based on the transcripts.
Problem: too much duplicate information in any type of publication and too much fluff
Problem2: YouTube/transcriptions
Probably wouldn’t pay for such a service, but would be very happy using it. Perhaps some channel promo / email based ads for discovery or recommendations.
thanks and agreed! we see a massive distrust in traditional legacy media, and hopefully PodSnacks can give readers more control about how they access those voices.
any way to expose api access for the keyword search feature? would be cool to build this into some topic summaries delivered via email I'm building for my team
I've been building PodSnacks because I found it overwhelming to keep up with podcasts across tech, business, and science. PodSnacks uses LLMs to summarize the most popular episodes from shows like Lex Fridman, Acquired, All-In, Invest Like the Best, and more.
You choose your favorite shows, and we email you short, high-signal summaries — no audio to skim through, no endless backlog guilt.
So far:
126K+ episode summaries generated 92K+ hours of podcasts processed 48–50% open rates 2,900+ early users
Still iterating and adding features like bundling by theme, language translations, and audio feeds.
If you're the kind of person who wants more inputs without more noise — would love for you to check it out.
Always open to feedback from HN!