Your own daily news podcast, from your sources
Proofite reads the sources you picked (newsletters, Substacks, RSS feeds, Google searches) and every morning turns them into one podcast episode of the length you choose, written for the ear and voiced by two hosts or by one. There's a 30-second sample on this page in Zephyr, the default voice: hear it before you decide.
- Your only real free time is behind the wheel, at the stove or at the gym, when your hands and eyes are busy and your ears aren't.
- You tried an app that reads emails aloud: a flat voice reciting every greeting, every sponsor block, every 'if you enjoyed this email'.
- You'd listen to your newsletters on the commute if they came as one show, not as forty separate audio files a week.
Point your sources at Proofite
Create an inbox per topic and subscribe your newsletters to its address. Next to them go RSS feeds, checked every 20 minutes, and Google searches, rerun four times a day; an OPML file brings up to 300 feeds at once.
The AI writes for the ear
Each morning it writes the day's briefing from what came in, then a script from that briefing. The script opens on the day's most consequential fact and uses only facts the briefing contains.
Press play
Two hosts by default, or one voice if you prefer, for 2 to 20 minutes. Listen in the Proofite app, which notifies you when the episode is ready, or in Apple Podcasts, Pocket Casts or AntennaPod through your private feed.
Why read-aloud apps make the backlog worse
Apps that read newsletters aloud turn a reading backlog into a listening backlog: one audio file per newsletter, each read from start to finish. Fifteen subscriptions make fifteen files, greetings, sponsor blocks and 'forward this to a friend' included. Listening is slower than skimming, so that pile grows faster than you can clear it.
Read-later apps do that job well for single pieces: Matter markets a voice it describes as hard to distinguish from a human, and Readwise Reader turns articles, PDFs and ebooks into audio in dozens of languages. But they read, in full, things you already picked. With forty unread emails, a good voice reading all forty still means forty listens; the cutting has to happen before anything is read aloud.
Text-to-speech vs. a spoken digest
Text-to-speech reads an existing text in full. A spoken digest is a new text: a script the AI writes from the day's briefing, which has already merged your sources, and only then hands to a voice.
Proofite does the second, in English, Italian, Spanish, German, French or 6 more. The script keeps the briefing's order: the top story first and in depth, the others in short blocks, the quick hits compressed near the end. Side by side:
- TTS: one file per newsletter. Digest: every source in one episode.
- TTS: as long as the author wrote it. Digest: 2 to 20 minutes, you decide.
- TTS: prose written for a screen. Digest: a script written to be heard.
- TTS: sponsor blocks read out with the rest. Digest: sponsored blocks never reach the script.
What goes into the morning episode
The episode is built from whatever reached your inboxes since the last one: newsletters, RSS feeds and Google searches, plus links you saved by hand.
Coming from another reader? Import an OPML file from Feedly, Inoreader, FreshRSS or NetNewsWire: you get a preview grouped by your folders, choose what comes in (up to 300 feeds per import), and see which feeds didn't make it and why.
- Newsletters: each inbox has its own address, such as ai-3f9a1c@in.proofite.com. Subscribe there and your personal inbox stays out of it.
- RSS feeds: paste a site's address and Proofite finds the feed itself, so there's no hunting for /feed URLs.
- Google searches: news or general queries that rerun four times a day and bring new results into the digest.
- Saved links: paste a link into Read later on the Overview, or share it from the Android app, and it's guaranteed a place in the next briefing.
Two hosts, or one voice
New accounts start on the Two-host show. The anchor brings the facts and the numbers, the co-host asks what you would ask, and the day's main story is told as an exchange that includes what it means for someone with your job or interests. Everything else comes as quick headlines, one line each.
If you'd rather have a single narrator, choose One voice in Settings → Podcast: the briefing read in full by one voice, with the signature tune and a sting before the quick hits. Either way the script is written from a digest that already applies your feedback, so a topic you muted stays out of the episode too.
Hear the voice before you decide
The 30-second sample on this page is Zephyr, the default voice, reading a short sample script in the one-voice format, with the same style direction the episodes use. In Settings → Podcast you can play each of the nine voices (five female, four male) before choosing, press 'Hear the pair' to hear the two hosts together and, if you listen in English, pick the Accent: American, British, Australian, Irish or Indian. Until you pick one, the accent follows your time zone, so an English listener in Europe hears British English.
Each episode opens with your signature tune, two seconds of music that stays the same every morning ('Change melody' picks another), and a short sting marks the quick hits; the episode notes link every story to its sources. The iOS and Android apps notify you when text and audio are both ready. If seven episodes in a row show no sign of being heard, in the app or from the feed, the podcast pauses itself instead of piling up, and one tap resumes it.
Listen to a sample briefing
30 seconds, the product's real voice.