If you publish news, you already have an RSS feed. Turning that feed into a daily stream of short-form videos used to mean a dedicated editor; today an AI news video generator does the same job end-to-end — script writing, voice-over, vertical editing, captions, publishing. This guide walks through how it works and how to set it up on FeedReels.
What an AI news video generator actually does
An AI news video generator ingests structured text (an RSS item, a URL, a snippet) and outputs a finished 9:16 video: AI-written hook, native-language voice-over, animated captions, B-roll or avatar shots, and an export ready for TikTok and Reels. The key word is "news": the system is tuned for short-lived items where speed-to-publish matters more than cinematic polish.
Why RSS is the right input
RSS feeds are stable, machine-readable, and free. Every news outlet, niche blog, and aggregator exposes one. Pointing an AI video generator at an RSS feed is the closest thing to a self-driving channel — new article in, finished video out, no copy-paste.
The five-step pipeline
1. Connect one or more RSS sources in Sources. 2. The script writer rewrites each new article into a short spoken script (hook, 2–3 facts, CTA) in your chosen tone and language. 3. A voice-over is rendered with a premium voice, your cloned voice (Creator+), or an Argil AI avatar from the Avatar Studio. 4. Auto-edit assembles the vertical 9:16 video with animated captions burned in. 5. The video lands in your Queue with an email preview — approve once and FeedReels publishes natively to TikTok and Instagram Reels through the official OAuth APIs.
Editorial control where it matters
Full automation does not mean zero oversight. Approval is a one-click email; if a draft is off-tone you can regenerate, edit the script, or skip it. A configurable delay between publishes paces the channel so you don't dump 20 videos at once.
Languages, voices, and avatars
FeedReels supports 12 languages with native voices. Stick to one voice or one avatar across the channel — consistency is what makes a feed feel like a brand, not a content farm.
Costs and infrastructure
Every video shows the underlying provider cost (script LLM, voice, render) so you always know your unit economics. The whole stack is Swiss-hosted, idea-to-video benchmark is around 90 seconds for voice-only and ~3 minutes with an avatar.
Get started
Add an RSS source in Sources, pick a tone and a voice, approve the first video, and the channel runs on autopilot from there. Pair it with the Avatar Studio for one-off ideas that don't belong in a feed.