AI-native podcast studio · EN · ES · FR · DE
Turn expert knowledge into conversations people want to hear.
Create source-grounded, multi-speaker B2B podcasts from your existing research and content — with clear claims, licensed voices, and approval before anything is published.
Human-approved: AI researches, plans, writes, verifies and produces. A person stays responsible for sources, voice rights, script approval and publication.
- 00:14 · Host
From August 2026, anything that publishes synthetic audio has to say so — that is the Article 50 rule. C1 · supported
- 00:31 · Analyst
Labelling, not licensing — and the label has to survive the export. C2 · supported
“Providers of AI systems generating synthetic audio, image, video or text content shall ensure that the outputs are marked…”
How it works
One brief. A reviewable episode, not a black box.
- 01
Brief and sources
Describe the episode and add the material you hold the rights to use: PDFs, articles, notes, public links. Each source is checked, checksummed and indexed into citable passages.
- 02
A script with a claim ledger
Two or three personas discuss the topic at the audience level you chose. Every material claim is linked to a passage in your sources or labelled as opinion. Unsupported claims block the script.
- 03
Review and approve
Read the dialogue with its claims beside it. Cite a passage, mark an opinion, or send the script back with notes. You approve one exact, immutable version.
- 04
Licensed voices
Each persona speaks with a voice whose consent is on record. Listen turn by turn, flag one line, and regenerate only that turn. Mastered to broadcast loudness.
- 05
Publish with disclosure
The episode, transcript, chapters, show page and RSS feed go out with the synthetic-media disclosure each destination requires — now or on a schedule. Takedown is one control away.
What you get
Everything a recurring show needs, from one brief.
- Script
A multi-speaker conversation, segmented into chapters, in the audience's language.
- Claim ledger
Every material claim with its cited passage, verdict and version — defensible after publication.
- Mastered audio
Per-turn synthesis with licensed voices, loudness-normalised and technically checked.
- Transcript and chapters
Timestamped transcript and chapter markers on the public episode page.
- Show page and RSS
A public show page and a directory-ready feed served from a CDN, with stable episode identifiers.
- Disclosure and audit
Synthetic-media disclosure per destination, plus an approval trail with named versions.
Launch languages: English, Spanish, French and German — one show per language, each with its own personas and voices.
Why not a free generator?
Audio a legal or brand reviewer can sign off on.
Free tools make a two-host audio file from your documents — and stop there. No feed, no claim-to-source trail, no approval workflow, no voice licensing, no disclosure management. Provencast produces a recurring show your reviewer can approve in minutes and defend afterwards: every claim traced to a source you provided, every voice licensed, every destination’s disclosure rules handled, published to your feed on schedule.
Source-linked claims
Each material claim points at a passage in a source you provided, or is marked as opinion.
Approval trail with named versions
Scripts and audio are immutable versions; approvals name the exact version they cover.
Licensed voices with recorded consent
A voice renders only while its consent is valid. Expired consent blocks rendering and publication.
Per-platform disclosure
Publication refuses a destination whose synthetic-media disclosure is incomplete.
Founding partner program
Three design partners. A managed show, operated through the product.
We produce your show with you for three months: a defined source, review and revision workflow, an editor on the episodes, and weekly feedback while we learn what your reviewers need. Places are by application — we qualify material quality, review speed and rights readiness before we start.
- Setup
- €750, one time
- Monthly fee
- €1,250
- Minimum term
- 3 months
- Included output
- 4 episodes a month, up to 15 finished minutes each
- Revisions
- One script revision and one turn-level audio revision per episode
- Feedback
- Weekly in the first month